I took a sabbatical in 2025 to figure out the next chapter of my career. I ultimately turned my attention to energy–specifically, electric power. I coalesced my thoughts into the essay below. Much of my thinking was inspired by posts from Casey Handmer, Austin Vernon, Brian Potter, and Noah Smith. Insights are theirs; errors are mine.

Since writing this in July, I joined Gridmatic to accelerate deployment of cheaper, cleaner energy onto the grid.

Thank you to Michael Retchin for encouraging me to share this essay. I'm a few months delinquent, but better late than never.

Why I want to work on energy

Our ability to harness and use energy is directly correlated with almost all measures of human development. These trends have held over millennia; you can see the fall of the Roman Empire in the data. Energy consumption has accelerated since the industrial revolution, as have most measures of well-being, including life expectancy, economic output, and total factor productivity.

The last few decades—since the 1970’s—are the first time in human history we’ve enjoyed GDP growth in the US without a corresponding growth in energy usage. That said, things haven’t exactly been great since then. We ride on public transit from the 1920’s. We get water via pipes from the 1940’s. We fly in airplanes from the 1960’s. We take high speed rail from... oh wait.

Energy limits our ability to drive change in the physical world—from desalinating water to moving people around a city to manufacturing goods. The public cloud and AI build outs have instigated the first increase in US electricity consumption in over two decades. That said, this increased consumption is, if anything, a return to normalcy. In the early and mid 20th century, electricity consumption increased rapidly.

We used this power for public transit, assembly lines, smelting, street lighting, household appliances, air conditioning, elevators, and more. Similar to the bandwidth glut of the early 2000’s, an energy glut will open new, unpredictable, second-order opportunities.

Energy liberated 90% of humanity from the life of a subsistence farmer. It replaced the doomsday prophecies of Malthus and Ehrlich with mass food production, manufacturing, and transit. To realize a future of increased abundance, the US must return to the Henry Adams Curve.

The US energy market today

The US energy market is absolutely massive. Electricity accounts for about a third of energy consumption. Most of the remainder comes from petroleum products and natural gas. Residential and commercial users rely on natural gas for space heating, water heating, and cooking—over 50% of US houses still use natural gas heaters. In addition to space heating, industrial customers use natural gas for manufacturing processes (lots of very hot furnaces). Transportation almost entirely relies on petroleum-based fossil fuels. In addition, petroleum fuels the production of many materials—everything from plastics to fertilizers to perfume.

I expect this breakdown of energy usage to change dramatically in the next two decades. As electricity becomes cheaper, we will see the electrification of transportation and of HVAC. Right now, the transition is slow and mostly outside the US.

Trends in electricity markets

Solar drives falling generation costs

Electricity generation costs continue to fall. Solar primarily drives this decrease, though there are a few dark horses: fission (small modular reactors), fusion, and geothermal. (It’s cheap to pull geothermal energy out of old fracking wells).

I’m most excited about solar because of the learning curve effects: the price of solar photovoltaic modules drop as cumulative shipped volume increases. In the 20th century, we became really good at building megafactories that churn out cheap products. Much of what was formerly expensive—flatscreen TVs, personal computers, etc—has become insanely cheap. Solar PV cells allow factories to manufacture energy; I and many others are convinced they will continue to become cheaper.

Storage is becoming cheap for the first time

Traditionally, the only good way to store electricity was pumped hydro. These facilities are hard to site (you need a mountain or a deep mine), so their use has been limited. Lithium ion batteries not only enjoy the same learning curve effects as solar PV modules but also enable more solar adoption.

Solar + Batteries

Solar and batteries complement each other. Batteries ameliorate solar’s daily intermittency (left). As a result, the vast majority of near-term capacity additions (above) and projects in the interconnection queue are solar and battery facilities. Most new solar and battery additions are in Texas and California (right). This location-dependence results from how the Texas and California grid operators (ERCOT and CAISO) handle new generation; it has little to do with technological limitations (or how much sun Texas gets). There’s still so much whitespace.

Positive feedback loops

Exponential growth requires positive feedback loops. I list a number of potential drivers of the learning curves for solar pv cells and batteries. More produced → cheaper price → more demand → therefore, more produced. The beginning of exponential growth is already evident. What will we do with cheaper electricity?

Electrification of transportation. Electric vehicles will become cheaper than internal combustion vehicles. Their design is far simpler, their reliability is better, and their variable costs are lower. As the price of electricity and batteries fall, EVs will become even cheaper. More demand will drive further price reductions as battery production increases and charging infrastructure proliferates. Internal combustion car improvements have plateaued. It’s only a matter of time before EVs dominate. Just look at China (see next section).

Electrification of other existing use cases. Much of non-electrical energy consumption in the US goes into heating—mostly HVAC but also cooking and water heaters. Natural gas prices have remained largely steady over the last decade, after a dramatic earlier decrease due to hydraulic fracking. Eventually we’ll turn to electric heating and cooking (req. ~60% price drop), driving more demand for electricity & generation, and fueling the learning curves.

New use cases become economically feasible. Most industrial processes require electricity. The cost of electricity determines the economic feasibility of these processes. (In 1940, production of aluminum and magnesium alone required 1/7 of all electricity in the US!) Newer examples include bitcoin mining, and desalination in the Middle East, where water is relatively expensive and energy is relatively cheap. Cheap enough electricity totally changes the calculus of many processes. We can desalinate the Salton Sea to harvest huge amounts of lithium in our backyard. We can supply water to Southern California without destroying natural habitats. We can economically operate vertical farms that deliver fresh, clean produce to megacities. We can even synthesize carbon neutral natural gas directly from water and atmospheric carbon dioxide. Extracting fossil fuels out of the ground, in the long run, is only getting more expensive. At some point, we’ll be synthesizing these still-essential materials directly not only because it’s more sustainable but also because it’s just cheaper.

The vicious cycle. Both solar and battery-produced energy are mass-manufacturable and therefore enjoy learning curves effects. This is their most important attribute. (Small modular nuclear reactors might benefit similarly; fusion does not.) As demand for electricity increases, electricity will get cheaper, driving more demand, spurring more production, further decreasing the price. The best industrial revolutions are those that fund themselves; they cannot forever rely on government subsidies. I believe solar and batteries are nearly there, and I look forward to the energy glut ^[1] that will follow. China, I believe, is a harbinger of what’s to come.

China: the electrification bellwether

Coal, solar, wind, and hydro account for 90% of China’s electricity generation. Solar power now dominates new electricity generation. This lead isn’t because China cares about the environment ^[2]; it’s because solar is cheaper.

Cheap power has prompted an electrification of end use cases. EVs are becoming cheaper than regular cars, driven by cheap electricity and batteries. Buildings and industry, traditionally heavy users of natural gas, are also shifting to electricity. I believe China is a harbinger of what’s to come, as the economics of solar and batteries there begin to diffuse to the rest of the world.

Energy market structure

The energy market is mind-blowingly massive: a multi-trillion dollar market globally. Yearly profits in oil and gas are on the order of hundreds of billions of dollars, with most revenue going to the upstream players. (It pays to dig holes.) Power is a ~$500B market in the US, which I expect to grow over the next few decades. There are three major categories of players in these markets: energy producers (generation); energy consumers (distribution); and those that move energy from the first to the second (transmission and/or trading).

Electricity markets

In an electrified world, there will be a huge opportunity for companies that can internalize risk on behalf of end users. These companies will facilitate the large-scale use of very cheap but highly volatile generation sources (my hypothesis: primarily solar) and will ameliorate the impact of the aging and sclerotic transmission infrastructure. The grid originated to mitigate risk. New sources require an evolution in how it is managed and operated.

Producers: I believe legacy producers will get squeezed. This is already happening to the manufacturers. Similarly, there’s now a glut of grid-scale storage batteries in TX. Learning curve effects are great for consumers but put producers on an accelerating treadmill.

Consumers: End users do not want to run trading desks. They want their electricity at cheap, reliable prices that they can plug into Excel spreadsheets. Entities that are willing to internalize risk on behalf of end consumers will play an increasingly valuable role in this supply chain. This role becomes more important as electricity generation shifts to more volatile sources, especially given sclerotic transmission growth.

Risk managers: In a solar-dominated grid, integrated load service entities—ones that control both generation assets (e.g., battery farms) and end-user relationships (e.g., PPAs with large industrial or commercial entities)—will become the dominant players. These players have the best access to load information through their retail contracts, and they can better hedge price volatility by owning generation assets. Similar to the oil and natural gas industries, the electricity industry will likely consolidate, with major players owning various categories, defined by the generation technology, customer profile, geography, and so on.

Final thoughts

We’re in a one-in-a-century transition of humanity’s marginal energy source. New technologies are decreasing the price of electricity generation, and new use cases are increasing electricity demand. These trends show no signs of abating.

At the turn of the 20th century, Standard Oil enabled the US to transition from coal to petroleum as the marginal energy source. The internal combustion engine replaced the steam engine. This transition gave us everything from ubiquitous lighting to air travel. I believe that the next Standard Oil will emerge in the coming decades, unlocking a new era of gains in human capability and prosperity.

Acknowledgements

Thank you to Alex Kurland, Konstantine Buhler, Logan Engstrom, Logan Spear, Max Franzblau, Michael Retchin, Ryan Cohen, and Zach Diamandis for helpful comments on earlier versions of this essay. I'm also grateful to the many others who gave me advice during my sabbatical.

Footnotes

[1]	I expect these changes to play out over a ten to twenty year horizon. That said, the early signs are already here. The US is projected to increase solar by ⅓ in 2025, and batteries often supply ~25% of CA electricity in the evenings. (This number might be out of date soon.)

[2]	China leads the world in coal power additions.

Supplemental Info & Figures

This year's JP Morgan report had lots of great figures. I included many below.

Worldwide solar growth is rapid, well ahead of most predictions.

US solar growth and capacity has not yet caught up. Panels in the US are 2-3x more expensive per watt than in the rest of the world. More expensive panels require axis trackers to justify costs, driving up labor and other materials costs. Less expensive modules (around ⅓ of current costs) will quickly reduce fieldwork and balance of system costs, which will further depress utility scale solar price. I hope we hit a “tipping point” in the next 5 or so years.

Transmission line growth in the US is pathetic, resulting in high congestion costs. As far as I can tell, the fixes are policy: eliminate much of NEPA, and give FERC siting authority for high voltage transmission lines, similar to its authority for natural gas lines. In the meantime, we’ll likely see more growth within individual states (TX, CA) and more distributed energy generation and storage. I’m optimistic this policy problem will be solved in the medium term.

Electrification of heating still doesn’t make sense, but with an order-of-magnitude drop in electricity prices, we’ll see large-scale deployment of electric heat pumps. I suspect we’ll see this start within the next decade.