This is not news to anyone reading this, and most readers could likely rattle off the high-level drivers of this (e.g., some version of modularity / economies of scale / learning curves).

But I realized something recently: I have always focused on the right side of the chart, thinking ‘Now that solar is cheap, what can we do with it? How can we make it even cheaper?’

I didn’t think to ask the more basic question: Who on earth was buying enough solar panels at $100 per watt to get us down the learning curve in the first place?

‘How Solar Energy Became Cheap’ by Gregory Nemet tells that story1, and lays out a framework through which to think about other similar innovations.

The key was a ladder of niche markets, each providing the step-up in scale (and requiring a step-down in costs) that eventually created a world-beating technology.

Share

Space, the first frontier

Like many mid-century technologies, solar photovoltaics in their current form were invented at Bell Labs. Building on early experiments from the 1800s using selenium, Calvin Fuller, Gerald Pearson, and Daryl Chapin developed the first practical silicon-derived solar photovoltaic (PV) cell in 1954.

In a way, it is no surprise that this breakthrough occurred in the same environment that birthed the transistor. Both relied on a p-n junction in doped silicon for their mechanism, and for much of the next 50 years solar PV would piggyback on progress in integrated circuits as computers and microchips became ubiquitous.

Initially though, the technology was wildly expensive (~$3,500 per watt in 2024 dollars) with no proportionally valuable use case.

Until the space race.

In 1957, the Soviet Union launched Sputnik 1, the first artificial satellite in orbit. Early satellites like Sputnik (and Explorer 1, the US’s first successful satellite) were powered by silver-zinc batteries. These batteries, though state of the art, were heavy and could only provide days worth of power.

In 1958, the US launched Vanguard 1, its second satellite and the first to include solar PV cells. Where Vanguard’s batteries depleted within 20 days of launch, its solar PV cells continued to power the craft for a full 6 years.

This difference in performance was critical, and solar PV quickly became the primary energy source of most spacecraft2, often in conjunction with onboard batteries. In parallel, the space industry became the largest customer of solar PV cells over the next 15 years.

As space applications continued to push improvements in cell efficiency, durability, and weight, solar PV started to see deployment in other niche applications like navigation buoys, rural telecom repeater stations, and offshore oil platforms.

This next niche beyond space allowed for a much wider diffusion of experience with the new technology, while providing demand for continued scale-up of production. These also provided the first (somewhat) cost-conscious buyers of solar PV. Where the space race had national pride on the line (and a parallel disregard for cost), other buyers for extreme applications were commercial enterprises, and valued solar’s reliability and relative-cost-improvement versus alternatives like stand-alone batteries or combustion engines.

These two niches marked the first phases of solar commercialization from the late 1950s to the early 1970s, but they weren’t enough on their own to move solar PV much down the cost curve. That would require larger terrestrial markets.

The oil crisis and technology-push

In 1973, the first Arab oil embargo3 showed just how vulnerable the US was to an external energy shock.

Suddenly, oil was expensive. This catalyzed policy support and government investment in a wide range of energy technologies to bolster the US against future shocks. This included the establishment of what became the US Department of Energy along with the launch of ‘Project Independence’ by then-President Nixon to relieve the US from reliance on external energy sources by 1980.

Solar PV was not a primary focus of these efforts, given its relative nascency and high cost. Those honors went to nuclear, coal, and synthetic fuel investments. Still, significant absolute resources were poured into solar PV research and scale up. The Solar PV R&D Act of 1978 provided for ~$1.5B over 10 years towards research into crystalline, thin film, and concentrating solar arrays, a strong boost for fundamental PV research.

On the commercialization front, one key US government effort was the ‘Block Buy’ program. Between 1975 and 1985, a series of 5 block purchases of solar PV with increasingly stringent specifications saw the price per watt fall towards $10, a valuable early example of government market commitment mechanisms.

In parallel, countries like Japan began to experiment with their own solar PV programs as part of their ‘Project Sunshine’. Solar PV was no longer just a US concern. Japan was also a subject of the Arab oil embargo, and was even more vulnerable due to its status as an island nation without significant domestic energy resources (e.g., coal, oil, gas).

This became suddenly relevant when the US stepped back markedly from solar PV following the 1980 election of President Reagan and the crash in oil prices in the early 1980s. Never again would the US lead in solar PV production, but the industry had taken root overseas.

Japanese industrial policy

Japan kept the light of solar PV alive through the 1980s and 1990s. As the US retreated, Japan drove a progression of breakthroughs and iterative optimizations as the country’s industrial companies, most notably Sharp, continued to push the technology forward.

This was a time of continued expansion into progressive niches: more buoys, lighthouses, oil wells, oil pipelines, railroad crossings, off-grid residences, pocket calculators.

Then, in the 1990s as the niche markets became increasingly saturated, MITI (Japan’s Ministry of International Trade and Industry) pioneered the first subsidies for residential rooftop solar. These programs were the first to explicitly focus on rooftop solar. They also introduced subsidies that declined over time, a pattern that has been copied by many subsidy programs since.

MITI also facilitated the adoption of net-metering of solar PV systems on residences, simplifying the process for households to interconnect and capture value from their solar PV systems.

Taken together this provided the next big jump in demand for solar PV and helped to establish it as an emerging mass-market technology. Vertically integrated manufacturers like Sharp stepped in to provide cells to this burgeoning market.

This was helped by parallel advancements in and explosive demand for silicon-based computer chips through the introduction of personal computers and the internet over the same time period. Even as solar PV increasingly established itself as an independent industry, it continued to be able to draw on the accumulated experience and equipment from the adjacent chip sector.

German demand shock

In the early 2000’s the Germans picked up the torch in a major way with the passage of their Erneuerbare-Energien-Gesetz (EEG) renewable energy law. First passed in 2000, this provided support for a range of renewable energy technologies, including feed-in tariffs that paid generators above-market market rates for the energy they produced.

Strengthened in 2004, this law supercharged the adoption of rooftop solar. Per Nemet, “these revisions sparked the EEG’s ‘gold rush moment’. Installations grew by a factor of four from 2003 to 2004 and transformed the industry.” The policy would catalyze over 30GW of installations between 2004 and 2012 (at a cost of over 200 billion Euros, or ~5000 Euro per German household).

This scale pushed solar PV into the mass-market, with equipment manufacturers developing specialized production equipment specific to solar (not modified from computer chip production). To help cater to the booming demand for solar PV, equipment vendors even went so far as to develop and market ‘turnkey’ solar PV production packages. These deployments, supported in commissioning and training by the vendors’ technical staff, helped new entrants quickly join the fray as panel suppliers scrambled to meet demand.

In fact, demand grew so quickly that German and Japanese solar PV suppliers couldn’t scale up fast enough to meet it, ceding market share to new Chinese rivals like Suntech. These competitors leveraged equipment manufactured in Germany to undercut German solar PV suppliers through lower costs and an ability to scale up more quickly.

When the financial crisis came in 2008, Chinese government support for the solar PV industry helped indigenous suppliers continue to grow and expand just as their German competitors saw a capital crunch.

Framed as Germany’s ‘Gift to the World’, the EEG was modified in 2012 to be much less generous to solar PV. By that point Germany had catalyzed a transformation in solar PV from niche to mass-market, while leadership once more passed to a new country.

Chinese hegemony

And now we get to the part of the story that most people already know.

Over the last 15 years, costs have continued to tumble down as the solar PV industry has become increasingly concentrated in China. From silicon production through module assembly, the Chinese solar PV supply chain is robust, specialized, and efficient, cranking out more solar PV modules in 2024 than the entire industry between 1975 and 2019 (~588GW versus ~580GW).

And this production has not just been for export. China has also been the most prolific installer of solar PV capacity, in part thanks to its own feed-in tariff first introduced in 20114.

This sustained growth has created a truly global market for solar PV, with cost reductions creating fast-growing markets in countries like India, Brazil, and Pakistan, and empowering solar maximalists to shift the Overton window towards a solar-powered society.

Recognizing the strategic importance of solar PV, countries like the US have sought to protect their own domestic solar industries through tariffs and tax credits. This has caused some production and assembly to shift back or to other countries in southeast Asia like Vietnam, but the center of gravity for the industry remains in China with little signs of that changing.

Takeaways

So to recap, who was buying all those solar panels?

• US - space (1960s) and R&D (1970s)

• Japan - industrial niches (1970s to 1990s)

• Germany - mass-market (2000 to 2012)

• China - global market (2004 to present)

Where else has this pattern repeated?

In a way, this is an industry-level version of Clayton Christensen’s theory of disruption. Start with a product that performs poorly on existing competitive dimensions (e.g. cost per unit of energy), but outperforms on something critical (e.g. weight and longevity for solar PV in space). Leverage that differentiated advantage to grow volumes and bring down costs. Rinse and repeat until you’ve taken over the world.

One of the most well-known examples from an adjacent industry: Elon Musk’s ‘Secret Master Plan’ for Tesla released in 2006.

In it, Musk outlined how Tesla would use its Roadster to serve the niche, high-end market for sports cars with high torque electric motors that could beat cars like Porsche and Ferrari. From there, they would move on to a more mainstream 4-door vehicle (which became the Model X and Model Y) before ultimately releasing a mass-market vehicle (the Model 3)5.

This iterative progression from niche to mass-market enabled Tesla to become the largest EV manufacturer in the US, even if its global leadership is now being eroded by Chinese competitors.

In this context, the path for solar PV from invention to world domination seems remarkably contingent, with the industry swinging from country to country even as costs declined steadily and volumes grew. It’s amazing that it happened as quickly and steadily as it did.

In contemplating the lessons from his own book applied to Direct Air Carbon capture (DAC), Nemet bemoans the 70 years it took for solar PV to get to today. But it’s not clear how much a perfectly calibrated technology-push and demand-pull approach can accelerate this iterative process, even if the need for climate action is as urgent as many believe.

The challenge is applying the recipe to ‘commodity’ markets like energy and carbon capture.

Jumping too quickly to scale can end in tears, as the Cleantech 1.0 boom and bust showed.

In this sense, the framework can also provide a useful razor to evaluate emerging technologies. Those with a clear ladder from niche to mass markets may have a shot, those trying to jump straight to mass market are taking a riskier, more binary bet.

Among nuclear fission startups, for example, this lens would favor folks like Antares and Radiant Industries targeting micro-reactors for military and off-grid applications, or mid-scale players like Last Energy (~20MWe) or Aalo Atomics (~50MWe) targeting datacenters. Others like Terrestrial Energy (~380 MWe) or NuScale (~924MWe) are more likely to struggle since they can’t help but compete immediately with existing mass-market alternatives.

For DAC, it’s not obvious there is a natural niche buyer equivalent to the industrial niches that powered solar PV beyond space applications. Highly-interested corporate buyers like Microsoft and Stripe can create a flourishing of early stage startups, but without the next tranche of buyers the market will falter. Taking this lens, it may make more sense to bet on technologies that serve industrial CO2 users first with a clear path to broader use, even if they can’t immediately scale up to a million tonnes per site.

Simple isn’t easy and laddering up from niches to world domination is no surefire thing, but at least it’s possible as shown by the growth of solar PV over the last 70 years. That said, the path will likely take longer than many hope, particularly for climate technologies, and investors and operators should orient around milestones and markets on the way to their bright vision, lest they come up empty-handed.

Further reading:

How did solar power get cheap? - Brian Potter

• Part 1

• Part 2

The deal reflects an unavoidable reality: when it comes to innovative infrastructure, startups often can’t wait for 3rd-party developers to deliver bankable projects and must become developers themselves.

In this post I’ll walk through why Electric Hydrogen had to make this acquisition now, and why this pattern repeats across similar sectors.

Subscribe now

Thesis: 3rd-party project development requires maturity

Electric Hydrogen has secured ~$798M in funding to-date2, and are already deploying their systems at sites like Infinium’s Roadrunner eFuels plant in Pecos, Texas. They could have just worked with Ambient Fuels if they wanted, along with any other developer.

So why make this acquisition?

Electric Hydrogen’s customer pipeline likely couldn’t deliver the project volumes they wanted3 due to a lack of viable 3rd-party hydrogen projects. Electric Hydrogen thus needs to develop their own projects to keep momentum going and prove out their technology.

Without this, they were likely to succumb to the global headwinds against hydrogen; with this, they have a shot of pushing through the valley of death to the other side.

To illustrate why this happened, we can look at a sector where 3rd-party project development works relatively well: utility-scale solar.

For large solar projects in the US, project developers4 will identify and secure land for their project, design the system (at increasing levels of detail), shepherd the project through permitting (including environmental review), apply for interconnection (and pay for grid upgrades if necessary), select an EPC5 for construction, procure the underlying equipment, negotiate a power-purchase agreement (PPA) with a utility / independent power producer / corporate buyer, secure financing (including tax credit transfers) from project financiers (based on the projected costs and secured PPA), line up insurance on the system (as necessary beyond warranty), oversee construction, and turn the system on.

That’s a lot of activities, but they are pretty specifically not betting on whether the panels work.

Solar PV is a mature technology. The panels have predictable performance over a 20-year life6, and can be secured from any number of competitive suppliers. The PPAs are negotiated to match the asset life. The EPCs have experience doing this work, and are willing to take on lump sum, turnkey contracts with wraps. The asset managers and pension funds that back these projects know what they are getting from a risk and return perspective7.

Importantly, neither a shovel of dirt is turned nor the bulk of the money committed for a project until the project has reached ‘final investment decision’ (FID).

Nothing gets built until every box is checked. The design, permits, land options, interconnection studies, PPAs, EPC agreements, and capital stack are all known in advance of the commitment of ~80-90%+ of overall project costs. This minimizes absolute risks, while allowing for capital-efficient developers to drive the process.

In utility-scale solar, project development is a business of predictably manufacturing assets with a known and desirable set of characteristics, not of taking technology risks8.

Antithesis: Innovative projects misalign incentives

How does that model translate to sectors that are seeing technical innovation?

Not particularly well, and that’s a problem for the Electric Hydrogens of the world.

Folks like Electric Hydrogen (and their VC backers) would prefer to focus on scaling and deploying their breakthrough innovations, not chasing down permits or selecting EPCs.

The latter was likely deemed ‘outside their core competency’9.

It likely seemed risky enough to develop a series of breakthrough innovations, design something as complex as a ~100MW PEM hydrogen electrolysis plant, make the design repeatable and scalable, manufacture the underlying equipment and subcontract for key balance of plant components, and oversee deployment and integration at customer sites.

Having to actually develop the projects is a completely different set of skills, and there are alluring examples like solar near-to-hand that show those skills profitably existing within distinct entities.

Unfortunately, it just doesn’t work. Worse, it actually can’t work due to the incentives involved.

The challenge here is threefold.

1. Risk-averse backers

Project finance and infrastructure investors are in the business of, to a first approximation, never losing money. Unproven technologies are risky, and the bar for a technology to become proven is closer to ‘has already been deployed 3+ times in this configuration at this scale’ than anything that might seem reasonable to the uninitiated.

2. Asymmetric upside

Under conditions of technical innovation, the upside for the project developer and the project-level backers won’t match the the technical innovator, creating a problematic misalignment of incentives.

The first few facilities will be incredibly valuable for the technical innovator, because they (hopefully) prove out the new technology and the learning curves that had to-date been just ambitions. But under normal conditions, a 3rd party developer and project backers don’t see any of this upside. They are limited to the returns from their individual project. Those returns can still be high if the technology truly is a breakthrough, but it’s generally the difference between a 18% and 30% return, whereas a startup might see their value double or more overnight with the successful deployment of a first-of-a-kind facility.

For project financiers, chasing that higher return in the face of technical risk can feel like picking up pennies in front of a steamroller: not worth it.

This is true even for in-house development by a build-own-operate player with a strong balance sheet. Classically, this is the model used in developing deepwater or otherwise remote oil reserves. Only the Exxons and Equinors of the world can carry the complexity and risk of developing big, technically innovative projects. But even balance-sheet developers face the same asymmetry, and true arms-length deals are relatively rare.

3. Paradox of promise

Perhaps counterintuitively, in shifting commodity markets like hydrogen and SAF it can be just as hard for project developers using proven technology if someone else is promising technical innovations.

Why would anyone sign a 10-year offtake agreement for low-carbon hydrogen at $4.50 per kg today if there are multiple other players promising $3 per kg in ~2-5 years? In a commodity market, if the latter turns out to be real it will disadvantage the early movers.

This paradox of promise from competing approaches, even if unproven (or unrealistic), can gum up the customer pipeline for developers like Ambient Fuels.

—

Take all this together, and 3rd-party project development works in conditions of technical maturity, as in utility-scale solar. While the technology is still being proven out, the 3rd-party model is hard to square.

The upshot: Electric Hydrogen risks having no customers, no matter how good their offering in isolation.

Synthesis: Internal project development for innovation

So Electric Hydrogen had to act. Projects weren’t maturing fast enough to be their customers, and so they needed to do it themselves.

In order to control their fate and prove out their approach, they need to develop their own initial projects. In parallel, 3rd-party developers like Ambient aren’t particularly viable without access to differentiated technology, and are ripe for the picking.

In this context, the acquisition of Ambient makes perfect sense.

The combination strengthens Electric Hydrogen’s development capabilities immediately, and allows them to maintain forward momentum during a critical development window before the current 45V begin-construction deadline on Dec 31, 2027.

Ambient Fuels (and their backers) gets access to a leading hydrogen technology that can be slotted into their projects, allowing them to deploy their capital against a proprietary stream of high-potential projects, while also maintaining exposure to upside as the tech gets proven out.

The biggest surprise is that it took so long.

Electric Hydrogen's acquisition of Ambient won't be the last. Watch for other climate startups to follow suit, or be acquired themselves.

Further reading:

Unlocking Capital for Climate Tech - Breakthrough Energy Ventures

First-of-a-Kind series by Sightline Climate

I am of course talking about batteries.

Because of the dramatic fall in lithium-ion battery prices over the last 10 years, they are starting to show up in all sorts of unexpected places, and this has important implications for how the future energy system will work.

Today, I’ll briefly outline the 4 places batteries are being integrated into the home, and try to reason through how this could continue to evolve.

But first, why is this happening?

First, people really don’t like losing power or the use of their appliances. As homes get more electrified, there is increasing desire for outage protection or other forms of backup power so that a wind storm doesn’t knock them back to the 1800’s.

Second, there is an increasing economic incentive to shift power usage throughout the day. Since the grid is built for the peak, many utilities have implemented steep Time of Use rates to disincentivize power use during peak evening periods. Traditionally, homeowners have 'solved' this problem by e.g. turning off their A/C or using their dryer at other times, an inconvenience.

But now, batteries give homeowners more options to manage their bill without changing their lifestyle. This is generally a side-benefit, with systems ‘oversized’ for performance and backup power reasons, but an increasingly important one.

The upshot is that we should expect increasing proliferation of batteries in the home across 4 distinct points.

1) In-line backup power (e.g., Pila)

Ever worried about the food in your fridge rotting during a power outage? Buy a battery from Pila Energy and worry no more.

These batteries are designed to sit in between your home wiring and key loads (e.g., a refrigerator) providing device-leve backup power in case of an outage.

But, at a cost of ~$1,299 per 1.6 kWh unit, or ~$800 per kWh of capacity, these batteries have the potential to be cycled more frequently to offset peak evening use. Assuming a full-size fridge draws ~100 W per hour of use, cycling this daily during a peak periods (and charging during off-peak periods) would save a homeowner ~$30 per year1 or ~$300 over a ten-year appliance life.

The big benefit though is the ease of set-up. No installation or home rewiring required for backup power of key devices. The bill management is just gravy.

2) In-device backup power (e.g., Impulse, Copper)

Boil water in second? Check.

Complete daring feats of culinary wizardry like frying eggs with no oil or tempering chocolate without burning? Check.

Continue to work in a power outage? Check.

The cooktop from Impulse Labs is the leading example of our next type of battery: those built into devices themselves.

While this is foremost for performance reasons (few household 240V circuits can deliver the 10kW of power needed to boil water that fast), these batteries also provide backup power in the even of an outage and, you guessed it, the ability to avoid energy charges during peak hours if so configured.

Assuming one uses ~1 kWh while cooking each evening, the savings come to ~$75 over the course of the year, or ~$750 over a ten year appliance life2.

The main benefit here is the brilliant cooking itself, with the backup power and bill management potential happy parallel benefits.

A brief interlude on home wiring and the fire code

In European countries like Germany, homeowners are allowed to install up to ~800 W worth of solar panels on their balconies or other outdoor space without wiring changes. The panels include AC microinverters and just plug back into the normal home wiring.

This is not allowed by the National Electrical Code (NEC) in the US, though there has been some effort to allow this in some parts of the US (e.g., Utah).

As a result, the in-line and in-device batteries like those from Pila and Impulse are limited to off-setting the specific loads they're connected to. If this were to change, and they could feed power back into the house to offset other loads, the opportunity for and value of these types of batteries could be significantly higher.

3) Whole-home battery storage (e.g., Tesla Powerwall)

The next category is what most people think of when they think of home energy storage: stationary batteries that connect through a home's electrical panel to support household loads.

There are different specific system architectures—which may involve a 'smart' electric panel or not, paired solar panels or not—but in general these batteries are designed and sized to handle loads across a household through existing wiring.

These systems tend to be larger-still than device-specific batteries, in the ~5-15 kWh range3, so that they can provide 1-3 days of outage protection if necessary. Given this capacity, these batteries provide another powerful way to manage a home's energy use around peak periods.

While the cost per kWh is often lower (in the ~$500-1,000 range4), these systems are more expensive in absolute terms because of their size and their need for professional installation by a skilled installer5.

Assuming one uses ~3 kWh during each evening peak, the savings come to ~$220 over the course of the year, or ~$2,200 over a ten year battery life6.

This value will vary more than in the previous cases because most systems like this are: 1) paired with solar panels, meaning their charging cost is lower; 2) set up to feed power back onto the grid. The value then will depend on the specific system architecture and the programs a given utility or electricity provider may have in place.

4) EV bi-directional charging (e.g., Tesla Cybertruck)

The final category of battery also sits in one's garage, only on wheels. EV batteries are the biggest of the bunch, ranging from ~57 kWh for a base Tesla Model 3 to ~212 kWh (!) for a Hummer EV, as I've written about before.

Unfortunately, right now only a few models (e.g., the F150 Lightning, the Tesla Cybertruck) are equipped for bi-directional charging (when equipped with the right house-side equipment). This gap exists for a host of reasons, including uncertainty that should be resolved in the 2026 National Electric Code, but I fully expect wider deployment of these features going forward.

As with the other batteries, to-date these systems have been marketed for outage protection but can provide the appropriate bill management if configured correctly, with similar value at stake as the whole-home example above.

The EVSE path has a couple of big benefits: 1) people are buying EVs anyway, this bill management is gravy; 2) manufacturability of cars creates more space for learning curves / less to install with custom local labor.

The value created should more than pay for the incremental cost of the equipment with these capabilities, with the absolute size of the batteries creating large resources for grid support or other value-added services.

So who's going to win?

I think there's a role for all of these, and they are not mutually exclusive. One might have a whole home backup through a Tesla Powerwall and still want the high power of an Impulse stove, or use their F-150 Lightning for bill management but still want their Pila battery to ensure backup power for their fridge.

That said, for pure scale of impact I would bet on the devices that have benefits beyond backup power (e.g., Impulse, EVs) and provide peak shaving as the cherry on top. These will get more traction, faster, from customers less focused on the pure economic benefits, while leaving the door open for big benefits down the road.

The impact is already being felt, with individual customers getting better stoves, faster cars, and more resilient home electricity. In the future, these batteries will come together to have a meaningful impact on the grid as well.

I've focused primarily on hardware architectures today, as much to clarify the archetypes for myself. In a future post I hope to dig further on the business models and software that are driving these systems, and where the incremental opportunities could be.

Subscribe now

For decades now, renewable energy has been subsidized in one way or another, in one market or another1, while it has progressively reached scale and seen costs tumble down. The era of subsidies is now ending, and solar PV and wind energy in particular will need to compete on their own merits.

Fortunately, it is likely that solar PV and onshore wind2 in particular have already reached escape velocity3 in key markets and applications. While the future pace and specific types of deployment will matter desperately to many existing companies (some of whom will fail), the overall trajectory is clear.

Renewables are staying in the US energy mix, and the coming shakeout will create opportunities for those positioned to thrive in this altered context.

So what is going to happen?

Solar PV and wind energy deployments will reset

The phaseout of the 48E / 45Y tax credits for solar PV and wind has been accelerated to qualify only systems ‘in construction’ by July 4, 2026 or in service by the end of 2027. These in-service dates are tight for large projects which often sit at the mercy of interconnection queues and permitting for their specific construction and in-service dates. Big projects that will qualify in time are already in-flight.

Safe harboring could extend the ‘in-construction’ path somewhat, but recent Treasury guidance has sought to clamp down on this practice by requiring ‘physical work’ to claim the beginning of construction. Previously, purchasing ~5% of the project’s value in e.g. equipment upfront was sufficient to safe-harbor a project for years and still receive tax credits. With these changes, this approach is riskier and more expensive, focusing on actual site specific work (like laying foundations) that often comes after permitting and interconnection approvals.

As a result, over the next ~18-24 months the project pipelines will likely soften in key segments. This will put pressure on developers and technology / service providers that are leveraged operationally to a certain rate of deployment. There will be cancelled projects and likely layoffs as participants adjust to the new shape of the market.

The only silver lining is that this will happen relatively quickly. If the safe harbor provision were kept as before, there would be an overhang of projects waiting to get built through the early 2030’s (advantaging those developers with the capital now to invest in safe harboring their pipeline).

As it stands, pricing and project pipelines will reset more quickly, keeping everyone on a more level playing field.

Residential solar will shift to third-party ownership

Residential solar is generally bought in one of three ways: outright in cash, with the proceeds of a loan the homeowner takes out, or not at all (with the system owned by a third party and leased to the homeowner). Residential solar purchased with cash or loans fall under a different tax credit (25D) than third-party owned systems (48E), with the former phasing out at the end of this year (2025), two years ahead of the 48E credit.

This difference in treatment will mostly hurt the independent installers that form the long tail of the solar industry, the ‘mom and pop’ shops of the residential solar world. Big players like Sunrun4 already deploy primarily ‘third-party owned’ systems (TPO) that will continue to receive tax credits through 2027 (or beyond with safe harboring5). They will be at an advantage and likely gain share at the expense of smaller installers.

The securitization and support infrastructure that underpins third-party ownership is non-trivial to spool up quickly, but I expect to see more installers shifting in this direction (alongside a large-ish contraction in the industry as installers exit rather than adapt).

Software and financial partners able to facilitate this shift should see demand.

Battery deployment will continue to accelerate

The tax credits for energy storage have been maintained, and the boom in batteries will continue. Particularly as the pipeline of new generation from solar and wind slows (and all the gas turbines go to data centers), batteries of all scales will be increasingly valuable in shifting energy around demand peaks.

More tactically, the differential treatment between storage and generation will create interesting project-level design dynamics. For example, most solar PV systems today are actually solar + storage with batteries co-located to shift energy to higher value evening hours.

It is not yet obvious how balance of system costs and soft costs (e.g., sales, design) will be allocated across the two, with the potential to shield much of the overall system costs under the energy storage tax credit. We already see companies like Sunrun repositioning themselves as energy storage companies (that also happen to deploy solar PV).

The wild card here will be the Trump administration’s ramp up of Foreign Entity of Concern (FEOC) requirements in tracing the provenance of system components. Given how reliant the global battery supply chain is on China (even for cells and modules assembled in the US), this could create pockets of tightness in the market as supply chains are reworkd.

Batteries, energy storage developers, and the software to manage them all will see increased demand over the coming years.

Software (broadly defined) will become more important

Because we’re making it harder and more expensive to add new generating capacity, we will have to get more out of the assets we already have.

This will mean investments in software to better orchestrate and maintain those assets, and refined market design and rate structures to incentivize grid-compatible behavior.

This will take many forms, from virtual power plant (VPP) and demand response (DR) platforms to vegetation and asset management tools to new tariffs focused on peak load. Market changes will be slower, and unfortunately no region will look like Texas any time soon, but expect continued plodding improvement.

These trends are already very much underway on all fronts (it’s not like it was easy to build infrastructure before), but it will get a boost relative to the counterfactual of IRA continuation.

Where does that leave us?

Perhaps surprisingly given the rhetoric on all sides, we find ourselves with mostly the same trends as last year, just weakened in spots.

Renewable energy, and solar in particular, will continue to be deployed. Batteries will continue to be deployed. Software will continue to be necessary to make the whole, increasingly decentralized system work. Nuclear and geothermal will get their chance. Coal is not coming back in any sustained way.

To quote Churchill6, “this is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning.”

Yet an ‘all of the above’ approach to energy dominance—which the US should undoubtedly pursue—has plenty of room for the continued growth of otherwise ‘green’ technologies like biofuels. Along key vectors, these energy sources are simply better than the alternatives.

What follows is a bit tongue-in-cheek, but there are important non-sustainability reasons for continuing to invest in biofuel production, particularly for liquid fuels like methanol and jet fuel.

America is an unmatched energy superpower.

We are the global #1 in oil production, #1 in natural gas production, and #2 in electricity production.

You might wonder then why the Trump administration is so focused on “energy dominance”? They realize that cheap energy is core to human flourishing. You can never have ‘enough’.

Put differently, America should be the shining beacon on the hill burning people’s eyes because of its absurd power output.

In this spirit, in his January 20^th executive order “Unleashing American Energy”, Trump gave us his energy priorities: “oil, natural gas, coal, hydropower, biofuels, critical mineral, and nuclear energy resources”.

Maybe you’re already familiar with these changes, nodding along to yourself: “Oil? Yes! Natural gas? YES! Nuclear reactors with big beautiful American flags? FUCK YEAH!!!”

“…biofuels…? Like, ethanol in gasoline?”

Yes, but also so much more.

With recent technical advancements, we stand at the precipice of a new era in liquid fuel production, one that isn’t always obvious to folks not following the space closely. We are nearing the point where we can turn things like household garbage and sawmill waste into jet fuel and methanol at costs competitive with traditional hydrocarbons.

This relies on a range of technologies, from biomass gasification to hydrothermal liquefaction to fermentation, with each converting biological feedstocks into carbon and hydrogen rich chemical precursors that can then be converted into methanol, jet fuel, diesel, and more.

As a complement to nuclear, gas, and solar energy, these will form an important cornerstone of American energy dominance. Even without their sustainability benefits, biofuels have a range of positive attributes that make continued investment a priority.

1. Supports American farmers and foresters

Next-generation biofuels—the kind that can compete head-to-head1 with traditional hydrocarbons on cost—are often made from cellulosic biomass like agricultural and forest waste (think sugar cane husks and wood chips).

Right now, this waste is of little value. Often it is just burnt on-site. Using it to make biofuels provides another source of revenue to the American farmer, one that is detached from the vagaries of global crop prices.

Even better? We can take the waste from better forest management and turn it into literal jet fuel, preventing wildfires in the process!

2. Provides energy security

The US uses ~340 million tonnes of biomass each year—mostly corn for ethanol and wood burned in specialized power plants—but we let another ~350 million tonnes of ag, forest, and municipal waste go to…waste.

This excess alone would have been enough to fully supply the jet fuel needs for the US in 20232.

Beyond this, the DOE estimates there could be another BILLION tonnes of biomass available from expanded energy crops and new production pathways like algae.

Without importing anything, or diverting existing production, we could QUADRUPLE our country’s jet fuel production with biomass.

3. Builds on US petrochemical leadership

Building billion-dollar chemical plants is not trivial, and that is exactly what is required to capture this biofuel opportunity. While not quite chip fabs, they are near-apex technologies for civilizations.

Thankfully, the US already has the engineers, operators, and developers to go do this, on the order of ~250,000 people working directly in the oil and gas industry per the Bureau of Labor Statistics3, not even including the specialized construction trades that physically construct these facilities.

Other countries don’t have this skill base, and will struggle to reap the biofuel bounty alongside the US.

4. Can be sold to Europeans for a big profit

Even if our government isn’t focusing on reducing our carbon emissions, Europe sure is.

They’ve put in place serious mandates for airlines to use ‘sustainable’ fuels, already at millions of gallons a year ramping up to ~20% of all jet fuel by 2035. Yet they can’t serve this demand themselves.

With a history of expensive electricity and expensive natural gas, their chemical industry is an also-ran. This won’t change any time soon; new plants will be slow and expensive to build in Europe. To hit their green mandates, European companies will be forced to import biofuels from the US.

We are already rushing to sell them as much LNG as we can ship out. We should be doing the same with biofuels.

—

Even without focusing on sustainability, biofuels have many advantages for the United States. They aren’t a silver bullet4, but will become an increasingly important part of our energy mix.

It is good that the Trump administration recognizes this, and continues to support this vital industry as it helps unlock a new era of American energy dominance.

This will save consumers and utilities billions of dollars, while unlocking faster electrification and decarbonization. 

But! The technologies alone aren’t enough; getting the right regulatory and business models in place will be critical to deploying these tools.

In particular, I expect approaches that focus on large loads and minimize customer acquisition costs will win out in the long run. The unfortunate corollary: some existing approaches, and the companies pursuing them, will fail.

We can see this through two examples: smart thermostats, and managed EV charging.

—

Load-at-stake: why smart thermostats are a sidequest

Smart thermostats were the first ‘smart’ devices that gained widespread adoption in US homes, and opened up the potential for space heating and cooling systems to provide grid services. Mostly these have been used for demand response (DR) programs, where a customer might allow the utility to turn off their A/C a couple times a year1 during peak load events in return for a fixed payment or bill reduction.

This was a big improvement from early residential demand-response programs that required special equipment from the utility, but the overall potential here isn’t that big. The loads at stake aren’t large relative to electric vehicles (EVs), and the ability to cycle them repeatedly is low (an A/C that shuts off every hot day is useless). People don’t like when someone messes with their thermostat.

As a result, customers expect disproportionately large payments to participate in these programs. This means that the value smart thermostats can provide as dispatchable grid assets gets eaten up by customer payments (or limited by strict program constraints). This makes thermostat aggregation a low-margin, low impact business.

Renew Home, one of the leading thermostat aggregators in the country, is aiming to collect 1 gigawatt of capacity in Texas by 2035. By comparison, Texas will see more than 6 gigawatts of battery capacity installed this year alone.

Approaches that treat smart thermostats as an add-on piece to some broader energy management system, rather than a valuable load to be controlled on its own, will likely win out long-term. The loads are too small and the customer costs too high for anything else to make sense.

Customer acquisition costs: managed charging is a feature, not a company

For many electric utilities the rapid adoption of electric vehicles is a nightmare.

A single EV can increase a household’s annual load by ~50%. Worse, if drivers do the natural thing and plug-in to charge when they get home from work, that charging load will hit the grid at the worst time–during the evening peak.

It’s no surprise then that a number of companies like WeaveGrid, ev.energy, and Optiwatt have popped up to help. 

Each seeks to leverage an EV’s (or in some cases the charger’s) APIs to control the charging behavior to avoid peak times. This can be done as part of either demand response programs (which shut off charging during peak events), or to optimize around time-of-use rates (where customers pay different rates depending on the time of day).

But these are Gen 1 solutions and have a lot of the same issues as thermostats. Where demand response is the main use case, the value gets eaten up by customer acquisition costs. Customers just don’t like their device’s behavior out of their control. For time-of-use customers, the savings can be more meaningful and direct (helping customer acquisition), but the algorithm is dead simple (just don’t charge during peak times) and leaves little room for the startups to monetize their offerings.

Instead, the end-state looks more like auto manufacturers2 working directly with software platforms like Leap and Axle Energy to provide managed charging as a feature to their customers, rather than as a standalone business.

This streamlines customer acquisition and the related costs, while boosting the overall controlled load by capturing a higher share of customers. This will create a much more effective tool for utilities to manage peak events, and for customers to save money.

It will take years for the grid of the future to take shape, but the outlines are increasingly clear.

Minimizing customer acquisition costs relative to load will be critical to winning as a DER provider or aggregator. This will require more consumer control (where they want it), and a focused customer acquisition funnel (likely by partnering with the original seller).

Getting this right will accelerate the deployment of these tools and increase their positive impact for the grid.

Thanks to Emma and Julius for reading an early version of this.

No, not Communism; utility death spirals.

Utilities were faced with growing legions of subsidized rooftop solar customers who weren’t paying enough into the grid to cover system costs. Some were even leaving the grid entirely (otherwise known as grid defection). Their share of system costs would then fall on non-solar customers, causing more people to switch until the utility’s business had been irreparably eroded…

Except this didn’t come to pass, mostly because defecting proved too costly for customers once regulatory supports were reduced.

Rooftop solar was relatively expensive, and only produced power during the day. Batteries to make the system work 24/7 were even more expensive. Regulations were restructured to avoid unduly favoring renewables. As a result, regulators and utilities mostly stopped worrying about grid defection and utility death spirals.

But Hollywood loves a sequel, and the specter of a utility death spiral has returned to haunt California.

In fact, for some households grid defection is currently the economic choice. If recent trends continue, it will only become more viable for more households over the coming years. 

The time is now ripe for Californians to start leaving the grid.

On one hand, this has potentially dire implications for local utilities. Grid defection at scale makes a utility death spiral increasingly likely. 

On the other hand, this is an exciting moment. The cost of renewable energy has gotten so low, and the efficiency so high, that a household can reasonably cover its needs from onsite generation alone. No more power outages, no more rising utility rates or impenetrable bills, just clean steady electricity from the sun in the sky (and the battery in your garage).

True energy independence is within reach for many Californians.

Share

Why now and why California?

The price of solar photovoltaic (PV) modules—the panels in a rooftop solar system—has fallen dramatically as global production has ramped up.

Residential solar prices have fallen alongside (albeit not as quickly due to the large drag from soft costs like customer acquisition, which I dug into here).

Battery cell prices have also dropped dramatically, though that hasn’t yet fully translated to a drop in final installed costs for residential customers.

While solar costs are going down, residential electricity prices have been going up, particularly in California.

Importantly, these are nominal dollars1 so this includes broader inflation. But that makes the drop in solar and battery costs even more impressive. They have been dropping even through increases in the overall price level. Utility rates, however, just continue to rise.

In fact, California now has some of the highest residential electricity prices in the country (highlighted in red below).

At the same time, California is rich in solar resources. In the below map—which shows solar irradiance that reaches a horizontal surface from direct, refracted, and reflected light—California is at the left end of what is effectively a solar bullseye in the US southwest. In the red areas, more output is possible from a given set of solar panels.

Other states in the southwest have lots of sun, but their current electricity prices are low enough to beat off-grid setups. Some in the northeast have high electricity prices, but their (relative) lack of sun make them less appealing for solar-backed grid defection (though they could be next as equipment prices come down.

Taken together, we see that California has a unique combination good solar resources (high solar irradiance, shorter stretches of cloudy days) and high electricity costs that make it a prime target for grid defection 

Interestingly, Hawaii will likely face the same defection, but right now its island location makes the cost of solar and batteries more expensive, and its solar resource worse2.

It's important to note that expensive electricity isn’t necessarily the fault of California’s utilities. In recent years they have had to deal with a lot, including:

• High cost of natural gas generation due to limited local gas production and pipeline access

• Aging equipment and systems that were built out rapidly in the post-war years

• Rapid return of load growth (growth in demand for electricity) from electrification

• Increasing wildfire risks, and increasing costs associated with mitigating these risks

• Increasing climate risks from flooding and heat3

• Challenges hiring and retaining craft labor (e.g., linemen) in key metros due to rising cost of living4

• Explosion of grid planning complexity and interconnection study volumes in the face of all of the above

These are all hard to deal with, particularly when utilities are constrained by a century of regulation and (justifiably) conservative operating cultures.

But customers don’t really care. They just want cheap electricity that comes out of the wall when they expect it. If there is a better and cheaper option, people will choose that.

The household economics of grid defection

It can be easy to talk in sweeping abstractions about straight lines on graphs. Let’s talk about a specific household to illustrate the dynamics at play.

Similar to my previous article on solar net billing, I’m going to use a house in Temecula as a proxy for the type of ‘SoCal suburbs house’ that is a prime customer for solar + storage (and since we need a physical site for the solar resource estimation, which I’ve done using NREL’s PVwatts API5).

Against this, we can draw normalized hourly load data from Southern California Edison for 2023 for a single family residence. The annual demand is ~6,200 kWh, or ~17 kWh per day, which we’ll be aiming to power on an hourly basis across the year.

To serve this load, we will build a system comprised of three things6:

1. Rooftop solar panels: ~$2,300 per kilowatt of capacity

2. Stationary battery storage: ~$1,000 per kilowatt-hour of capacity, charged by the rooftop solar

3. Backup storage / generation (either a diesel generator or an EV with bi-directional charging): ~$2.50 per backed-up kWh

For our system, we’ll look at 3 scenarios: a baseline built on the unaltered hourly load data, and two scenarios that include added electric vehicle charging load (given the ongoing shift to EVs) with different times for charging (day or night). Additional key assumptions are outlined in the footnotes7.

How much will this off-grid system cost?

Running a quick optimization, we get the following sizes and costs:

First, note that these systems are expensive! The absolute cost (before subsidies) is pretty large, on par with a new car or moderate home renovation. This is why ~70-80% of systems are financed or third-party owned.

Second, these systems aren’t that big in the context of a normal residential solar installation, which averaged ~7.4 kW in capacity in 2023 (per Lawrence Berkeley National Lab). Our off-grid systems are overbuilt relative to the expected load, but we wouldn’t expect issues fitting the needed solar panels on most roofs, or putting the batteries in most garages. The big difference in going off-grid is the inclusion of larger batteries.

Finally, a single EV adds a lot of load above our baseline, and the timing of its charging is pretty important to the overall system design. While charging at night is fine for grid-connected EVs, charging during the day (while the sun is shining) is much more efficient for our off-grid system. This makes specific customer behavior important to the final decision on defection.

But how does all this compare to what we might pay to the local utility?

Below we can see the cost-per-kWh across our different scenarios, with the height of the bars representing the cost per kilowatt-hour for energy during the first year of installation8. The off-grid costs are shown in blue/red/green/white and the utility costs in grey.

The cash costs in blue/red/green are lower than the utility costs, so if you’re angry with your electric utility and have some spare cash lying around, go off-grid! You can likely see bill savings today.

But around 80% of residential solar + storage systems aren’t bought outright, they’re financed. Once the cost of financing is included (the grey outlined box, based on financing rates of 8%), it isn’t economical to go off-grid just yet.

This is consistent with recent research from Western University that finds economic grid defection viable in multiple California locations before accounting for financing costs.

This level of cost parity is enough to entice some to switch, but likely isn’t enough to start a wave of defection.

Can this change in the near future?

Imagine that over the next 3 years, utility rates continue to grow at their recent historical trajectory of ~3% per year above inflation, battery costs come down to ~$600 per kWh, solar PV costs fall to ~$2,000 per kWh, while financing rates fall to 6.5%.

Then we end up with the following:

Going off-grid becomes immediately economical, and would only become more attractive as utility prices keep marching up9.

In this scenario, our model homeowner could expect to save ~$15,000 over the 25 year life of their system by going off-grid.

Is it realistic to expect such a future?

I’m most confident in the coming battery price declines. $600 per kWh is what you’d pay for the battery in a new Tesla Model 3 (and you’d get the car for free10). Cell prices for lithium iron (LFP) batteries are approaching $50 per kWh; installed battery prices are going to come down.

We’re already starting to see this with Tesla’s Powerwall 3 expansion units. These 13.5 kWh units rely on the inverter in a lead Powerwall 3, allowing them to avoid duplication of equipment for larger systems with multiple battery units. With this simplified architecture, these add-ons are priced at ~$500 per kWh (before installation).

The cost of solar PV should also come down, but at this point the panels themselves make up a minority of total costs. Bringing down soft costs will be slower, particularly given the challenges in early customer acquisition for an off-grid system.

For reasons already mentioned, utility rates are likely to rise at the level of inflation or above. The appeal of off-grid systems will hinge on how much ‘above’ those rate increases end up being. Increases consistent with the last few years will quickly bring off-grid systems into viability; only sustained price discipline will keep this risk at bay.

The path of interest rates is hardest to forecast. Rates have come down from recent highs, but it’s not obvious where they will settle. Much will depend on the policies of the incoming Trump administration.

Each of these factors will impact the speed at which grid defection will occur. But the overall trend is clear.

Who will shepherd Californian’s to energy independence?

Sunrun and Tesla

Residential solar companies like Sunrun and Sunnova are built to sell and service residential energy systems, and I expect them to lead the broader rollout of these off-grid systems once the economics work11.

Sunrun in particular already sells most of their systems as part of a subscription. Once that subscription (and the underlying system) can cover all of the energy the customer needs, it may not make sense to pay for a grid connection anymore. The subscription model also allows Sunrun to take advantage of Investment Tax Credit (ITC) Adders that further reduce the effective system cost.

I expect them to start with their existing customers (for which they have extensive data and a pre-existing relationship), piggybacking on a repowering12 or system upgrade to go off-grid entirely.

These efforts will be boosted by the broader adoption of bi-directional charging capabilities for electric vehicles (EVs). When paired with the right equipment, this allows an EV to power a home just like a stationary battery might. This can significantly expand the effective battery capacity of a system, either reducing the cost (by allowing a smaller stationary battery) or eliminating the need for a backup generator13.

Right now, this capability is only really found in a few trucks like the F-150 Lightning and the Tesla Cybertruck, but the expectation is that additional models with bi-directional charging are coming in the next few years (potentially alongside code refinements in the 2026 National Electrical Code to formalize rules around this).

On this front, I expect Tesla to lead. They have the vehicles, and an existing home energy business from their SolarCity acquisition. They are well-placed to orchestrate an off-grid shift, and I wouldn’t be surprised to see Musk pitching homeowners on a path to energy independence.

Tesla Powerwalls are already being used in off-grid-ready systems like the one shown below.

Once either Sunrun or Tesla start to offer off-grid systems at scale, the rate of grid defection will accelerate markedly.

California grid defection: when, not if

So, we’ve established that economic grid defection is already possible in California on a cash basis. As solar and battery prices continue to fall, and utility rates rise, this will only become more appealing over the next 5-10 years. It is a matter of when, not if.

On the question of when, there are a number of factors that could slow the arrival of household energy independence, including:

• Increases in interest rates: The effective cost of off-grid energy is heavily dependent on the cost of financing. Any increases in interest rates will make these systems disproportionately costly relative to utility rates14.

• Tariffs / tax credit reductions: On the campaign trail, Trump and his team talked extensively about increasing tariffs on foreign-made goods and removing parts of the Inflation Reduction Act (which provides generous tax credits to solar PV and battery projects). If these changes come to pass, this would significantly alter the economics of going off-grid.

• Other regulatory changes: Once it becomes clear that grid defection is a serious risk for local utilities, I expect there to be some sort of regulatory response (positive or negative).

  One potential avenue: air quality regulations. California regulates the use of installed diesel backup generators15 if grid power is still on, but it’s not clear how this applies to homes without a grid connection. It wouldn’t be hard to make one kind of off-grid system design illegal for ‘air-quality’ reasons, though this wouldn’t preclude homes from using a bi-directional EV for backup.

The interesting wild card is consumer appetite. While I’ve centered the arguments for defection around economics, for many people there are other considerations.

For some, there is clearly appeal in independence and self-sufficiency; these are deep-rooted American values after all. Homeowners frustrated by ‘public safety power shutoffs’ every Thanksgiving16, or worried about a multi-day outage following a major earthquake, will be more willing to buy some peace of mind with an off-grid system. Decreasing prices make this option more available to more people.

On the other hand, many will see risk in an off-grid system. Going off-grid shifts electricity from something that is (basically) always available (even if expensive) to something they need to think about. No amount of probabilistic weather modelling will overcome some people’s anxiety about a long stretch of cloudy days, in the same way that some have ‘range anxiety’ around electric vehicles, even if they only ever take short trips.

This will be overcome eventually, just as range anxiety has been allayed by larger EV batteries and more numerous charging stations, but expect early adopters of off-grid systems to be embracing adventure and independence as much as they are seeking lower power bills.

The future rests on California homeowners

What happens to our existing utility system once grid defection gathers pace?

Honestly, I’m not sure.

In many discussions of utility death spirals, much hinges on the elasticity of consumer defection. Say 10% of customers leave, and half their costs are fixed and get piled onto the customers that remain, making the latter’s bills go up by ~5.5%. If that’s a big enough increase for the next tranche of customers to leave, then they will, and the spiral continues. If not, then the system stabilizes at a new equilibrium.

This is now complicated by ongoing price declines in solar and battery equipment. The current process doesn’t rely on a feedback loop with utility rates to cause people to defect. To the extent that the systems keep getting cheaper, more people will leave each year whether or not the death spiral effect holds.

This makes defection hard to stop among those with the resources to leave.

But there is a limit to this process. Some people and businesses will not be able to leave at an acceptable cost, for example because they lack the space for enough solar panels to cover their demand. 

An optimist might argue this group is large, and that grid defection will ease load growth pressure on the distribution grid. This will free up capacity to provide more energy to these remaining customers, growing total energy usage overall.

A pessimist would predict a slug of stranded customers paying swollen bills17 to keep the remaining grid infrastructure operational after everyone else defects. This would fall disproportionately on urban residents and commercial customers; large industrial customers would move18.

Ultimately, this is an empirical question. Over the coming years, we’re going to find out which is right. My hope is that California’s utilities and regulators rise to the challenge, and this strengthens the system as a whole. If not, the long-term impacts are decidedly murkier.

In their last energy crisis, Californian’s recalled their governor. Pacific Gas & Electric has already gone bankrupt once from wildfire damage. Should it come to pass, the disruption from grid defection could prove more painful than either.

For individual homeowners though, the future is bright. Thanks to the declining cost of solar panels and batteries, they can seize true energy independence for themselves. Over the coming years, I expect many Californians will do just that.

Further reading:

• S Sadat, J Pearce (2024): The threat of economic grid defection in the U.S. with solar photovoltaic, battery and generator hybrid systems

• Navon et. al (2023): Death spiral of the legacy grid: a game theoretic analysis of modern grid defection processes

• Rocky Mountain Institute (2014):The Economics of Grid Defection

Dynamists put their faith in experimentation, ongoing competition, and iterative evolution towards ‘the good’ without knowing exactly how we’ll get there. They value ongoing progress but can come off as a bit simplistic when pressed on the details of how their policies will lead to flourishing.

By contrast, Stasists seek stability and order—either through a return to a previous ideal period, or through technocratic control of ongoing change—in pursuit of the ‘one best way’ to organize society. They often deride the Dynamists as naive and out of touch with human nature, believing that without rigorous control we risk spiraling into chaos.

Now, which best describes the culture of the electric system?

No points for guessing: the mindset of most electric utilities and regulators is deeply Stasist and technocratic.

Arguably, this is for good reason! The traditional grid hierarchy—large central generators and naturally monopolistic distribution managed by a central operator—practically begs for stability and top-down control. The alternative risked grid collapse or rolling blackouts from mismatches in generation and load.

This centralized structure further bolsters calls for strong regulatory oversight. ‘How do we know this monopoly utility isn’t just padding its pockets? We need a legalistic process of regulatory oversight to protect people’.

Virginia Postrel, who first minted the Dynamist / Stasist framing in her 1998 book “The Future and its Enemies”, said as much in pointing out that:

[Stasist technocrats offer] complex regulations to make the world simple. Through a million tiny, specific rules, decreed and adjusted to cover each new situation as it arises, they promise to mold the future to the one best way. (p. 113)

But now, the world is changing, and the Stasist paradigm is struggling to keep up.

Load is growing again, often in hyperlocal pockets that strain both generation and distribution infrastructure1. Generation is becoming more distributed, with new energy technologies like solar PV and batteries upending the traditional logic where large central generators were most efficient. System complexity is growing rapidly.

The embedded Stasist paradigm, with fossilized layers of process and regulation, has made it hard to respond to changing customer needs and to make use of new technologic affordances. The result is slow change and rising costs.

The original technocrats simply wanted to manage change. But the apparatus they created provides a million ways to stop it altogether. (p. 23)

If this is the status quo, and it isn’t good enough, how can we inject more dynamism into the system and accelerate the energy transition?

Two recent conferences, the Progress Conference and DERvos, show what a Dynamist approach to energy could look like, one that emphasizes experimentation, broad rules, and competition to accelerate progress.

Share

Progress Conference 2024: Thinking Big

Hosted at the beautiful Lighthaven campus in Berkeley, the Progress Conference was far-reaching and exploratory, ranging across biology, urbanism, AI, energy, and regulatory renewal (and weather, self-driving cars, mechanism design, religious-hued sci-fi, and much more). Virginia Postrel herself even gave a talk.

A big part of the magic of the event was the architecture of Lighthaven campus itself. Comprised of multiple buildings and in-between nooks and courtyards2, the space itself felt ambition-expanding in its allowances for unscripted side discussions (that sometimes expanded to 40+ people).

Within this constellation hung interviews by Dwarkesh Patel with Tyler Cowen and Patrick Collison—the two progenitors of the ‘progress studies’ movement with their 2019 Atlantic essay “We Need a New Science of Progress”—where they shared their perspectives on progress.

But what can this tell us about energy in the more literal sense?

Within the multiple talks on energy abundance and related topics (e.g., permitting), discussions at the event demonstrated a key Dynamist pillar: allowing for experimentation / alternative paths to progress, and a forbearance on specifics. To be more concrete: where disagreements existed, people mostly avoided saying “X won’t work”, in lieu of things like “we think Y is most likely to work because of A and B”.

This is important, because the bear (pessimistic) case for something almost always sounds more intelligent than the bull case. One can sound very erudite and wise in describing why something won’t work. It sounds deeply naive to say “we’ll figure it out!” even if this latter attitude is what drives progress forward.

The benefit of this forbearance was amply demonstrated in a conversation between Casey Handmer (founder of Terraform Industries and probably the most well-known solar maximalist working today) and Bret Kugelmass (founder of the fission startup Last Energy).

Here were two men whose companies and professional reputations were tied to their chosen technologies, and who might be expected to have trenchant criticisms of the other’s approach. But in the event, they spent most3 of their time agreeing on the importance of energy abundance and the multiple paths to get there.

While one might be enamored of one technology or another (apparently lots of men love nuclear!), as a society we shouldn’t actually be trying to decide on the “one best way” to create energy abundance. The future of energy is big enough for multiple technologies to win, and it will likely look at least as diverse as it does today.

We actually need to run the experiment and see which new approaches prove most fruitful (and where).

DERvos 2024: Making It Happen

If the Progress Conference focused on the potential for radical transformations over the next 5-25 years, DERvos was decidedly more focused on the 1-5 year future of energy.

Held at the Brooklyn Navy Yard, the event brought together a growing slice of the electric industry focused on distributed energy resources (DERs) like residential solar/storage/EVs and commercial & industrial microgrids (along with the software to power all of it).

The attendees were mostly those working in the trenches of the energy transition, looking to deploy, deploy, deploy their solutions into the market today (or very soon). The vibe was very much: ‘It is still early; if you let us go faster, we can do so much more’.

Across multiple panels, much of the discussion centered around two other key pillars of Dynamism: the value of broad rules and competition.

For experimentation and exploration to be effective, people and businesses need 1) to be allowed to do things, and 2) to get feedback on their efforts (through competition).

Both are currently challenges in electricity markets across the US, where complex regulatory constructs and incumbent monopolies often block new types of activity. Even where new things are allowed, a lack of price signals and competition slows feedback and hampers the ability for better solutions to win out.

DERvos’s closing regulatory panel drove this point home, with Jacob Mays’ call for “full-strength prices and mandatory contracts4” to clear the way for innovation in electricity markets.

The logic here is that electricity is a unique commodity, and much of the market is structured around serving peak demand.

In practice, the marginal megawatt on the hottest afternoon of the year is way more valuable than on a normal day. But in many places, that value is not properly priced5. With much of the innovation and progress in electricity related to new intermittent generating tech (e.g., solar PV), battery storage, and EVs, this is a huge block on forward progress.

Where broad rules and competitive markets exist—in Texas for example—the introduction of new services and technologies has been rapid. Texas leads the country in deployment of renewables, and is seeing significant innovation in its retail energy markets. As a result, Texas captured outsized mindshare at the conference as a hopeful portent of what might be possible in other jurisdictions.

But where new entrants can’t participate in key parts of the market, then progress will be underpowered where it occurs at all.

The Path Forward: Embracing Dynamism in Energy

So where does that leave us?

A few key pillars stand out:

1. There is no “one best way” to produce electricity and create energy abundance

2. Broad rules are needed to create space for experimentation and new approaches

3. Competition is welcome, and can provide feedback on the best approaches

Embracing each of these Dynamist tenets would provide a significant boost to progress in energy, even if it is not obvious exactly where they’ll lead. We need to ease our grip on the Stasist paradigm that has become entrenched in the electricity sector, one where deterministic management seeks to minutely control the system as a whole.

The grid is becoming too complex, the requisite knowledge too dispersed to be solely planned and controlled by technocrats (whether they be public utility commissions or utility executives). Prices and competition can surface dispersed information and incentivize the best approaches more ably than any single person or body.

This may sound abstract and idealistic, but even incremental movement in a Dynamist direction will have an outsized impact on the cost and effectiveness of the grid.

The electric system will continue to require control and management beyond most other industries, but weaving in more of the Dynamist spirit will help accelerate progress towards a more abundant energy future.

Other coverage of the Progress Conference:

Dean Ball - “Be Embraced, Ye Millions”

Scott Alexander - Notes from the Progress Studies Conference

Kevin Kohler - Americans are from Musk, Europeans are from Greta

Packy McCormick - What do you do with an idea?

Lynne Kiesling - Progress and Its Enemies

Noah Smith - How long can we sustain economic growth?

Disclosure: I attended the Progress Conference as a Roots of Progress fellow, and had my accommodations and ticket covered by the Roots of Progress Institute. I will definitely be applying to attend again next year at whatever the going rate is; it was an excellent conference on the merits.

This idea is archaic, once rooted in fact but now obsolete. Its persistence skews otherwise productive conversations about the future of energy, and we need to retire the term like the coal plants that originally powered it.

Share

But let’s back up, where does the term come from?

It originates from two things: the minimum amount of load (energy) that is always needed on the grid, and the operating characteristics of traditional thermal power plants (like those run on coal).

First, it is true that there is a stable level of load that is basically always needed for a given region.

In the chart below, the green line shows the number of hours where a given load was needed in California. There are effectively no hours below ~18 gigawatts—the baseload level in red. At least this much load was needed during every hour of the entire year.

This part is still true, a meaningful amount of load is always required somewhere on the grid.

The gap in understanding comes with the second factor: belief that it is inherently more efficient to match this load with big thermal generation.

This comes from a historical contingency, where large stable loads matched very well with traditional coal-fired and nuclear power plants. These burn coal (or fission) to heat water, creating steam that then drives a turbine to generate electricity.

In both cases, getting the water up to temperature can take hours, and these plants do not handle fluctuations in demand well. They are much more efficient running steadily at a given rate. But, when operated at that steady rate, these large plants were historically some of the most cost effective available.

As a result, in most places a layered system of generators was built with large, slow-but-efficient coal and nuclear plants designed to serve the ‘baseload’ that would always be required. On top of this, faster-but-more-expensive gas, oil, and hydroelectric generators were layered.

Thus, baseload became a design shorthand within the power system, born from the economic and operating characteristics of specific generation technologies.

But that set of characteristics no longer holds.

For one, we are well on our way to phasing out coal:

In the US, this has been an economic story more than anything else: natural gas has been world-historically cheap for the last 20 years. It has become cheaper to build and run a new combined cycle gas plant (CCGT) than continue to operate an existing coal plant. Power generation has shifted accordingly.

Big coal-fired and nuclear plants are no longer the lowest cost power, requiring a stable base to operate effectively.

Modern generation technologies are also much better at ramping to match fluctuations in load.

Combined cycle gas plants can achieve partial output quickly based on simple combustion of gas, even if they still heat water to reach their full output. Nuclear reactors have also gotten better at safely following fluctuations in load. Batteries are quickly proliferating as responsive tools for managing fluctuations on the grid.

This has significantly reduced the efficiency penalty from ramping up and down in non-coal generators.

So why are we still talking about baseload?

This myth has persisted because it is satisfying.

A certain type of person will look at the distribution of load above, and think:

There is always 18 gigawatts of load, it just feels most efficient to match that with a big set of generators that run all the time.

Wouldn’t it be so much simpler and aesthetically pleasing, if we kept this layered system of load-following and peaker plants to handle loads above 18 GW?

We can design it with certainty, and be confident the generation will be there when we need it. We could even match up big zero-carbon nuclear with this base load!

But you have to actually check the relative costs, and aiming for lowest cost overall increasingly leads to a messier mix of resources.

Subscribe now

Say you have a single 100 megawatt (MW) combined cycle gas turbine plant that serves a constant 100 MW of load (e.g., a datacenter training a model to write non-rhyming poetry).

The gas plant has fixed costs of ~$150,000 per megawatt-year (e.g., capital costs, financing, overhead), and variable costs of ~$35 per megawatt-hour (e.g., fuel, maintenance).

If the plant runs at full capacity for the whole year, it would accrue total cost of ~$45.6M. Spreading this across the 876,000 MWh of energy produced, and we get an effective cost of $52.12 per MWh.

Now, say that we look into adding a 100MW solar plant into the mix. The solar plant produces ~100MW for 6 hours each day at a levelized cost of ~$20 per MWh.

The result looks like this:

Interestingly, the average cost for energy from the gas plant goes up from ~$52 to ~$57 per MWh, even as the overall cost goes down to ~$49.

This occurs mechanically because we are spreading the fixed costs over fewer megawatt-hours. This is something people seem to point at as ‘renewables undermining baseload’ resources, but the overall cost is lower and the gas plant still gets paid enough to operate. The solar plant should be built, even as it intrudes on stable ‘baseload’.

This leads to a broader point: If the levelized cost of a new generator is below the marginal cost of the existing generators, adding it to the system can lower average costs. This holds even if the headline cost of the existing generation goes up.

As a result, with new cheap generation technologies like solar, baseload is no longer a useful frame. In sunny regions, solar can be cheaper than the marginal costs of even low cost gas plants. The same holds of wind farms in places like west Texas.

To be sure, this won’t be the case everywhere, and there will be an important role for gas, nuclear, and geothermal generation as we seek to decarbonize. But it will be because those are less expensive in specific regions, not because there is some slug of baseload that needs to be served.

Instead of the layered metaphor of baseload, we need to think about a tapestry of generators that weaves in and out throughout days and seasons. This will not be deterministic–solar and wind cannot be ramped up at will–but a probabilistic tapestry.

The system will appear messy, with more volatility in pricing and more complexity in long-term resource planning, but the end result is lower cost, more abundant energy for everyone.

Clinging to the myth of baseload will not help us get there.

Further reading:

Brattle Group - Advancing Past “Baseload” to a Flexible Grid

CleanTechnica - Elon Musk Harpooned Baseload Power

Uber changed this. By recruiting drivers outside the regulated taxi structure, and implementing surge prices during periods of high demand, they were able to flex supply to better match demand. Their app-based platform also gave riders much improved transparency into the price they would be paying, and the expected length and route of their ride.

The result: a revolution in intra-city transport that is still reverberating.

We are at the beginning of a similar revolution for the electric grid.

In my last piece, I talked about how the infrastructure of the grid is built around peak load. This is the equivalent of taxis constantly circling just in case there’s a sudden storm, with all the attendant cost that entails.

Today we’re going to talk about how distributed energy resources like EVs and batteries will disrupt this model—and how a flurry of new businesses are seeking to capitalize on the opportunity.

Subscribe now

First, a quick recap on the problem to be solved.

Electricity is vitally important to modern civilization, and Americans get very angry when their power goes out or if it’s more expensive than expected.

This makes life hard for grid operators and utilities, because demand fluctuates a lot hour to hour and season to season. There are extreme spikes in demand that must be handled by the grid.

This looks something like the above, which shows the distribution of 5 years of hourly load in California1. ~87% of hours (the line in green) are below 30 GW of load, but there is a fat right tail of extreme outcomes. While ~99% of loads are below 40 GW, the peak on September 6, 2022 reached more than 51 GW.

Historically, these types of spikes have been solved by ‘overbuilding’ capacity, which creates significant costs even when done well. This model is now under increasing strain with the return of load growth and generally rising electricity prices.

But what if there was a way to better use our existing capacity by managing around these peaks?

Enter: distributed energy resources, or DERs

DERs are devices like smart thermostats, home batteries, and electric vehicles whose load or generation can be changed to help manage around peak times.

To be a bit more rigorous, we can define a DER as:

• Anything that can change the net load

• At a given house or business (behind the meter)

• In response to a control signal

This would include2 things like curtailable loads (e.g. smart devices like thermostats, water heaters, stoves, EV chargers) whose demand can be temporarily be lowered around peaks. This also includes batteries (either in EVs or stationary storage) which, with the right equipment, can power other loads in a home or feed power back to the grid.

From these basic building blocks, much can be built to manage around peak periods and lower effective costs.

The earliest examples at consumer scale (first deployed between the 1970’s and 1990’s) were control modules that attached to and could shut off home air conditioners. In return for a small payment or a discount on their electric bill, a consumer might allow a utility to control their A/C a dozen times a year during peak events. Growing up in California, my family saw programs like these as a great way to save money in return for temporary discomfort.

These programs have since matured, and have been effective at shaving the worst of the peak. But customer enrollment is generally expensive, and these programs are limited in how much they can scale up. An air conditioner that pauses most hot afternoons is an air conditioner that doesn’t work.

So what has changed? Why now?

First, there are more DERs than ever before. In 2023 the Department of Energy estimated ~30-60GW of capacity was already deployed, equivalent to ~1% of peak load today. They expect this to grow rapidly, with an estimated ~50-150GW in capacity added each year between 2025-2030.

Second, the load shifting capacity per DER is larger than ever, thanks to EVs in particular. Where a single home air conditioner might be curtailed to save ~1-2 kW of load3, turning off a Level 2 EV charger4 might save ~7-8 kW. If that EV has the capability5 of feeding power back to the house, its large 60+ kWh battery could offset the whole home for hours at a time. 

Across DER types, significant capacity is being added, with Wood Mackenzie (an energy data firm) estimating DERs could top ~217GW in potential capacity by 2028 (equivalent to ~30% of current peak load, ~10-20% of the forecast peak).

People are also more likely to agree to these DERs being used. While there’s a direct negative impact to having one’s air conditioning shut off, for most people it is easy to delay charging their EV until later at night or to use their home battery to help support the grid (while still retaining some charge in case of an outage).

Finally, information technology has advanced rapidly. Broad penetration of smartphones, embedded computing (e.g., in vehicles), and wireless networking6 now allow the control and orchestration of these devices in real-time at scale.

Suddenly, we aren’t looking at artisanal programs with great effort expended to enroll customers, install devices, identify peak events, send curtailment signals, and reimburse participants. We have fleets of devices that can shift gigawatts of load at the speed of a push notification.

This is the difference between having to call a cab to your house and hoping it will show up, and using an app to summon a car within minutes, knowing its price upfront and tracking its progress in real time. Not only is the latter a better experience, it is so much better that it gets used much more frequently.

For DERs, this will mean a shift from niche programs to platform-enabled participation in markets. The opportunity thus expands dramatically–from a dozen peak events a year to ongoing activity7 on a day-to-day basis. These activities will vary somewhat by the type of DER, but there is a role to play for each of them.

This variety unlocks a range of new business models built around DERs, with many increasingly focused on aggregation of DERs into ‘virtual power plants’ (VPPs) for market participation.

Even if the basic concept of ‘send signal to DER to help balance grid’ is pretty straightforward, the specifics are already starting to look different across regions depending on their respective regulatory structures.

In relatively deregulated energy markets like Texas, much of the consumer-scale action is taking place among retail energy providers that buy power on wholesale markets to provide to their customers.

In these markets, companies like Octopus Energy and David Energy are enrolling customers that have DERs and controlling those devices as necessary to avoid spikes in wholesale energy prices. This allows them to offer a lower effective cost of energy to their customers, while also helping to stabilize the broader grid during times of stress.

The opportunities for savings here are so dramatic that companies like Base Power and Branch Energy are even willing to front the cost of batteries for their customers as a way to unlock additional DER capacity.

In other markets like California, DER aggregators like Weavegrid and Optiwatt and residential energy companies like Sunrun and Tesla are leading the way. Sunrun for example has sharply increased the portion of their projects that include battery storage, reaching ~54% in Q3 2024. In parallel, Sunrun’s focus on subscription sales (where consumers pay a semi-fixed rate while allowing Sunrun control of installed DERs) allows them more flexibility to optimize the operation of customer DERs on the back-end.

California in particular has also been a promising market for commercial & industrial DERs and microgrids (where a facility might be able to operate islanded from grid power during an outage). This can take different forms, from a basic offering like Voltus’s that mostly leverages existing equipment, through a new microgrid from Scale that aims to offer a net savings on energy costs, to an Enchanted Rock microgrid that uses market participation to offset the cost of backup natural gas generators for critical facilities.

Finally, in regions with full vertical integration–where a single utility controls generation, transmission, distribution, and retail sales–the use of DERs at scale will likely be slower to ripen. While the devices themselves might still be deployed (some people will want EVs regardless), without the right regulatory and economic arrangements in place they will not reach their full potential.

This points to another important parallel with Uber: not everyone will agree on the best regulatory and economic treatment for these new tools. While I don’t think we’ll see the electric equivalent of taxi strikes and rideshare bans, the current veneer of cross-industry agreement could easily crack.

For example, DERs stand to undermine the value of ‘peaker’ plants that currently run a few times a year to support grid peaks. While this might be more economically efficient for the system as a whole, the owners and proponents of those generators are already lobbying against the reliability of DERs and in favor of capacity payments to keep these plants open. Expect to see more lobbying from incumbents around the appropriate use of DERs.

As DERs reach scale, we may also begin to see complaints about their disproportionate use of grid equipment and capacity (rather than the benefits DERs can provide). For example, grid investments may be temporarily concentrated in wealthy neighborhoods that adopt EVs faster, creating a perception that poorer customers are subsidizing the rich. While not always well-founded, these types of arguments have held sway in the past.

In parallel, scale could bring bruising fights between DER providers that echo those fought by Uber against companies like Lyft and DiDi.

The general DER offer structure—where a consumer sees a simple lower rate (or appealing payment), and all the complexity of DER operation is handled on the back-end (within parameters that a customer agrees to)—creates a potential opportunity for commodified price competition. While consumers might benefit from lower prices, as they did with Uber, they might also be unpleasantly surprised by unexpected surge prices8, dead batteries allowed by fine print, or even bankruptcies when providers cut too deep on price.

The regulatory and economic context will need to be rebuilt as these devices get deployed and harnessed, to say nothing of the technical back-end9 required to reliably orchestrate fleets of devices

On net, the results of ‘Uber-ization’ will be massive.

The thoughtful deployment and engagement of DERs will allow us to quell grid peaks, the traditional bane of the electric grid. This will reduce the incremental investments needed to manage load growth, while lowering the effective price of energy across the country. New markets will be unlocked, and upstarts will face down century-old utilities.

The combination of smartphones and apps upended the taxi market; the rise of DERs will fundamentally rewire the provision of electricity.

Leave a comment

Big thanks to Rob, Quade, and Steve for reading drafts of this. Any errors or omissions remain my own.

Further reading:

• Department of Energy - Pathways to Commercial Liftoff - VPPs

• NARUC - Aggregated DERs in 2024

• Ryan McEntush (a16z) - Decentralizing the Electric Grid

• Lynne Kiesling - Decentralizing the grid

How would you avoid ever running out?

Well, you would build extra capacity. People mostly want frozen desserts in summer, so you could overbuild your factory to prepare for the busy season. You could buy extra delivery trucks, custom-made for efficient trips to neighborhoods and businesses. You could try to build some freezers to stockpile, but you worry the craving for sweet treats is too great for storage to do much good.

Worse: all of this is starting to get pricey. People will complain if their dessert is too expensive, something about you ‘ruining their childhood’.

The cherry on top: there are new people moving to town. You’re not sure how much ice cream they’ll want, or even where they’re going to live and how many more trucks you’ll need. It’s nearly impossible to make everyone happy.

—

As crazy as it sounds, this is the position utilities and grid operators find themselves in.

The whole electric system is built to serve peak demand. Billions of dollars are deployed to keep the lights on and the A/C humming during the hottest minutes and hours of the year.

This is harder because electricity is more complicated than ice cream^{[citation needed]}. Electricity is expensive to store, and generation needs to be finely balanced1 with fluctuating loads (uses of power, from phone chargers to foundries). If this balance gets too far out of whack, even for a short time, the whole system can collapse2.

This can happen in an uncontrolled way, but more commonly a grid operator will take parts of the system offline in a rolling black out3 to save the rest. This happened repeatedly in California in the early 2000’s. People were so angry about this failure to balance generation and load that they recalled the governor and elected Arnold Schwarzenegger.

The upshot is that the system is actually pretty under-utilized by design, with spare capacity more than 99.99% of the time. Over the last 5 years, California’s average load has been just ~50% of its peak at 4:57pm on September 6, 2022

This fat right tail (in green) is what everyone is planning for, and why there is so much ‘extra’ capacity.

Imagine that you, an ice-cream-magnate-turned-grid-operator, have been bopping along for years (more than a decade actually) without getting close to 50 gigawatts (GW) in coincident load. Then, a heat wave in early September pushes you above 51 GW in load and threatens rolling black outs. Suddenly you’re sending out an emergency alert asking people to pretty please turn off their A/C.

To complicate matters further, load is now growing for the first time in 20 years, putting more strain on the system during peak hours. Datacenters (which mostly run 24/7) and EVs (which might charge during peak times) are some of the biggest worries for grid planners.

These worries are not idle: a datacenter plopped down in a suburb might request the same 200 megawatt (MW) interconnection as a steel mill’s electric arc furnace; a charger for a semi-truck can draw the same 1MW as a small factory, but ramp up instantaneously4.

Couple this with aging equipment, and maybe it is not super surprising that electricity rates are rising?

–

But wait, there’s more!

The balancing of generation and load only really covers the ‘factory’ portion of our example. What about the ice cream trucks?

For electricity, these are the transmission and distribution lines that carry power to homes and businesses, alongside the transformers, switches, fuses, and circuit breakers that link them all together.

This equipment is also capacity limited, but by heat rather than real-time balancing.

Overloading an ice cream truck will make it break down; pushing too many electrons through a wire will make it melt.

To protect against this, individual pieces of equipment are limited in how many electrons—how much current—they are rated5 to carry. To stay safely within this rating during peaks, the distribution and transmission system might be planned6 against a ‘1 in 10 heat storm’ (to reflect high loading and high ambient temperatures), or an N-2 emergency situation (where 2 key nodes are out).

Extreme heat from climate change is expected to further limit thermal capacity during peak moments, requiring additional buffers.

The result is further under-utilized capacity in transmission and distribution equipment, while expensive upgrades are undertaken to manage through temporary spikes.

—

This all sounds pretty bad. We have a system focused on managing through its worst moments, at the cost of unused capacity the rest of the time, seeing increasing demand for the first time in a generation.

But this is also great news; there is lots of capacity in the existing system!

There is disproportionate value in things that bring down the peak (or at least avoid making it worse), allowing us to better use what we already have. And, increasingly, we have the tools and technologies to do this.

More and more homes have smart devices–thermostats, water heaters, stoves–that can shift their load to help avoid daily and seasonal peaks. Home batteries and EVs can increasingly feed power back into the system at key times. At commercial and industrial scales, microgrids with onsite generation and batteries can shift huge amounts of load temporarily.

These distributed energy resources (DERs) are the equivalent of installing ice cream fridges in homes and businesses: by moving capacity and storage to the edges of the system, the whole thing can be used more efficiently.

To be sure, these are no panacea. In particular, planning around these new technologies, many of which are just now being deployed at scale, is particularly uncertain. Without knowing how they will perform in practice, it is risky to rely on these tools to defer more traditional investments.

Further, because of the complexity involved, new software stacks are needed to orchestrate things at scale. Nor are technology solutions sufficient, regulatory and market structure evolution will also be necessary.

But cost pressures and customer expectations will push utilities and grid operators to better use existing capacity, if only to appease customers that want to charge their new EV or train the next ChatGPT. Done right, we could even be able to bring rates per kWh down (or at least help it rise less quickly).

We should embrace the peak and the buildout it has forced. We can use the spare capacity to enable our electrified future.

In my next post, I’ll talk a bit more about what this means from a business model perspective, particularly for all the new companies trying to carve out pieces of this opportunity.

Thermal capacity (a bit more detail for those who care)

Electrical power (generally measured in watts) is a function of voltage (measured in volts) and current (measured in amperes, or amps).

Power = Voltage x Current

Different parts of the grid operate at different, but specific, voltages. That might mean 500,000 volts for a long-distance transmission line, 12,000 volts for a local distribution circuit through a neighborhood, or the 120 volts at your wall outlet (if you’re in North America).

Because these target voltages are fixed by the system design, to get more power the current must increase. This is a challenge, because increasing the current will increase the resistance—the ‘friction’ of electron flow—which means more heat.

Electrical resistance (measured in Ohms) is a function of voltage and current.

Resistance = Current / Voltage

Incidentally, this is why long-distance transmission lines are run at higher voltages: the resistance (and thus the line loss) is reduced for a given level of current.

This heat is the key constraint on the individual components that comprise the grid–wires, switches, and transformers. This makes heat the constraint on moving energy around within the system.

For example, overhead lines sag between the power poles they hang on. Raise the current, and thus the heat, and the line will sag more. Add enough current and the line might sag into a tree (causing a ground fault, starting a fire, etc.). In theory, if the current goes high enough, the wire will fall apart under its own weight.

This looks a little different for each component, but none will fail as immediately as the overall system under a load mismatch.

Unfortunately, it is also incredibly energy intensive in its current form.

Current leaders like Climeworks and Heirloom estimate they need ~2-3 megawatt-hours of energy to remove and compress a single tonne of CO₂ for storage.

To put that in perspective: I could drive a Tesla Model 3 from Los Angeles to Nathan’s Famous Hot Dogs at Coney Island, realize I forgot my favorite hot dog eating shirt at home, drive all the way back to get it, and only upon my return to Coney Island would I have used the same energy3 as removing just a single tonne of carbon dioxide. 8,500 miles of driving versus 1 tonne of carbon removed.

Scale this up to 1 billion tonnes4 and ~2,000 terawatt-hours of energy will be needed each year for DAC carbon removal. This equates to ~10%5 of total current global electricity consumption just to support carbon removal (on top of existing load growth from EV’s, industrial heat, data centers, etc.).

This makes energy use a key constraint on the scaled deployment of Direct Air Capture carbon removal, and a potential bottleneck on broader climate efforts.

In this post, I’ll step through how to think about this energy constraint, and how different startups are trying to tackle it. No matter which approach wins out, a huge amount of low-carbon energy will be needed.

What makes carbon removal energy intensive?

Direct Air Capture is not a one-step process, and it helps to first unpack where exactly all this energy is going.

Generically6, most DAC carbon removal approaches reduce to the following steps:

1. Collection - Air passes over or through a collection medium that traps the carbon dioxide

2. Regeneration - The collection medium is heated7 or otherwise processed to release the trapped carbon dioxide

3. Compression - The released carbon dioxide is compressed (often to ~100+ atmospheres) for use or storage underground

The energy required for regeneration in particular is immense, with some approaches requiring temperatures of up to 900 degrees Celsius. Across these steps, ~20% of energy is used in collection (e.g., fans to move air) and compression, while ~80% is used in regeneration. This energy generally comes in the form of electricity (which can be turned into heat), though some approaches use natural gas or geothermal heat directly.

The result is an energy intensive process. Leading players like Climeworks, Heirloom, Carbon Engineering estimate they use ~2-3 MWh per tonne in running this process, though all are working to reduce this.

$100 per tonne will not be easy

At the same time, many stakeholders believe that a target cost of ~$100 per tonne or lower is necessary to accomplish carbon removal at scale. This puts a hard ceiling on how much our energy can cost for any given energy requirement.

For example, if our carbon removal process requires 2 MWh of energy per tonne, and can cost at most ~$100 per tonne, we must have energy costs below ~$50 per MWh. This is before even considering any other operating or capital costs.

Right now, the average price for industrial electricity in the US is ~$70-80 per MWh, which already puts us above the target cost at ~2-3 MWh per tonne. That grid average also comes with carbon emissions8; the power for DAC needs9 to be from low-carbon sources like solar, wind, geothermal, or nuclear.

Just as importantly, reducing energy needs or finding cheaper energy leaves more headroom to spend on all the other operating and capital equipment costs10–the fans, kilns, and compressors themselves, not to mention the permitting, financing, overhead, and land costs (near geologically suitable underground storage). All of this costs money.

Some have claimed that achieving removal at $100 per tonne is simply not possible given current cost structures and energy requirements.

In the face of this impossibility, there are 100+ startups trying to find a way.

Many ways to try

As you might expect across so many contenders, there are a range of approaches seeking the optimal tradeoff between complexity, capital costs, energy costs, and operating costs. 

To oversimplify, we can center these efforts around energy along two vectors:

1. Use less energy, often at the expense of higher capital costs

2. Use cheaper energy, often from renewables that don’t produce 24/7

Each is taking their own path towards both, with real uncertainty about which will win. No one has yet built a one megatonne DAC plant, let alone a thousand.

So what are some of the main ways people are trying to solve this?

Single loops (Heirloom)

First up is a (relatively) simple single-loop chemical process based around calcium carbonate from limestone (a very cheap material).

Calcium carbonate (CaCO₃) is ground up and heated to release CO₂, becoming calcium oxide (CaO, also known as quicklime). Calcium oxide is highly reactive11, and when combined with water (H₂O) forms calcium hydroxide (Ca(OH)₂) in a process known as slaking. This calcium hydroxide binds CO₂ in the atmosphere, forming calcium carbonate (CaCO₃) and arriving back at the beginning of the process.

Each time the calcium carbonate is heated, the released CO₂ is collected and stored to neutralize its global warming potential.

This process, commonly associated with the startup Heirloom, uses cheap limestone but requires temperatures of 900 degrees Celsius (C) during heating (often referred to as calcination). That said, there are significant potential energy efficiencies possible through heat recycling from calcination and slaking (which is exothermic, releasing heat).

Compared to other pathways, this approach is not as reliant on continuously operated fans, making it well matched to intermittent renewables like solar photovoltaics (PV). As a result, folks like Heirloom are well placed to benefit from falling solar PV costs.

This chemical pathway is also being used by Terraform Industries, somewhat famous for the solar maximalism of their founder Casey Handmer.

Double loops (Carbon Engineering)

One step more complex and we get a double-loop chemical pathway of calcium carbonate and potassium hydroxide (KOH), the latter being a more effective collection medium for atmospheric CO₂ (at the cost of a more complicated process).

Liquid potassium hydroxide (KOH) is reacted with atmospheric CO₂ in large air contactors powered by horizontal fans. This results in a ‘CO₂ rich’ solution of K₂CO₃ and water (H₂O). When mixed with liquid calcium hydroxide (Ca(OH)₂), pellets of solid calcium carbonate (CaCO₃) precipitate out.

While the refreshed liquid KOH heads back to the air contactors, these pellets can then be heated to 900C in a kiln / calciner to release their load of CO₂ for storage, creating calcium oxide (CaO) to be slaked with water (H₂O) back into calcium hydroxide (Ca(OH)₂).

This use of potassium and calcium in a double loop is most commonly associated with Carbon Engineering (CE), a part of Occidental Petroleum.

Interestingly, these designs often include natural gas to generate the heat for calcination. This adds complexity in the form of CO₂ scrubbing and an air separation unit to create O₂ for combustion12, but allows for cheap (in the US) natural gas to provide the bulk of the energy needed for the process.

This approach appears well-suited for a symbiotic relationship with sustainable fuel manufacturing. The availability of pure O₂ from hydrogen electrolysis makes it more cost efficient by eliminating the need for an air separation unit, and provides an automatic customer for CO₂. While this pathway may not make sense for all DAC, it will likely carve out an important niche in places like Texas with cheap natural gas and nearby chemical manufacturing.

Even more complex looping processes are possible, with folks like Holocene leveraging two overlapping loops of amino acids and guanidine to enable their carbon removal. Other notable startups here include: Phlair, Greenlyte. Each of these trades off the complexity of their systems with input costs, CO₂ reactivity, and energy required for material regeneration, among myriad other factors.

Solid sorbents in a box (Climeworks, CarbonCapture)

Solid sorbent approaches–like those favored by Climeworks and competitor CarbonCapture–trade material handling loops for modules built around sorbent filters which bind CO₂.

This requires higher continuous energy use (at least in current form), with continuously operated fans needed to maintain a pressure differential across the air contactor.

Importantly, the solid sorbent heating cycles run on low grade heat (~80-120C) and could potentially be run directly from geothermal plants, leaving expensive 24/7 electricity to the fans. This could significantly reduce their effective cost of energy as geothermal technology advances.

The real upside potential will come from improved collection materials which, if they are achieved, could significantly reduce energy requirements. This includes non-heat alternatives like the moisture-swing adsorbent approach by companies like Avnos.

Electrochemical cells (RepAir, Mission Zero)

The biggest wildcard, and least mature niche, involves electrochemical cells produced by startups like RepAir and Mission Zero. These use charged cells to attract and bind CO₂, then release the CO₂ when the charge is changed. No heat is required to regenerate the contact medium.

They claim energy requirements of ~600-800 kWh per tonne of carbon removed, but it is unclear how scalable these systems are or how expensive they would be to construct upfront (their capital costs). Without additional information, I would put these at the high capital / low energy requirements end of the spectrum. This could win out but, as with all of these technologies, that remains to be seen.

Taken together, a clear pattern emerges. Approaches with higher energy requirements must be able to rely on cheap, intermittent renewable energy. Those with higher capital costs must have very low energy requirements. Those stuck in the middle will struggle to reach scale.

One path forward is nuclear or geothermal power, whose high availability would favor high capital / low energy approaches. Both sources would allow for continuous operation, both are promising, but neither have proven their ability to scale in a cost effective way.

For my money, I’d bet on the cheap and cheerful DAC plant that runs on ever-cheaper, ever-scaling solar. Because carbon dioxide is diffused globally, DAC plants can go where the sun is, maximizing solar’s advantage over other forms of energy.

Even with reductions in energy usage, whichever approach wins out will likely use prodigious amounts of energy to rebalance our atmosphere. At scale, this will be an important way to use clean energy and ingenuity to combat the worst effects of climate change.

Big thanks to Rob, Ben, Grant, and Andrew for reading drafts of this. All errors or omissions are my own.

Further reading:

International Energy Agency (IEA) - Direct Air Capture

Ben James - Energy Fundamentals of Carbon Removal

Austin Vernon - The Future of Carbon Dioxide Direct Air Capture

Related papers:

McQueen N et al (2021) A review of direct air capture (DAC): scaling up commercial technologies and innovating for the future Prog. Energy 3 032001 doi: 10.1088/2516-1083/abf1ce
(overview of Carbon Engineering solvent approach and Climeworks solid sorbent approach)

Keith D, Holmes G, St. Angelo D, Heidel K (2018) A Process for Capturing CO2 from the Atmosphere, Joule, Volume 2, Issue 8, 2018, https://doi.org/10.1016/j.joule.2018.05.006.(Carbon Engineering approach)

Beuttler C, Charles L and Wurzbacher J (2019) The Role of Direct Air Capture in Mitigation of Anthropogenic Greenhouse Gas Emissions. Front. Clim. 1:10. doi: 10.3389/fclim.2019.00010
(Climeworks approach)

McQueen N, Desmond MJ, Socolow RH, Psarras P and Wilcox J (2021) Natural Gas vs. Electricity for Solvent-Based Direct Air Capture. Front. Clim. 2:618644. doi: 10.3389/fclim.2020.618644
(cost estimates for different energy sources for Carbon Engineering approach)

Young J, McQueen N, Charalambous C, Foteinis S, Hawrot O, Ojeda M, Pilorgé H, Andresen J, Psarras P, Renforth P, Garcia S, van der Spek M (2023) The cost of direct air capture and storage can be reduced via strategic deployment but is unlikely to fall below stated cost targets, One Earth, Volume 6, Issue 7, 2023, https://doi.org/10.1016/j.oneear.2023.06.004.
(cost estimates across leading technologies)

Enter: thermal batteries.

In this piece, I’ll talk about what they are, why they’re compelling, and where they’ll likely gain traction first.

Subscribe now

Thermal batteries1, like those produced by Rondo and Antora Energy, use resistive electrical heat (toaster wire) and large thermal masses (bricks2) to supply zero-carbon heat for industrial uses. They can charge up when electricity is plentiful and prices are low, and can discharge over time to supply steady heat3 to industrial facilities. No longer must we rely on big gas boilers for paper manufacturing and cracking hydrocarbons in a refinery, nor gas furnaces for metal smelting and cement calcination.

This matters because right now this industrial process heat is a major contributor of greenhouse gas emissions, accounting for 7-9%4 of the US’s 2022 emissions, and up to ~17% globally. These emissions have been seen as otherwise hard to abate, given the high cost of alternatives.

Direct electrification is challenging, because the 24/7 electricity many industrial facilities need is expensive (and emissions intensive without some form of storage). For example, assuming ~$0.07 per kWh electricity and ~$5.00 per MMBtu for natural gas5, and 100% and 75%6 efficiency respectively, just the energy costs of outputting 1MWhth in heat would be ~$70 with electricity but only ~$23 with gas.

Heat pumps (which compress and expand refrigerant to shift heat rather than create it) are more efficient7, but they can’t reach the high temperatures needed for many industrial processes. Available models can reach up to a max temperature of ~165 Celsius, but ~80-100C is more common. This leaves a gap for the ~70% of process heat needed at temperatures above ~100C.

Source: Energy Innovation Policy & Technology - Thermal Batteries

Thermal batteries help solve both of these problems.

First, they can achieve much higher temperatures than heat pumps. Antora claims an achieved temperature of 1,800C, while Rondo is targeting delivered heat of 1,500C. The upper bound here is generally in the heat transfer equipment8; the bricks themselves are very stable over repeated cycles. These temperatures are a big improvement over what is possible with heat pumps, even if limits remain for applications at very high temperatures (e.g., metal smelting).

Source: Rondo

Second, thermal batteries can be charged when electricity prices are low, reducing their effective cost of electricity. They can charge up during the day or late at night while avoiding peak evening periods9, helping stabilize the grid while reducing their effective cost of energy. In parallel, they steadily discharge to provide heat. One analysis by Energy Innovation found that this optimized charging could lower the effective cost of electricity per MWh by ~55-65%.

This dance works because the roundtrip efficiency of thermal batteries is high at ~95%+ (due to slow heat losses), and the cost per unit of storage media is relatively low at ~$5-20 per kWh (a fraction of the cost of lithium ion cells).

This creates a compelling path to decarbonizing industrial process heat.

So, given all these benefits, why don’t we hear more about mass deployment of these systems?

While I believe that wave is coming, the economics of these projects are just starting to work.

We can think of project viability in terms of:

• Cost of electricity versus natural gas

• Electricity price volatility (intraday, interday)

• Subsidies / taxes

• Site-level constraints

Starting with only the relative cost of energy, we can see how different US states fare as potential targets for thermal batteries. Even assuming a 50% reduction in effective electricity prices through arbitrage, and 75% efficiency for natural gas, only a few states10 currently see thermal batteries breakeven on a pure energy cost basis.

For example, Washington (in the bottom right of the chart) has relatively cheap hydroelectricity and relatively expensive natural gas. This makes thermal batteries potentially viable on a cost of energy basis11 (below the line in the chart). By contrast, California (top right of the chart) has more expensive natural gas and much more expensive electricity, making it a less appealing target for thermal batteries without subsidies.

Once we factor in the ~$18.00 per MWhth cost12 of the thermal battery system itself, and the grid-tie to power it, no states appear economically viable without subsidies.

So thermal batteries aren’t (yet) cost competitive given low prevailing gas prices in the US, but they are MUCH more appealing in Europe, where the cost of gas is higher.

Predictably, this is where we see many of the first announced projects landing, with Rondo developing 3 customer sites across the EU13. The Europeans will likely get most of the hot brick action in the near-term.

That said, this input price comparison only gets us part of the way there. Solar and wind generation, and the intermittent energy they create, will be the real driving force for thermal battery adoption.

Solar and wind are intermittent resources, meaning their outputs fluctuate and can’t on their own be dispatched to follow changing grid loads. On a sunny day, solar output might be high but it will be zero in the evenings when loads may be peaking due to a mix of residential and industrial demand . Similarly, there may be periods in spring when wind and solar output is high but seasonal heating and cooling loads are muted.

These swings have even brought about something previously unthinkable: negative electricity prices. For wind and solar plants, the marginal cost of producing power is effectively zero. At the same time, there are thermal plants that ‘must run’ because they are necessary for grid stability or because they can’t ramp up and down quickly. Add in some transmission congestion and friction in day-ahead markets14, and the last few years have seen recurring low and even negative prices when loads are low and renewable generation is high.

Source: EU Agency for the Cooperation of Energy Regulators

This pattern of intermittent low or even negative electricity prices15 at least some of the time creates prime opportunities for thermal batteries.

As renewables penetration increases, hot bricks can and should be there to soak up the incremental low cost generation. There is even the potential to run these systems ‘off-grid’16 with direct ties to solar PV or wind plants. While this would require overbuilding capacity to deal with output fluctuations, the ongoing cost declines for renewables make this increasingly viable.

Here too, Europe may have the edge. Their high average electricity prices are partially downstream of their high natural gas prices. When looking at just energy from renewables, the EU has much more globally competitive electricity costs. Thermal batteries will take advantage of the resulting price volatility (or off-grid installations) to drive down the cost of decarbonized industrial heat.

The pace of deployment will be further affected by subsidies (or carbon taxes) that incentivize the decarbonization of heat. This includes the 45X subsidy in the US (still being finalized17) which could reduce thermal battery costs by up to ~$45 per kWh of battery capacity. The EU has its own supports18, with funding through entities like the European Investment Bank (EIB) helping finance early projects.

The introduction of carbon pricing, or carbon border adjustments, would provide further incentives to switch away from hydrocarbon-generated heat.

Finally, assuming favorable high-level economics, the viability of any specific project will come down to site-level nuances.

The first projects will likely be retrofits to existing facilities. This will bring in site-level infrastructure constraints around existing heat systems, which will require varying degrees of modifications to accept heat from a thermal battery. A thornier problem could be the available headroom on grid connections19; charging these batteries requires a lot of electricity and if grid upgrades are required that could slow projects down by years.

Over the medium term, I expect we’ll see purpose-built plants designed around thermal batteries with off-grid renewable resources set up just to convert sunlight and wind into industrial heat.

But where does that leave the US?

Blessed as we are with low cost natural gas, thermal batteries are not yet obvious winners on purely economic terms. While subsidies will help, the declining cost of solar PV (and to a lesser extent wind) will likely dictate the pace of deployment here.

This makes the southwest and the central wind belt the hot brick hot spots.

Source: National Renewable Energy Laboratory

Source: US EIA

These are the regions where I’d expect to see thermal batteries get traction first. California and Texas in particular already have high solar penetration, with the latter adding capacity quickly, creating daily price fluctuations that thermal batteries can harness.

So: industrial process heat has long been seen as hard to decarbonize, but hot bricks are going to change that. While the economics don’t currently pencil out everywhere, they work in enough places for the technology to get proven out and start making a difference for decarbonization.

Expect a quiet revolution in industrial heat over the next 10-20 years.

Further reading:

Austin Vernon - The Case for Brick Thermal Storage

Energy Innovation - Industrial Thermal Batteries

International Energy Agency (IEA) - Heat

Thanks to Rob, Shreeda, Ben, Grant, Steve, Riad, and Katie for reading drafts of this. All errors or lapses in judgment are my own.

Installation is proving to be a key bottleneck for deployment. 

First, the pool of skilled electrical and HVAC1 labor is not growing fast enough to match the growing demand. Some forecast that achieving the goals laid out in the Inflation Reduction Act (IRA) will require 2 million more workers focused on just clean energy deployment and building efficiency, while the industry struggles to add just a fraction of the 9 million jobs the IRA could create overall. To be more concrete, California has a stated goal of 6M heat pumps deployed by 2030. With just 1.5M deployed today, the incremental 4.5M heat pumps represent a big ramp for the ~35,000 HVAC workers in the state, causing delays and higher prices.

Second, many households are finding the capacity of their existing service panel (where breakers are located and power is routed to individual circuits in a house) too small to support new heat pumps or EVs. Where a service panel upgrade is needed (e.g., for the ~30-35% of homes with panels 100 amps or smaller2), utility permitting and line-side upgrades can add significant time and cost to electrification projects. In a 2022 study commissioned by PG&E and SDG&E in California, the upgrade process took up to 10 months and cost up to ~$30,000 in instances where significant line-side upgrades were needed.

Accelerating electrification in the face of these constraints will require avoiding and simplifying installation as much as possible, rather than just optimizing existing processes (e.g., through measures like SolarApp+).

Avoid

Avoiding installation entirely is the first line of defence, going a long way towards shifting these technologies to being bought, not sold. A few examples of this include:

Balcony solar - In many parts of Europe, it is legal (code-compliant) to purchase solar modules online and (with included micro-inverter) plug them directly into a home power outlet to help offset energy usage. This practice is booming in Germany in particular, with more than ~500,000 devices installed and registered with the government. At ~$0.50-1.00 per watt3 for some packages, this is very cost-competitive with more traditional installations, albeit limited to 800W of total output.

Unfortunately, this approach is not (currently) allowed by the US National Electrical Code4, which has no specific allowances for backfeeding power into the home through a normal outlet, though US wiring could theoretically support small inputs (<800W) like those seen in Germany.

Battery-backed home appliances: Devices with built-in batteries can charge at slower rates than their peak power usage, allowing existing 120V outlets to be used for high performance applications like induction stoves (which would traditionally require a 240 volt / 30 amp circuit). 

Impulse Labs is the viral recent example, showing off on Twitter how their 10kW cooktops can boil water in less than a minute (and cook lots of other stuff) while relying on a normal 120V outlet. This magic is enabled by a built-in 3kWh battery, allowing for performance far exceeding what would otherwise be possible with the ~1.8kW that a 120V / 15A circuit could provide.

This battery further allows the homeowner to avoid using grid power during peak rate periods, while also providing resilience during blackouts (a common concern when electrifying key home appliances). Similar to balcony solar, reforms to the NEC clarifying outlet-level backfeeding requirements would unlock additional capability to support household loads through backfeeding.

This battery-based approach could be viable for other devices like electric water heaters, heat pumps, and dryers that cycle on / off or are otherwise used intermittently but at high rates and would traditionally require the installation of a 240V outlet to support their electrification / installation.

Simplify

Where complete avoidance of installation is impossible, there is still significant value in limiting and simplifying that installation effort. In particular, avoiding service panel upgrades and utility permitting / line-side upgrades can significantly reduce the cost of electrification for a household.

Energy management devices: Ultimately, some electrification will require loads too large to avoid installation entirely (e.g., Level 2 EV chargers, larger capacity heat pumps). But, the NEC has a provision (section 750.30) wherein the load doesn’t count towards service panel sizing if it’s behind an appropriate energy management device. These devices come in a full spectrum of complexities, from a simple 240V splitter like that offered by NeoCharge, through Stepwise’s EVtap that uses a current transformer to manage power on an individual home circuit, to full ‘smart panel’ replacements like those from Span or Lunar Energy. 

Each of these seeks to serve new large loads that come with home electrification without the need for expensive service panel upgrades. While most require some level of electrician labor to install, this is significantly reduced relative to the upgrades often required for a larger panel.

Zero-export solar + storage: As adoption of residential solar increases, the backfeed of power onto the grid goes from a valuable service to a potential risk to grid stability5.

In some geographies like Hawaii and Australia, utilities have already moved to ‘zero-export’ regimes, where new residential systems can serve the load at a given home but not feed power back to the grid6. This approach should be adopted more broadly as an option to simplify permitting and installation of residential systems.

For many customers in places like California, reforms to net metering (NEM 3.0) have slashed payments for feeding power back to the grid. In many instances customers could be better served by a simpler permitting process and a zero-export / self-consumption approach to their system. But in practice this type of application can fall under a different–more expensive and more complex–application process designed for much larger systems (e.g., C&I microgrids like those developed by Scale Microgrids). Formalizing zero-export schemes for residential solar + storage systems could accelerate their adoption by trading shrinking export subsidies for faster deployment.

Rapidly deploying technologies like residential solar, EVs, heat pumps and other ‘distributed energy resources’ is a key part of achieving the clean energy transition. These technologies are also increasingly appealing on their own terms, beating traditional technologies like internal combustion engines and thermal power plants on performance and cost.

At the same time, many of the constraints on deploying these DERs exist for a reason. Electricians require training and apprenticeship to safely and effectively do their work. Upgrading the grid requires similarly skilled labor while also often requiring power outages. Permitting, however clunky, exists to manage the disruption that can come from construction and new development.

Developing products and systems that understand and work within these constraints, and respect the tradeoffs they represent, will be just as critical as efforts to alleviate the constraints themselves. Over the coming years, I expect to see more products and offerings that fit this pattern: avoiding and simplifying installation to avoid expensive labor, permitting, and delays, accelerating the energy transition in the process.

Not only are financing costs large, their recent fluctuations go a long way towards explaining current dynamics in the residential solar market. The slowdown in installations, the shift towards subscriptions and leases, and the bankruptcy of Sunpower are downstream of interest rates as much as seemingly more proximate causes like California’s shift to it’s Net Energy Metering (NEM) 3.0 approach (which reduced payments for energy exports back to the grid).

Financing costs are large

We can see this most clearly with a worked example. Say an independent installer quotes you $3.00 per watt for an 8 kilowatt (kW) rooftop solar system, bringing the overall system costs to $24,000.

If you got a subsidized 0% interest loan1 for the 20-year life of the system, your monthly payments would be ~$100.

At 4% interest, available just a couple years ago for systems secured against the equity in the home, the payments rise to ~$147 per month for a nominal cost across the system life of ~$35,319.

At 8% interest, roughly in line with the last ~12-18 months, payments reach ~$204 per month for a nominal cost of ~$48,889–more than double the overnight capital cost of the system.

Put differently, a system financed at 8% interest is nearly 40% more expensive in nominal terms than one financed with a 4% loan, and more than twice as expensive as one with a subsidized 0% loan.

This matters when nominal 10-year Treasury yields have jumped from ~1.5% at the start of 2022 to a peak at ~4.9% in Q4 2023 and bank prime rates have jumped from ~3.25% to ~8.5% over the same period.

Even bluechip residential solar companies like Sunrun have seen the yields on their securitizations rise from 2.28% in September 2021 to 6.78% in September 2023, for an increase of ~4.5% over the course of 2 years.

These are stiff headwinds which directly impact the cost of residential solar energy for the vast majority of customers who lease or finance their systems.

Solar PV is differentially exposed to rates

Importantly, these financing costs fall harder on solar PV systems.

There are two relevant comparison points here: other generation technologies, and prevailing electricity tariffs from utilities / munis / coops.

At the utility scale, the marginal non-solar watt is coming from wind or combined cycle natural gas turbines (CCGT). In this comparison, solar PV assets have structurally longer duration due to their very low O&M costs. With their moving parts, wind turbines require more maintenance, while CCGTs require both maintenance and ongoing fueling, meaning their costs are relatively more evenly spread across the life of the asset. As a result, the overall cost of solar PV projects is relatively more sensitive to changes in interest rates than alternatives2.

Gas turbines are not realistic options for most homeowners though. For them the relevant reference point is the cost of electricity from their local utility. In this, solar PV has a similar disadvantage, with a larger share of utility tariffs devoted to ongoing operating and capital costs rather than financing costs, buffering the impact of interest rate changes.

As a result, though retail electricity rates have seen steep increases in recent years, those increases have not been on par with the ~30-40%+ increase implied for solar PV by the rise in financing costs.

Subscribe now

Cost increases kill marginal projects

Importantly, the rise in financing costs erodes the value proposition for customers.

We can think about the value of a residential solar system as the difference between the cost of the system and the value of the electricity produced—either from self-consumption / avoided electricity tariffs or from export payments.

Let’s assume an energy value of ~$0.15 per kWh for our model 8kW system, derived from a combination of avoided utility payments and energy exports. If our 8 kW system achieves a capacity factor of ~20% over 20 years of operation, we get the following ‘levelized cost of energy’ (LCOE) under different financing costs:

A project that made sense at 4% interest rates is distinctly unprofitable at 8%.

The universe of potential solar PV projects can be defined more generally in terms of these same variables: a site’s solar resource (captured here by the capacity factor), a project’s capital costs, its financing costs, and the value of energy produced.

Over the last decade, the decline in capital costs and the rise in utility rates had steadily grown the number of residential solar projects that have positive expected value. The recent sharp rise in financing costs, coupled with the shift towards lower export rates with NEM 3.0 in California, has temporarily reversed this expansion as marginal projects become unviable.

We can see the impacts of both of these effects in the chart below (drawn from this very interest report on the CA residential solar market over the last year):

We see declining installs in Q1/Q2 of 2023 as rates start to rise, a sharp uptick in installs before NEM 3.0 comes into force, then a decline as activity shifts under the new regulatory regime (and continues to be hampered by high financing costs).

Importantly, and seemingly overlooked in the report itself: installs started to decline before NEM 3.0 came into effect, showing how important interest rates were as a driver.

This rate-driven slowdown helped kill Sunpower

At scale, a company like Sunpower or Sunrun or Sunnova can be thought of as a complex machine that identifies positive value projects, convinces customers the projects are worthwhile, installs the necessary equipment, and services both the equipment and the customer for the next 20+ years, all underpinned by a carefully orchestrated financing pipeline.

When market conditions change sharply, this machine can break down.

As the residential solar value prop eroded due to financing cost increases and energy export value reductions from NEM 3.0, it became increasingly important to be able to reshape the customer offering to preserve upside.

This favored companies like Sunrun that had focused heavily on subscriptions rather than outright system sales, and had the capability to sharply ramp up storage attach rate (given the increased value of storage capacity under NEM 3.03). Subscriptions allowed Sunrun to optimize its financing structure on the backend while showing a simple, no-upfront-cost offering to customers. Storage, when integrated with and orchestrated as part of a residential solar system, created significant value for the system as a whole.

These two capabilities could be combined to maintain a compelling customer value proposition and keep the machine running. This advantage has been borne out, with Sunrun reporting sharp increases over the last 18 months in both subscription rates (~78% to ~95%) and storage attachment rates (~18% to ~54%).

Because it relied more heavily on 3rd party financing channels and exposed more of the value complexity to customers, it was harder for Sunpower to reshape its customer value proposition to stay compelling. This left its sizeable customer acquisition and system installation machine running at reduced effectiveness and bleeding cash. When the company ran into issues with its accounting and inventory controls, it lacked the financial flexibility to maneuver and was forced into bankruptcy.

Financing costs are hard to change directly

Unfortunately, despite the importance of financing costs, there are few direct opportunities to drive future savings.

The securitization of residential solar projects is a mature process, as are the tax equity maneuvers that often go along with them. 3rd party financing channels exist, both solar-specific and general, already exist. While future rate cuts will provide a welcome tailwind to the industry, the industry is not in a position to influence this directly.

This leaves developers to focus on reducing upfront costs to drive further value for customers, without neglecting the ongoing importance of financing costs to the overall picture.

But learning curves2 aren’t magic, and ‘solar’ isn’t some monolithic thing whose costs have declined monotonically. In this post I want to dig into what specific costs have been reduced, and where we might hope to see further declines.

I’ll focus here on the US3, drawing on data4 from the National Renewable Energy Laboratory (NREL) unless otherwise noted. We’ll look at utility-scale and residential PV in turn, with each demonstrating their own pattern of cost reductions and further opportunities.

Share

Utility-scale PV

First, let’s establish the classic cost decline chart for utility-scale PV in the US. As noted, the costs have come down enormously (-82% per watt DC between 2010 and 2023, or -12% per year).

In this case though, because the cost declines have been so precipitous, it can be hard to tell what’s still left.

To get a better view, we can look at how the split of costs has changed over time. As we can see below, everything has gotten cheaper but some costs have fallen even faster than others (per the compound annual growth rates in red).

In particular, module costs have fallen fastest (even in the face of fluctuating tariffs), while (field) labor and 'balance of system’ (BoS) costs (like rails / mounting and conductors / wiring) have proven harder to cut5 or even gotten more expensive in recent years. This has left the Labor and BoS as a larger share of the remaining costs.

This underpins the thesis for a number of interesting startups focused here, including:

• Terabase: Construction pre-fabrication and automation (along with related offerings) that purports to significantly increase labor productivity, including a panel wielding robot arm.

• Erthos: ‘Earth-mounted’ solar that trades individual panel yield for BoS and land efficiency. The thinking professionals approach to “pave the world with PV”.

Alternatively, companies like Origami Solar—which manufactures steel racking in the US—have taken the relatively high prices of BoS components and attempted to differentiate on supply chain stability and domestic content (which can help achieve IRA tax incentives) rather than cost per se.

More generally, it will be interesting to see if the industry carves out a role for distinct ‘integrators’6 that pre-assemble modules + wiring + some BoS offsite to speed labor in the field and potentially reduce BoS costs, or move in the direction of Terabase where major projects include on-site integration factories.

Surprising (to me at least) is the relatively low non-labor soft costs. As we’ll see in a second, this is a big part of the story for residential solar, but it seems utility-scale interconnection and permitting delays are painful but not a big cash cost relative to the project as a whole.

Overall, this is a story of cost declines across the board, and further progress will likely involve incremental improvements across cost types.

Share

Residential PV

We can perform the same analysis for residential solar, where costs have declined by -69% per watt DC between 2010 and 2023, or -9% per year.

Starting from a higher cost per watt, the declines have been impressive but not quite as steep as for utility-scale PV. The costs declines have also been more uneven, with the biggest improvements in Module and Labor costs.

The result is capital costs that are more than twice as expensive per watt as utility-scale solar.

Module and Labor costs are similar per watt between Utility-scale and Residential PV.

Inverter costs are meaningfully higher for Residential installations, likely a result of the competitive dynamics (and physical realities) of creating a plug and play system centered on someone’s garage (rather than on a concrete pad as part of a spec’d and procured ~$300M installation in the desert). This may grow more commoditized over time as industry interoperability standards become more widespread.

The story is similar for BoS costs. Theoretically, Residential BoS could be cheaper (there’s already a structure to attach the modules to), but the customized nature of these installations coupled with the relatively low bargaining power of individual installers leads to Residential PV being a price taker for components. The result recently has been a jump in BoS costs from ~$0.36 per watt in 2021 to ~$0.57 in 2023, likely due in part to supply chain disruptions.

As supply chains normalize and components continue to standardize, I would expect to see further improvement in each of these areas.

But what’s in that big brown Other bucket7?

For residential solar, the single biggest cost is customer acquisition.

In the United States today, residential solar is sold, not bought.

These are complex purchases worth tens of thousands of dollars, often involving a multi-decade contract. The result is sophisticated (and/or expensive) customer outreach (including both direct mail and in-person door knocking) and pipeline development (where leads are qualified, systems are designed, and customers are walked through their options before hopefully signing on the dotted line).

These customer acquisition costs have proven relatively resistant to improvement, and have ended up as the single biggest cost per watt.

Folks are tackling this at multiple stages in the process, including:

• Sales support tooling (e.g., OpenSolar): The sales process is complex. Sales and design tools that flow directly into creating a customized proposal (including potential savings) help reduce sales effort required per customer. (And the tools really are easy to use, as I found out writing my article on NEM 3.0).

• Solar marketplaces (e.g., Haven Energy): The marketplace is fragmented, with many distinct steps and capabilities between initial customer contact and system completion. Connecting customers and installers while smoothing the path from inquiry to system operation (often leveraging similar tooling and support as above) can reduce cost per customer by increasing yield.

• Subscription (e.g., Sunrun): High upfront costs are a big deterrent to potential customers, and ‘subscriptions’ or solar leases have proven an effective way to get customers past this potential hurdle. Per Sunrun’s latest quarterly filings, more than 80% of their customers opt for a subscription versus purchasing a system outright.

The final tailwind will come from rising utility rates. As electricity rates go up, the relative value of a solar (+ storage) installation goes up as well, likely easing the path to customer acquisition8.

At the same time, as module costs continue to decline we may start to see different paradigms for solar at home. For example, Germany is seeing growing adoption of ‘balcony solar’ wherein a small number of modules get plugged directly into the home through a normal outlet. While the absolute power output may be limited, this approach eliminates whole categories of costs relative to traditional rooftop solar.

As with utility-scale installations, the truth will likely be a combination of all of the above and more, but there remains significant opportunity to further reduce costs and increase deployment of these resources at all scales.

2030? 2035? 2045?

What if I told those batteries are already deployed, they just can’t be used? That’s the situation the Golden State finds itself in with its growing population of electric vehicles.

There are more than a million EVs in California, together comprising more than ~60 gigawatt hours of storage. Those batteries are effectively siloed today, unable to be used for anything but driving.

But once EVs start getting built with bi-directional charging capabilities, huge amounts of additional storage will be created every year, unlocking radical benefits for individual customers and for the system as a whole.

Share

Where we are today

There are ~30 million light-duty passenger vehicles on California’s roads today (per the California Energy Commission). At the end of 2023, ~1.5 million (~5%) of those were ‘zero-emission vehicles’ (ZEVs).

There are two types of ZEVs1, battery-electric vehicles (BEVs) like the Tesla Model 3 and plug-in hybrid electric vehicles (PHEVs) like the Toyota Prius Prime. BEVs have no combustion engine and run entirely on battery power, PHEVs have a smaller battery pack with a combustion engine to generate additional electricity for longer trips.

There were ~1.1M BEVs in California at the end of last year, and ~0.4M PHEVs.

If we assume a ~55 kilowatt-hour (kWh) battery pack for the BEVs (roughly in line with a Tesla Model 3) and a ~10kWh battery pack for the PHEVs (roughly in line with the Prius Prime), that equates to ~65 GWh of battery capacity in California EVs.

This is a huge amount of capacity.

As of May 2023, the stationary battery fleet within the territory of the California independent system operator (CAISO) was ~17.7 GWh, just ~27% of what these vehicles represent.

The highest instantaneous load ever experienced by CAISO peaked at ~52.061 GW, on September 6, 2022 at ~4:57pm. The load for that hour peaked at ~51.5 GWh, or ~80% of the theoretical capacity of these batteries2.

Where we are headed

This year, CAISO is expected to gain ~6.8 GW / 27 GWh3 of stationary batteries.

By comparison, ZEVs comprised ~24% of new car sales for Q1 2024 at just over ~100k vehicles. Annualizing this implies an additional ~19 GWh of vehicle-based battery capacity, bringing the total to ~84 GWh.

CAISO is aiming to hit at least ~23 GW / ~121 GWh of battery storage by 2035 (though given the current rapid pace of additions this could easily be surpassed).

If ZEVs reach ~35% penetration by that point (when they will be required to be 100% of new sales in CA), they would represent ~450 GWh of storage capacity.

At full ZEV penetration, the ~30 million light-duty vehicles in California would represent ~1,300 GWh of energy storage capacity4, enough to power 2023’s average load of ~25 GW for ~50+ hours.

This is a huge amount of potential storage, an amount that doesn’t seem to be fully recognized by all the relevant grid participants.

Share

They’re still cars, not BESS with wheels

At this point, you may be rightfully thinking that these are cars, and need to be used as such. Owners may not want to run down their batteries to zero to serve the grid. Totally fair!

The average driver travels ~40 miles per day. Assuming a ~240 mile range for our model BEV, this would consume ~9 kWh per day5. If we want to always maintain ~2 days worth of driving capacity, we would only be able to use ~37 kWh (~67%) of our ~55 kWh battery as storage capacity.

If we further assume the PHEV capacity is wholly dedicated to driving, current ZEV penetration implies ~40 GWh of available storage capacity in light-duty vehicles (versus ~64.5 GWh total capacity). At full potential, this would drop our available capacity from 1,300 GWh to 800 GWh (‘only’ ~32 hours of capacity at 2023 average load).

The bigger challenge is what it means to charge all these zero-emission vehicles that are driving ~40 miles per day.

This is not much of an issue just yet6. Our ~1.5M ZEVs using ~9kWh per day only add ~13.5 GWh of load per day, or less than 3% of the average daily load in 2023.

At 35% penetration in 2035 (in-line with the math above), this would imply a necessarily larger load impact of ~95 GWh per day (a ~16% increase over the 2023 average). At full ZEV penetration, the average daily load would be fully 50% higher (from ~600 GWh per day to ~900 GWh per day). All this in a industry that has seen effectively no load growth over the last 20 years.

Even worse is the uncertainty around when these cars will get charged. If everyone gets home from work at 6pm and plugs in their EV to recharge without constraints, the grid is going to swiftly and unceremoniously break.

This is a potential nightmare for utilities, and is the focus of ‘managed charging’ startups like Weavegrid, ev.energy, and EnergyHub. They are working with utilities, OEMs, and drivers to set up the software and process infrastructure to levelize EV charging to avoid creating load spikes (and reduce charging load during peaks).

This is also the target of more standard measures like Time-of-Use (TOU) rates where customers pay more per kWh consumed during evening peak hours than (for example) during the middle of the day when solar output is peaking.

So, these cars represent a huge risk to the grid if their charging all lands at the same time, and it is this risk that most people are currently focused on. This will get solved in the medium-term, it is the opportunity to leverage this storage capacity to proactively smooth load throughout the day that is the most exciting.

This is how we shift cheap solar from the middle of the day to all hours of the night. And we get a car thrown in for our troubles.

What’s holding us back?

The risk from EV charging are well known in the industry (at least in places like California); folks are actively working to manage this. The storage potential is huge but there are some key gaps that will need to be overcome for the potential to be realized:

Hardware layer: It is one thing to control an EV to moderate its own charging, its another to be able to feed energy from that EV back into a home (V2H) or back to the grid (V2G).

Very few EVs currently have the hardware to be able to do this, with the Ford F-150 Lightning and Tesla Cybertruck being notable examples. This is the biggest barrier today, and why we don’t see positive grid impacts from so many GWh of storage already deployed.

Many OEMs and utilities are performing pilots and testing these capabilities, and these should start to bear fruit soon. Given their leading position, a shift from Tesla7 (which they’ve started with the Cybertruck) could be a big momentum boost here.

On the charger side, there are further incremental requirements, with a bi-directional charger needed to effectively act as an inverter to feed AC power to the home service panel. This may also require more complicated electrical work at the home, depending on the current state of their panel and the local utility / regulatory environment8.

Making these changes will add incremental cost to these systems, but these costs are not game-changers relative to the cost of the battery.

Software layer: It is not obvious (at least to me) who should ‘own’ the behind-the-meter energy management system (EMS) that coordinates the charging / discharging of a vehicle battery alongside all the other loads in a home. It’s not enough just to have the necessary hardware in place, particularly given the importance of charging / discharging behavior in maximizing battery usage and value.

As mentioned, there are ‘managed charging’ startups that are building systems to orchestrate vehicle charging on behalf of utilities. This could seemingly be extended to handle the discharge behaviors from those same vehicles.

There are also residential solar companies like Sunrun that are now working to position themselves as ‘VPP companies’ to reflect their control over customer-side loads and storage through the EMS that comes with the systems they install (and may have contractual rights to control).

Finally there are companies like Span and Lunar Energy that are building software-enabled smart service panels that link with solar panels, storage, and vehicle charges along with allowing the orchestration of loads on a circuit-by-circuit basis within the home.

In the medium-term, I expect multiple providers to 'solve’ the software layers once the hardware capabilities exist for V2H / V2G power flows from BEVs.

Market structures / incentives: The cost of cycling EV batteries to support home or grid loads is not zero. Beyond the round-trip energy losses, there is battery degradation and loss of battery flexibility that vehicle owners would want to be compensated for.

That said, the jump in retail time-of-use rates between peak and off-peak rates (e.g, $0.25 to $0.61 per kWh9) already make this cycling worthwhile to support behind-the-meter loads (i.e. at your house), as I outlined here. Summer peak rates in California are very high, and EV-based energy storage can reliably help individual consumers manage their own exposure to these high prices.

Importantly, exports to the grid are not required to take advantage of this. Grid export has been a key issue (and focus area) for various bi-directional charging programs, but is overestimated in terms of its long-term importance relatively to offsetting behind-the-meter load.

In areas without TOU rates, the incentive structures will be much more muted and specific demand-response (DR) programs will likely be needed to capture the value of this storage capacity. Historically, retail capacity from these types of programs has been small (and/or expensive) relative to commercial & industrial DR programs.

Inertia / ‘customer experience’: Most customers don’t actually want to think about electricity this much. They would like their bill to be low and their appliances (and car) to work when they need them.

Most don’t want to participate in DR programs where someone else controls their battery, and EV OEMs are mindful of the customer implications if a car doesn’t work when and if it’s needed.

The key will be in making this as simple as possible for consumers to ‘set and forget’. TOU rates and an energy management system (EMS) should allow for this, with the customer setting up their preferences and guardrails at key life events (e.g., new car, new solar panels) and allowing the EMS to optimize energy usage in the background.

If the incentive is there, enough people will participate to make a big dent in peak loads behind-the-meter even without 100% adoption.

What could this unlock?

At scale, this has the potential to reshape how the grid should operate in places like California.

Solar panels—at residential, community, and utility scales—will charge EVs and other storage during the day to be discharged as needed during the evening hours and into the morning before the sun comes back up. Increasing EV penetration will supercharge the storage capacity available to serve this purpose, allowing for much higher solar (and wind) penetration across the system.

Dependent on the relative mix and location of stationary and EV-based storage, this could even avoid major grid upgrades through better utilization of existing system capacity throughout the day. In parallel, overall energy use will continue to grow, allowing for increased spreading of fixed system costs, lowering volumetric rates per kWh.

EV batteries will be a hugely important resource to decarbonizing the energy system, their potential just needs to be realized.

Thank you to Jason Crawford and Heike Larson for the conversation that led to this post. Any errors / omissions are entirely my own.

Unfortunately, GETs are not ‘one weird trick to double transmission capacity’.

In this piece I’ll step through a few of the main GETs that folks talk about, and outline the circumstances under which each is useful. This should give a better sense of where efforts can be focused to maximize positive impacts for the transmission system.

The very short version:

• Dynamic line rating: Most useful in regions that see winter peaking and have good winter wind resources; less useful in places like the US southwest.

• Advanced conductors: Faster and cheaper than adding a new line, particularly where there is unused capacity in the relevant substations; focus on lines <50 miles long.

• Advanced power flow controllers: Really useful in situations where network capacity is limited by a specific bottlenecked line with viable alternative paths; not so useful otherwise.

The benefits of GETs are very real, but their use cases are much narrower and limited than the headlines would suggest. These should not become a distracting panacea; we will still need to do the work of building out transmission and energy storage capacity alongside renewable generation to enable the shift towards a zero-carbon power system.

Why GETs?

Construction of new transmission lines has been sluggish (at best) in recent years, falling to a marked low of just ~138 miles in 20211. At the same time, transmission capacity is a key bottleneck on interconnecting additional renewable generation, and new loads like datacenter and electrified heavy industry.

Based on recent analysis by the Lawrence Berkeley National Lab, there are ~1.6 terawatts (TW) of generation and ~1 TW of storage capacity currently in interconnection queues across the country, more than the nameplate capacity of our entire current electric grid. ~95% of this generation is zero carbon, representing an expansion of ~5-10x of our installed renewable generation (if all projects were built).

Grid Enhancing Technologies provide one set of options to expand transmission capacity without the cost and time of new-build transmission, helping to interconnect these vital clean energy resources faster and at lower cost.

An analysis by RMI and Quanta, focused on the PJM region2 in particular, show the promise here:

But as we’ll see, their applications are narrower than this might suggest, creating the need for a “yes, and” approach to deploying both GETs and new transmission construction.

Dynamic line rating

The pitch:

Increased transmission capacities of up to ~40% for a fraction of the cost (and time) of building out new transmission lines.

How it works:

Many electric transmission lines are thermally limited in operation. As you run current through a given conductor, resistance will create heat causing the line to sag. Too much current and too much sagging creates a risk of dangerous faults3 (and potential damage to the conductor itself).

As a result, transmission lines are ‘rated’ to carry specific maximum quantities of current4 given their expected operating conditions (ambient temperature, wind speeds, solar irradiance, etc.) and allowable sag levels (e.g., before a line would risk hitting something).

The ‘operating condition’ assumptions driving these ratings have historically been static and set at relatively conservative levels (e.g., high temperature, low wind) to ensure safe operations at all times. This means that during most periods the current ratings are too low and don’t accurately reflect how much energy the line could safely carry.

‘Dynamic line rating’ (DLR) seeks to remedy this by adjusting the ratings to more accurately reflect current conditions, allowing grid operators to safely use the extra thermal headroom on cooler and windier days. This can be enabled through granular weather forecasts, as well as field sensors mounted on towers and on the lines themselves.

FERC Order 881 is already driving change here, requiring grid operators to move to ‘Ambient-Adjusted Line Ratings’ that reflect ambient temperatures beginning in July 2025. Many other startups (e.g., LineVision, Heimdall) and grid operators are working on ways to leverage sensors to integrate wind speed, realized line sag, and other variables to more accurately re-rate lines to maximize safe power throughput.

Where it is most useful:

While some version of line re-rating is required by FERC 881, the biggest impacts from this shift will happen in places that have:

1. Overly conservative static ratings:

   Most weather days will not be near the conservative conditions embedded in the prior static ratings. There will (likely) be meaningful upticks in thermal headroom on most days through the deployment of DLR, though this could vary across regions. Impacts will be more muted until planning can be updated to reflect this incremental capacity (see below).

2. Winter peaking

   The capacity of the grid is generally only limited on ‘peak’ load days. In hot regions like California, these all occur in the summer. In colder regions like the US northwest and Canada5, this can happen in the winter. Because DLR is most likely to increase capacity during cold, windy conditions this will be most important for regions that experience winter peaks due to heating load.

3. Good wind resources in winter months

   Many regions in the US see their highest wind production during cold, windy months, illustrated in the graphic below. DLR is a good fit for these areas, and over the medium-term could allow them to productively add more wind generation without requiring incremental transmission build-out6.

   

Limitations on usefulness:

The long-term impact from DLR will likely be constrained by the coincidence of peak days with cold conditions, limited substation capacity, and by the added complexity of modelling dynamic line ratings as part of the capacity planning process.

1. Coincidence of needs (or not): While it is great for places like Minnesota to be able to squeeze more wind power through a line on a winter peak day, DLR is unlikely to create meaningful extra capacity in summer-peaking regions. In places like California, Arizona, and Texas, transmission capacity will be limited by DLR on exactly the days and times when it will be needed the most.

2. Non-thermal constraints: Capacity of the system is only as great as the weakest link. In situations where that is the thermal capacity of the conductor itself, DLR will help. In practice, substation equipment7 or power stability8 issues may become the relevant bottleneck, limiting the impact of DLR. This will limit the impact of DLR to lines that are <50ish miles with incremental substation capacity.

3. Planning complexity: While DLR may unlock incremental capacity in day-ahead transmission scheduling, it is another thing entirely to plan for that capacity 10-20 years out. Significant effort will likely be needed for RTOs and ISOs to figure out how to integrate dynamic line capacity into their study process. While FERC 881 might push this to get solved, in the near-term some may choose to continue to plan against the more conservative operating conditions.

Advanced conductors

The pitch:

Increase the capacity of transmission lines by ~50-100% at a fraction of the cost of building new transmission lines

How it works:

Advanced conductors aim to tackle line sag from a different angle, by improving the conductor itself.

‘Standard’ conductors are Aluminum Conductor Steel-Reinforced (ACSR) and have been in use for a century. These use hardened aluminum wires around a core bundle of galvanized steel, with the aluminum acting as the primary conductor. They are well understood by utilities and their crews, and generally lowest cost.

‘Advanced’ conductors aim to improve on this proven design through higher tensile strengths, higher max operating temperatures, and lower resistance (at comparable ACSR operating temperatures). They generally do this through a modified core, with many employing composite cores. The Idaho National Laboratory has a great deep-dive on the specific technology variants here (from December 2023).

The upshot is that these advanced conductors can achieve higher ampacity without violating thermal sag constraints, albeit at a higher cost per mile than ACSR. Operated at the same temperatures as ACSR conductors, many can also achieve higher efficiencies / lower lines losses, reducing energy requirements and cost for customers.

Among other reforms to the generator interconnection process, FERC Order 2023 directed that the interconnection process should include evaluations of alternative transmission technologies like advanced conductors. This means that advanced conductors will be part of the repertoire of solutions that can be used to address constraints on interconnecting new generation. While not directly affecting load interconnections, this move should still boost widespread adoption of advanced conductors.

Where it is most useful:

1. Reconductoring of existing lines: Advanced conductors are most appealing for re-conductoring of existing transmission lines. For a relatively limited investment (and often simplified permitting, given use of existing rights-of-way), capacity can be boosted by ~50-100%. This is particularly appealing in built-up urban environments where there is limited space for new build construction.

2. Long-spans: Advanced conductors are also uniquely useful for line segments that cross long spans (e.g., over rivers or highways). Their minimal sag allows for shorter, cheaper transmission towers than would be required with ACSR.

This expectation is borne out by the types of projects that have been happening with advanced conductors, with many looking like9 this reconductoring by Southern California Edison of a ~230kV line with an ACCC (Aluminum Conductor Composite Core) conductor. In that case, the capacity could be expanded using existing towers and rights-of-way, which was particularly important in urban Los Angeles / Orange County.

Limitations on usefulness:

1. Non-thermal constraints: Similar to DLR, advanced conductors are mostly useful on lines that are thermally constrained via conductor sag (and have incremental substation capacity). This means lines <50ish miles long10. Power stability issues are even more notable as a constraint for advanced conductor usefulness, with many having higher resistance at their max operating temperatures11.

2. Limited track record of reliability: Utilities are deeply conservative organizations and generally want to see extensive evidence of reliability before broad deployment of new technologies. As the conductor manufacturers keep innovating, each new technology has a new burden of proof to overcome. This is not helped by the new installation procedures required of field crews that—performed improperly—can risk damaging the conductor12.

While #2 here will be overcome with time (and with prodding from FERC), non-thermal constraints are more limiting.

Advanced power flow controls

The pitch:

Avoid costly upgrades to deal with N-1 contingencies, install advanced power flow controllers instead!

How it works:

Most transmission systems are planned against N-1 and N-2 contingency scenarios, representing the loss of 1 and 2 nodes respectively. The lack of a given node (e.g., substation) shifts the grid’s topology and alters the flow of current across the network. This shifting of load can cause thermal overloads in lower capacity parts of the grid.

When these potential overloads surface during the transmission planning process, the traditional (expensive) solution is to bolster the offending line or substation (e.g., through reconductoring or new build construction). Advanced power flow controls (APFCs) provide a lower-cost alternative, using advanced power electronics to shift load away and avoid breaching thermal limits.

Even outside contingency scenarios, limited capacity on specific transmission lines can create congestion between low-cost generation and customer loads. Left unaddressed, capacity limits on congested lines means nearer, higher cost (often hydrocarbon) generation must kick in to fill the gap. One report from GridStrategies put these costs in 2021 at ~$7.7B for regions covered by regional transmission organizations (RTOs) and ~$13.3B across the US as a whole.

There are a number of specific technologies that fit under the ‘advanced power flow controls’ bucket, each working in their own way to manage reactance and impedance on a given transmission line. One prominent example, the SmartValve from Smart Wires, does this by injecting voltage out of phase with a line’s current to either increase its impedance13 (and push power onto alternative lines) or decrease its impedance (and pull power onto the target line).

APFCs are notable because they use modern power electronics to rapidly respond to grid dynamics in real-time, allowing for more fine-grained automated and manual control by grid operators. While Smart Wires originally launched with a conductor-mounted solution, most APFCs today are mounted in substations.

Where it is most useful:

1. Heavily meshed transmission networks: APFCs are more useful in heavily meshed systems with multiple alternative paths for power to flow (e.g., PJM in the US northeast), rather than systems with fewer alternative branches (e.g., CAISO).

2. High congestion areas: APFCs are particularly useful in areas with significant congestion, indicating there is value in expanding transmission capacity beyond merely bolstering reliability during contingency scenarios.

Beyond these general principles, there will also be grid-specific scenarios and use cases where shifting the grid topology through APFCs could be the lowest cost solution to a given grid need. APFCs are also addressed in FERC Order 2023, so should be deployed in generation interconnection situations where they make sense.

Limitations on usefulness:

1. Absolute transmission capacity: APFCs can shift power flow across alternate paths to maximize utilization of all available capacity. But they cannot create new capacity that doesn’t exist. While these are a useful tool in the toolbox to get the most out of our existing network, they are not a substitute for absolute increases of transmission capacity to interconnect incremental load and generation.

2. Availability of alternate paths: APFCs need alternate paths to shift power flow onto (or from). These won’t always be available, and the opportunity for APFCs may be saturated at a relatively low level. In a 2018 report, EPRI estimated that 13 targeted APFCs would resolve most APFC-addressable congestion within PJM.

Overall, these technologies well help us squeeze more capacity out of our existing transmission grid, with minimal incremental investment relative to new build-out.

But there are meaningful limitations around applicability, added complexity in planning, and full-potential impacts that should give pause to those who think that GETs will provide an easy step change in capacity. In particular, many GETs address thermal overloads in one way or another, but the long-distance lines used to interconnect renewables (particularly in the southwest) are limited by power stability not heat.

GETs are an important tool in the toolbox, but we still need to get to work build out transmission capacity.

By using ever-cheaper energy from renewables, their low cost approach to synthetic methane production has the potential to upend what we think is possible with zero-carbon energy.

In this article, I’ll aim to cover a few things:

• Summary: What is Terraform proposing to do?

• Market: What bet are they making?

• Customers: Where will early demand come from?

• Value chain: What role will Terraform play?

• Competitors: Who else is going after this?

• Differentiation: What makes Terraform unique?

• Other potential challenges

• Other potential tailwinds

• Next steps

Through this, I will show that Terraform has an ambitious but plausible path to ushering in a new age of energy abundance.

Summary: What is Terraform proposing to do?

They aim to pair solar PV with a ‘Terraformer’ module that ultimately outputs pipeline grade methane. For reference, their approach is described more fully here.

The Terraformer modules have three relevant sub-components:

1. Hydrogen electrolyser: Splits water into its component hydrogen and oxygen

2. Sorbent-based carbon capture + regeneration kiln: Captures atmospheric carbon dioxide, releases it upon sorbent regeneration

3. Sabatier reactor: Combines hydrogen and carbon dioxide to produce a stream of methane and oxygen

These are designed with two goals in mind:

• Run effectively on intermittent, direct-current power from a solar array

• Be as cheap as possible

The assertion is that when paired with low cost solar, these modules will produce carbon-neutral methane at a cost that beats traditional fossil fuels.

If this can be achieved, it will reshape global energy markets and usher in a new age of energy abundance. Even if only partially achieved, this will help humanity avoid some of the most expensive energy transition pathways in hard to abate sectors like aviation and shipping.

Market: What bet is Terraform making?

Let’s start by understanding the ‘Unit Economics’ section of Terraform’s April 20231 whitepaper and their June 2023 summary of the Terraformer Mark One2.

Terraform quotes a cost of ~$100,000 for a 1 MW module that lasts 5 years, and a long-term module cost of ~$30,000 per MW at scale. At current prices, a low cost solar plant to power the modules would run to ~$1.1M (~$20 per MWh over a 25 year life, before financing costs)3.

The key here is that most costs are coming from the solar plant, not the Terraformer modules themselves (even after accounting for the replacements every 5 years over the 25 year life).

As a result, Terraform is a highly leveraged bet on cost declines in utility-scale solar PV, impressive in its purity and its audaciousness.

While Terraform founder Casey Handmer outlines a number of other uses for low cost solar here4, it is the production of synthetic methane as a drop-in natural gas substitute that has near-term world-changing potential (if the forecasted cost improvements can be achieved).

This becomes clear in comparing current costs with the ‘long-term’ costs outlined in the whitepaper:

In this comparison, the cost of solar PV declines by ~91% and the cost of the Terraformer modules falls by ~70% for an overall cost reduction of ~84%.

Put differently: given current costs, a 1 MW installation would produce ~55,000 kcf of methane over a 25 year operating life at a capital cost (including energy generation) of ~$28-30 per MMBtu5 (before any subsidies).

Under the ‘long-term’ cost structure, the same volume of gas could be produced for just ~$4-5 per MMBtu.

Assuming a long-term module cost of ~$30,000 per MW, these costs would range as follows based on the expected $ per MWh for energy:

Are these costs achievable?

They are definitely ambitious, but the ongoing ‘learning rate’ of ~20-45%6 for utility-scale solar PV installations would imply they are achievable over the medium / long-term7.

In particular, solar PV deployments appear to be accelerating8, led by China, implying at least 2 doublings in cumulative installed capacity between now and 2030.

The Terraform approach seeks to ride the wave of cost declines this implies.

Pairing cheap solar with the Terraformer in particular will also allow for reduced / eliminated power conversion costs, as well as avoided soft costs like ISO interconnection9 and (ideally) reduced design costs.

Against those costs, what sorts of revenue are possible?

In the whitepaper, the team lays out 2 key revenue sources (assuming US-based projects):

1. Revenue from gas sales (asserted at ~$10 per kcf)

2. Tax credits, primarily the 45V (~$31.80 per kcf) and 45E (~$13.50 per kcf) tax credits from the IRA10

Taken together, this sums to an average value of ~$50-55 per kcf produced, meaningfully higher than the expected capital costs (before financing and O&M).

Interestingly, in the short term the bulk of value will be coming from the tax credits, not the sale of natural gas.

This is good, because natural gas is actually pretty cheap (before accounting for environmental impacts).

In the US, following the ‘fracking revolution’, the long-term price for Henry Hub gas is likely closer to ~$3-4 per MMBtu than ~$10 per MMBtu, though this varies a bit by region based on pipeline capacity11.

There are extensive reserves in existing US basins that imply a medium-term ceiling in this range for the marginal MMBtu.

Marginal pricing is higher in places like Europe and Japan, and is generally set by the price for liquified natural gas imports, which cost an incremental ~$4-7 per MMBtu (if coming from the US) for a net price of ~$8-10 per MMBtu (again, long-term, which is how things like LNG facilities need to be priced to get built).

Even at the long-term full potential asserted by the Terraform team, it is not clear whether an unsubsidized PV + Terraformer installation can outcompete fracked natural gas in the US12. That said, there is significant market opportunity between here and there that still makes this incredibly exciting.

With current subsidies, the Terraformer’s themselves have the potential to payback in just a couple years. This should drive enough experimentation for us to see what could be possible at scale.

Customers: Where will early demand come from?

Early customers are likely to be categorized along 2 vectors:

1. Higher willingness to pay for (carbon-neutral) natural gas

   1. In an absolute sense (e.g., Europe)

   2. Due to its carbon neutral pedigree (e.g., low carbon fuel standards, strategic corporate offtake, subsidy-supported prices)

2. Lower cost for marginal electricity

   1. Due to cheap incremental renewables (e.g., sunny Australia)

   2. Due to overbuilt solar capacity (e.g., curtailed solar PV, datacenter microgrids)

There are a whole range of archetypes that fit this bill, but a few concrete ones include:

• New datacenters (or other microgrids) with overbuilt solar generation and / or a preference for a carbon-neutral gas backup generator

• Gas purchasers subject to things like California’s Low Carbon Fuel Standards

• Gas purchasers in e.g. southern Spain whose alternative cost of gas is high, value of carbon neutrality is similarly high, and price per MWh of solar is low

• Gas utilities in the US southwest that want to avoid becoming obsolete (or regulated out of existence); particularly if pipeline constraints can be sidestepped

Value chain: What role will Terraform play?

Terraform asserts a long-term goal to be a manufacturer of Terraformer modules, leaving the roles of project development, operation, ownership, and maintenance to others.

While there are many ways this could evolve in practice, over the medium-term ceding these other roles will require a significant push to enable standardization of key elements across projects (in order to maintain Terraform’s low cost value proposition).

Alongside Terraform’s efforts to develop pilot projects and de-risk its technology, this could look like the development of:

• Standardized module specifications to streamline design and permitting through the reuse of common information (incl. integration with common design tools)

• Standardized module designs to allow for manufacturing at scale and to simplify / speed construction

• Standardized offtake / business terms to speed project development and financing (incl. capture of subsidies and tax credits as applicable)

In effect, Terraform needs to run down the path towards economies of scale that solar PV and battery energy storage systems (BESS) have followed over the last decade.

In practice, Terraform will likely need to play a much more active role in initial projects (both to make sure they get done, and to enable the iterations that will drive good standardization). Success here will depend on establishing offtakes with key end-customers and getting developer and financial partners comfortable with how these projects work.

Midstream gas companies will also be key stakeholders here, both in the deployment of gathering lines to aggregate production and as the transmission backbone.

It will be particularly interesting to see who the right operator of these assets is long-term. One can imagine a spectrum of operational complexity from low to high, with different players participating depending on what prevails:

• Lowest complexity (akin to standalone solar today): No real operational complexity. Maintenance is contracted out, with ownership (after commissioning / early year ownership requirements) sitting with financial sponsors.

• Low-Moderate complexity (akin to power plants / fracking today): Potential for IPPs13, shale producers, or utilities (with generation) arms to run portfolios of assets. There could also be new pure-play development / operating companies.

• High complexity: Traditional hydrocarbon producers (e.g., Shell), though the market likely wouldn’t take off if the Terraformer entailed high operational complexity.

Terraform’s goal is for their modules to not be any more complex for a developer, installer, or financier than the addition of a BESS component to a PV project.

Achieving this will allow for early Terraformer projects to leverage existing expertise and infrastructure to a great extent, even if the ultimate size of the market at full potential will support significant specialization.

Competitors: Who else is going after this?

Given their integrated approach across solar / hydrogen / direct-air carbon capture / syngas, there are a number of competitors whose offerings fully or partially overlap with what Terraform is proposing.

I’ve mostly focused here on the venture-backed companies hoping to bring something new to their respective markets. Given the maturity of the underlying technologies, there are a number of existing industrial equipment manufacturers, developers and EPC firms, and other energy and chemical players that participate in related markets. In expectation, these existing players are unlikely to match Terraform’s radical low cost approach.

Synthetic fuel equipment manufacturers (natural gas):

General Galactic: Founded more recently than Terraform, but appears to be following a similar approach. Key personnel come from aerospace backgrounds. Based in El Segundo. Raised a ~$1.9M pre-seed round in July 2023.

Rivan: Basically the London version of Terraform, though perhaps without the prodigious blog output. Minimal publicly available info.

Turn2X: Broadly similar, but leverages biogenic carbon dioxide. Claims a patented reactor design. Does not seem as married to the ultra low cost approach as e.g. Terraform and General Galactic. Based in Munich. Raised a ~$4.7M seed round in June 2023.

Electrochaea: RNG company focused on a biocatalyst in place of a Sabatier reactor; agnostic about carbon dioxide and hydrogen sources (imply use of standard off the shelf equipment). May not be sufficiently committed to low cost approach to move the needle long term. Based in Planegg, Germany. Raised ~$65M to-date, including a ~$41M Series D in January 2022.

Sunfire: More traditional manufacturer of electrolysis and reactor equipment, some leveraging process heat and steam. Equipment more tuned to factory scales and customers, not RNG installations in the desert. Based in Dresden, Germany. Raised ~$1B to-date, including a ~$235M Series E in March 2024.

Hydrogen / e-fuel project developers:

Tree Energy Solutions (TES): Project developer / energy company that pairs renewable-driven hydrogen electrolysis with biogenic carbon dioxide in a Sabatier reactor to generate “e-NG” (Sample project here). Based in Brussels, US office in Houston. ~$250M in funding (Series C of ~$150M announced April 2024).

Appears to be jumping straight to the gigawatt scale with a traditional high capex approach. Unclear how price competitive their end product is (at current and future energy prices).

Hy2gen: Project developer / operator looking across the ‘power-to-X’ spectrum from stand-alone hydrogen to methane, ammonia, methanol, SAF / Diesel / Gasoline. ~9 announced projects, concentrated in Europe. Similar to TES, appears to be following a traditional high capex approach, though with more variation in project specifics.

Based in Wiesbaden, Germany. ~$220M in funding, last financing February 2022.

Ambient Fuels: Project developer, focused on green hydrogen. Based in New York. Seed funding from SJF Ventures, ~$250M in project financing from Generate Capital.

Traditional energy companies like Orsted, Equinor, BP, and Shell are also pursuing scale projects here, though commitments appear dynamic as regulatory frameworks and subsidies are hammered out14. It has seemingly been challenging for traditional chemical manufacturing approaches to reach an attractive price point for long-term hydrogen off-takes, even with subsidies.

Hydrogen electrolysis:

Electric Hydrogen: ‘Manufacturer’ of ~100MW PEM hydrogen electrolysis plants. ~45-53 kWh per kg with output at ~32 barg. Online calculator indicates ~$800 per kW of upfront capex, with ~$250 per kW for electrolyser stack replacement down the road (roughly in line with the System X archetype I laid out here).

Based in Natick Massachusetts; office in the Bay Area. Raised ~$800M in funding (‘first hydrogen unicorn’) with a ~$380M Series C announced October 2023.

Ohmium: Manufacturer of PEM electrolysers. Modular cabinet / rack-based system designed to consume ~0.5MW per hour producing ~9kg hydrogen (~49-53 kWh per kg). Two configurations that differ on output purity and pressure (~10 and ~30 barg).

Based in the Newark, CA, manufacturing in Bengaluru. ~$295M in funding (Series C of ~$250M announced April 2023).

Hysata: Alkaline electrolyser manufacturer, focused on high efficiency cells (~41.5 kWh per kg). Unclear how far along they are in integrating these technological advances into saleable products (though the recent Series B will help).

Based in Wollongong, Australia. ~$140M in funding (Series B of ~$111M announced April 2024).

Verdagy: Alkaline electrolyser manufacturer, focused on commercializing a 20MW modular design. Currently ramping up manufacturing in Newark, CA, supported by DoE grant. Energy requirements per kg and capital costs are unclear.

Based in Moss Landing, CA. ~$156M in funding (~$73M Series B in August 2023, ~$40M DOE grant in March 2024).

H2Pro: Manufacturer with a proprietary E-TAC pathway (rather than electrolysis) focused on developing a high efficiency (~42kWh per kg) 25MW module. Targeting initial delivery in ‘2025 and 2026’.

Based in Caesarea, Israel. ~$107M in funding (~$75M Series B in 1/25/2022).

Bloom Energy (BE): While the overall business is focused primarily on their ‘Energy Server’ microgrid solution, Bloom has also been touting a solide oxide electrolyser that can produce hydrogen for ~37-44 kWh per kg in 1.2MW modules (and requires high temperature steam as an input). Unclear how much commercial momentum there is here, but they seem less likely to go bankrupt than Plug Power (PLUG) . Based in San Jose, CA.

Sunshine Hydrogen: ~$90M in funding (January 2024), based in Beijing. Little publicly available info (in English) besides their Crunchbase profile.

Direct-air capture:

In this category, I’ve focused on companies with solutions that can produce a stream of carbon dioxide for use in downstream industrial processes15. The general shape of these businesses is: capture carbon dioxide with sorbent (often with fans to increase airflow), apply energy (usually heat) to release carbon dioxide in a stream, do something with the carbon dioxide (often sequester it underground).

As with electrolysers, there are a variety of approaches that aim to optimize both capital costs and energy usage, though few seem as willing to waste energy to get as low capex as Terraform. These aren’t necessarily competitors, but potential sources of breakthroughs and market development alongside Terraform.

Climeworks: ~$800M raised, based in Zurich. Solid sorbent-based contactor, which allows for low-grade heat from geothermal for regeneration (~100 degrees versus ~900). Mammoth pilot facility removes ~36,000 tonnes of carbon dioxide per year.

Heirloom: ~$54M raised, based in San Francisco. Similar calcium carbonate / calcium oxide pathway as Terraform. Pilot project in place in Tracy California.

CarbonCapture: ~$115M raised (~$80M Series A in March 2024), based in Los Angeles. ‘Deeply modular’ approach that is flexible to accept new sorbents over time. Pilot project being developed in Wyoming.

Verdox: ~$100M raised, based in Woburn, MA. Novel system that uses voltage applied across the absorption medium to bind and release carbon dioxide. Interesting alternative to kiln / heat-based regeneration.

Carbon Engineering: Owned by Occidental Petroleum (OXY), based in Squamish, BC. Uses a potassium hydroxide solution to bind the carbon dioxide, which then goes through a number of steps to get to pure compressed gas release.

RepAir Carbon Capture: ~$11M raised, based in Israel. Uses selective membrane to capture and release carbon dioxide, without requiring high heat for regeneration; claims ~600kWh per ton of carbon dioxide removed.

Noya: ~$11M raised, based in San Francisco. Uses an activated-carbon monolith for the sorbent material which allows for more efficient regeneration.

Sustaera: ~$11M raised, based in Durham, NC. Relatively standard approach based on info released so far.

Differentiation: What makes Terraform unique?

Across this range of adjacent players and competitors, Terraform still has significant space to carve out a compelling offering.

In particular, their integrated approach has the potential to radically squeeze out costs, including soft costs around development and design, where others have been constrained.

Electric Hydrogen, for example, is pitching their PEM electrolysers as having industry-leading costs at current energy prices. They have secured significant funding to expand manufacturing. But by sizing their systems (modular in ~100MW increments) for use by major chemical manufacturers they are tying themselves to the utilization requirements of continuous process chemical plants and the custom design expectations of chemical megaprojects. This is great for decarbonizing ammonia production by displacing steam methane reforming, but will not foment an energy revolution.

Similarly, multiple carbon capture startups have proposed detailed plans for their pilot installations, but must manage delicate coordination across partners to provide the renewable electricity and Class IV injection capacity to make their projects possible, and the ongoing maturation of the carbon credit market to make their projects profitable.

I mean no criticism here; these are rational choices that reflect the historic reality of expensive energy costs and immature markets.

But by integrating across steps in the value chain—bridging cheap and simple solar to universal and transportable methane within a low-cost, modular package—Terraform has the potential to sidestep these constraints, repackaging ever-lower-cost PV energy for a globally tradeable end-market.

This will take significant product discipline, and likely more than a bit of luck, but there is a germ of potential here that could change the world.

Other potential challenges / questions

Aside from the steep learning curve to ride, there are a number of potentially more prosaic challenges Terraform (and it’s direct competitors) could face:

Gas interconnection / takeaway: While the Terraform approach conveniently sidesteps grid interconnection queues, the time and cost for gas interconnection is not zero. Building out gathering systems won’t happen overnight, particularly in regions like the southwest that don’t have them already (not being traditional hydrocarbon-producing areas). Trucking is an option here in the short term—composite tube trucks can carry more than 500 kcf of methane—but this will eat into initial cost effectiveness.

Either way, pressurizing gas for transport / transmission will add further costs, as could the addition of storage to levelize the flow of gas from Terraform projects.

Water consumption: Similar constraints and potential additional costs surround the consumption of water in the Terraform process (primarily from electrolysis). While there is potential to capture water as part of the DAC circuit, in the early days the trucking (or well extraction) of water could be the more cost effective option (though still non-zero).

Electrolyser efficiency in the desert: I’m sure the Terraform team has already thought of this (and maybe already solved it!), but many hydrogen electrolyser manufacturers warn of degradation in efficiency above ambient temperatures of 40 celsius / 104 fahrenheit. Given that a core part of the Terraform pitch is to site projects in sunny deserts, high heat will need to be dealt with (particularly given the tight linkages / lack of buffers between different module components).

Permitting: While the on-site use of solar power avoids the grid interconnection backlog, it does not obviate the need for other permits and permissions. It may need to be confirmed exactly how this type of mixed facility will be permitted, and whether the inclusion of natural gas production on-site might make this meaningfully harder. The need for permitting reform won’t be news to the Terraform team (or anyone else trying to build things in the US) but it could slow down the transition from early pilots to scale deployments in particular, potentially creating space for competitors.

Carbon accounting: There has been significant focus in recent years on the quality of carbon capture approaches and on the accounting for avoided carbon emissions like what is being proposed by Terraform. I’m reminded here of the moralizing in some parts over the use of carbon capture and storage to ‘delay’ the retirement of gas and coal plants. Terraform is getting ahead of this with a 45V life cycle analysis with Energywerx, but there may be ongoing cultural skepticism of carbon-neutral methane.

Low-cost competition: This falls into a potential ‘good for the planet, bad for Terraform’ bucket, but there’s a risk that at scale the market for low cost syngas equipment could be dominated by e.g. China rather than an American startup, at least in many markets.

The panels already come from western China. China is leading the buildout of solar PV over the next decade. China is a major buyer of LNG, with a relatively high marginal cost per MMBtu. The global energy market is huge, and there is space for many competitors, but strong low cost competition would imply long-term margin compression for manufacturers (similar to what’s occurred in panels, batteries, and other moderate complexity manufacturing). If this happens, it will not obscure Terraform’s achievements on behalf of humanity.

Other potential tailwinds / benefits

Beyond the already-mentioned PV learning curve, there are a few things that could help them in particular over the next ~18-24 months:

Early start: The Terraform team seems a bit earlier out of the gate than some of their more direct competitors like Rivan and General Galactic, though synthetic fuel is a crowded space more generally and it will be interesting to see how it evolves as larger projects start to come online.

Strong team: Casey has built a strong team to develop the initial prototypes, though Terraform will now need to build out the team further (likely following a series A) to establish the next tranche of growth.

Delayed hydrogen projects: As Treasury has refined guidance around hourly matching and additionality for renewable power sources for green hydrogen credits, ‘traditional’ scale approaches to hydrogen manufacturing have become less compelling while novel approaches (that natively match renewables) have gained ground. This will make Terraform’s approach to energy transformation relatively more appealing, while also potentially redirecting industry focus away from marginal hydrogen projects.

Carbon credit market maturation: There will be significant momentum from all the non-Terraform participants pushing the ball forward here. Terraform should be able to largely focus on figuring out how it fits into frameworks that others develop, rather than needing to create frameworks to capture the benefits it creates.

Interest rate normalization: Financing costs are significant for high capex / low opex assets like solar PV installations. If interest rates retreat from recent highs, that could provide a welcome tailwind16 to project financing for Terraform’s rollout.

Next Steps

Overall, the most exciting thing about Terraform is the extent to which it embeds the new logic of energy and markets circa 2024, rather than trying to wedge old ways of doing things into carbon-neutral shapes.

That will make their road easier, but there is still significant work to do. From my outside-in perspective, it seems like the team will need to be focusing on:

• Raising the Series A: The team is coming off an important milestone of demoing their end-to-end process. They appear to already be doing this, but it’s likely the right time to raise another round of funding to maintain their momentum towards scale pilots. Given the potential for rapid progress, it likely also makes sense to start cultivating project finance partners.

• Building out the team: This should focus on engineering capabilities to establish full-scale modules for deployment, and commercial capabilities to develop projects and manage customers / partners. As the technology gets de-risked it will be important to work with developers / offtakers / financiers to ensure the modules developed are deployable and bankable at scale. This may look like formal product management, or happen in concert with early project development.

• Developing pilot projects: Getting the first full scale module working in the lab will be important, but that shouldn’t preclude laying the groundwork for early field deployments. Given the time required for permitting and stakeholder coordination, the first / second / third deployments need to start being developed now to maintain forward momentum.

This will allow them to continue to prove out their technology while setting the stage for a period of explosive growth once their model is proven.

This pattern is generalizable: when utilization of an asset is low, the ‘inefficient’ process with lower upfront capital costs can be more cost effective overall.

This is relevant because solar and wind generation have two key characteristics:

1. Their cost is declining precipitously (~$15-30 per MWh and falling in key regions)

2. Their effective utilization is low (~20-25% for PV in sunny regions; ~40-55% for wind)

As a result, processes that can be run off low-utilization cheap renewable energy (particularly solar) can potentially beat higher capex processes run at higher utilizations. This shift to a renewable-powered industrial stack has the potential to create a step change in renewable energy deployment, but will require a rethinking of how we use that energy (as Casey Handmer outlines more extensively here).

Take two prospective hydrogen electrolysers: X and Y

X requires ~$1,000 in equipment costs1 per kilowatt (kW) of capacity, but can produce a kilogram (kg) of hydrogen with just 50 kilowatt-hours (kWh).

Y only requires ~$70 in equipment spend per kW, but needs ~80 kWh to produce a kg of H2.

Put simply: X is more expensive upfront, but less expensive to run.

So, under what conditions will system X be better? Given different energy cost assumptions, we can calculate how many kilogram’s of hydrogen must be produced before we are indifferent between the two approaches (and after which system X is better).

If we can procure energy at ~$0.04 per kWh, the two systems would reach cost equivalency after ~775 kg of hydrogen production (the darker green line in the chart).

At ~100% capacity factor (i.e. all 8,760 hours every year), we would expect a 1 kW version of System X to produce this in ~4.4 years, and a similar size System Y to produce it in ~7.1 years2.

At that production volume and energy cost, the breakdown of spend across the two approaches is:

Thereafter, System X would be producing power more cost effectively at ~$2.00 per kg in energy costs versus system Y’s ~$3.20 in energy costs.

Under these assumptions, System X is pretty obviously superior.

The wrinkle is this: it is challenging to procure energy for ~$0.04 per kWh from anything other than intermittent renewables (e.g., wind and solar). The use of renewables is also a pre-requisite for qualification for the maximum green hydrogen tax credits under the Inflation Reduction Act (IRA)3.

With this context, two things need to change about this scenario:

1. The capacity factor drops (e.g., ~20-25% for solar PV, ~30-55% for onshore wind4)

2. The potential energy costs could be even lower (~$0.02-0.04 per kWh and falling)

With this, we can start to understand how sensitive the comparison is to specific assumptions.

In the below, we see the # of years5 for costs to equalize based on potential utilization % and cost per kWh, with the red cells indicating shorter timelines (system A wins faster) and blue cells indicating longer timelines.

In general, any scenarios under which System X takes more than 15 years to ‘win’ is in effect a win for System Y once one factors in capital costs6 and market uncertainty.

We can take our toy model one step further7 and gesture at financial returns.

So let’s say we have access to energy from an incremental wind farm with a ~55% capacity factor at ~$0.024 per kWh. This is low cost, but within reason for the United States today (before subsidies).

At this price and capacity factor, over a 20 year system life we’d expect to generate hydrogen at a cost of ~$1.72 per kg for System X and ~$1.98 for System Y. If we’re able to sell our hydrogen for ~$3 per kg, we’re earning ~$1.28 per kg from System X and ~$1.02 from System Y respectively.

Another win for System X right?

Well, this is where the financial aspect comes in. From an pure cashflow perspective, for a given energy input and capital spend, System X pulls ahead around year 8.

But, once we discount these cashflows at ~12% to reflect a rough cost of capital, we find that System Y outperforms due to it’s capital light approach.

So, if I had a stranded wind farm (for some reason), I’d hook it up to System Y and get cash now rather than later.

To the extent one believes that energy costs will continue to fall, particularly for low-capacty-factor solar, this implies that the System Y approach will ultimately win out.

This matters, because big $$ are flowing into ‘green hydrogen’ production, and it is not obvious all the relevant stakeholders have internalized the ongoing cost declines of input energy. This will have a big impact on which projects will ultimately be cost competitive (particularly given the lack of forward purchases confirmed so far).

That said, this is great news for hydrogen end-users and the ultimate. Less expensive approaches to electrolysis, predicated on the intermittent availability of cheap renewable energy, will help pave the way to a low carbon future one way or the other.

Subscribe now