The Marginal Energy Source's Past
Humanity typically has one "marginal" energy source that determines how much work we can do and where it can happen. Availability and portability are key attributes. It usually becomes expensive because usage grows until supply is tight. Other sources might be important when available, but it is the marginal source that drives much of humanity's growth.
As authors like Vaclav Smil have documented, humans started with foraging in the hunter-gatherer phase. Supplies were limited and inconsistent. Next, agriculture provided humanity with grain, allowing the storage of food and the domestication of more animals. These animals, and the ability to feed them outside of wild forage, pushed energy, industry, and transportation beyond human muscle across a wide domain.
There were secondary sources of energy. Wood was mostly limited to heating. Wind and water power were sometimes available, but extremely context-dependent, and only applicable to oceans or rivers. Low productivity and the enormous land and labor inputs required for pre-industrial grain production limited growth and muscle power (human or beast).
Coal was next, and came of age during the Industrial Revolution. Steel, steam engines, railroads, chemicals, and electricity increased the productivity of a human by orders of magnitude. Its weaknesses were transportation cost and the inability to scale down in transportation applications. Today, almost every country with decent coal reserves is at least middle-income, but energy-poor countries often cannot afford the price of imported coal.
The Age of Oil
We undeniably live in the age of oil. Oil became increasingly important in the 20th century, but dominated after World War II. Cars, highways, semi-trucks, airliners, and plastic all grew at astounding rates in places like the United States. On the supply side, American prospectors found absurdly large reserves in the Middle East and elsewhere. Oil's transportation costs are minimal. It is simple to pump and has excellent energy density. Cheap energy could spread everywhere. That portability meant oil was not just transforming transportation and materials but powering basic industrial processes and fueling marginal electricity production.
Market Share of various ammonia feedstocks over time.
Unfortunately, oil's supply hit a wall in the 1970s. Prices post-1973 have been higher and more volatile. Oil remains the universal marginal unit of energy, but it can't meet the demand it would with a better supply. The global economy has reorganized to limit oil use outside of transportation, sapping labor and capital. Oil usage fell in process heat, electricity generation, and petrochemicals in favor of coal and natural gas. Only the most energy-poor places, like Hawaii, regularly burn oil for electricity due to the extreme cost. I find it jarring to see heavy fuel oil as the default energy comparison in 1950s and 1960s engineering reports.
Oil prices aren't the same.
It is also a part of the "What happened in the early 1970s?" meme. Did we collectively give up at once, or did the scarcity of our marginal energy source force us to reorganize around less productive energy sources?
The limits of oil have encouraged the development of alternative resources, though none have taken the crown. It is challenging to beat oil’s energy density, cost of transportation, ability to scale up and down, and flexibility.
The Importance of Complementary Technologies
Humans devise technologies to harness new energy sources for heating, transportation, work, and agriculture in a positive feedback loop. Animal husbandry, plows, and plant domestication are a few classic examples from the age of grain. Coal is of little utility without steam engines and blast furnaces. Humans mostly burned pools of oil in jars before turbines, petrochemicals, and the internal combustion engine brought oil into dominance.
A feature of complementary technologies is that inventors often optimize them for a specific energy source. Coal cannot easily power a car or aircraft. The gap between discovering a new energy source and how to use it most effectively can take decades (or centuries). Typically, the production technology for raw energy improves by serving an early market. Coal had heating, and oil had lamps. Lower raw energy costs and scale allow other markets to become viable.
Solar PV as a Universal Marginal Energy Source
Solar PV has features that could make it the universal marginal source of energy.
-
Solar Panels are Lightweight
The active material in a solar panel is only a few pounds, and the entire panel is light enough for one or two humans to carry. One modern panel produces the same amount of electricity in 30 years as 10 tons of coal. That is 300x more energy per mass. Even places with terrible infrastructure can cost-effectively import energy by purchasing solar panels.
Source
The mass of silicon per watt has fallen precipitously over time, down to 2 g/watt. A paper analyzing modern panels finds that the energy return on investment would be near 30:1 at a 12% capacity factor. Many of the largest producers hope to shrink wafer thickness again because emerging technologies like TOPCon do not lose efficiency at thinner wafer sizes. The grams per watt could easily halve. Items like aluminum frames will become the largest source of embodied energy.
-
Solar PV is Modular and Scalable
Energy use cases can range from charging a cordless electric drill up to powering a gigawatt-scale aluminum smelter when paired with the appropriate complementary technologies.
-
Sunlight is Widely Distributed
The concentration of fossil fuels presents issues because it can be too expensive to transport, and reserves are often in unfavorable places. The sun shines virtually everywhere on Earth, and even a gloomy place like Scotland receives half the sunlight that sunnier places get. Energy access can be universal.
The Need to Move Beyond the Grid
The electric grid is a premium service that is too expensive and inflexible to provide the marginal unit of energy for many applications. Hooking up to the grid adds several hundred dollars per kilowatt in power plant capital costs, can tack on years to the project timeline, and charges ruinous transmission and distribution costs. Wholesale electricity prices are now less than half of what customers pay.
Distribution and transmission are expensive.
Industry and materials are the most cost-challenged by current electricity prices; grid costs are a non-starter, and they already avoid using it. Aluminum smelters and other large electricity consumers typically use "captive" power plants or have subsidized pricing. Transportation is replacing expensive liquid fuels, but truck charging points are likely to incorporate microgrids to reduce demand charges.
There is talk of relying on negative electricity prices for low-cost applications, but trends do not favor this strategy. Users must still pay transmission and distribution fees or build infrastructure to tap into high-voltage power, even if wholesale prices are free. Subsidies and inflexible thermal power plants staying online for operational reasons often drive negative prices. Subsidies will not last forever, and batteries now have enough capacity in markets like California to handle evening ramps, reducing the need for gas power plants to be online during the day.
Solar will need to directly power many applications and processes without the electric grid to evolve.
Solar's Enablers
A pool of oil has little utility for anybody. But, the internal combustion engine, gas turbine, catalytic cracker, and oil-fired boiler powered massive economic growth. Similarly, a solar panel alone is limited in impact without complementary technologies.
It was not clear that solar PV could be inexpensive enough to be a base unit of energy until recently. A thoughtful book, “Taming the Sun,” written in 2017, proposed using sunlight more directly through photocatalysts or solar concentrators. The rapid drop in solar PV costs means technologies can hook into arrays of cheap solar panels instead of developing more complicated and less mature technologies. Storage or variable output is still required to solve intermittency.
There are four primary enabler categories.
Vehicles and Vehicle Batteries
Batteries fill the role of the internal combustion engine in a solar-dominated world. They turn raw electricity from solar power into work to power cars and other vehicles. Batteries have different tradeoffs than internal combustion engines. "Fuel" costs are lower even at grid prices, there is less powertrain maintenance, fewer emissions, and convenient nightly charging. But, energy density is less than that of liquid fuels, even after adjusting for efficiency.
Lower energy density means engineers must design vehicles for batteries, and the economic incentives to do this are strong. A car manufacturer bears the cost of adding efficiency features to gasoline-powered cars, but only the customer pays for fuel. There are rapidly diminishing returns to further efficiency. The math is different with battery-powered vehicles because the battery pack is much more expensive than a fuel tank. Efficiency gains pay off because they reduce pack size and manufacturing cost. The best-selling EVs are much more aerodynamic, shed wasteful mass, have low rolling resistance tires, and have features like heat pumps to maximize efficiency.
Autonomous driving will accelerate the trend to electric vehicles as most fleet operators favor EVs.
Stationary Storage
Stationary storage batteries solve most of the "the sun doesn't shine at night" issues and improve the utilization of transmission and distribution assets. They are ~100x larger than car battery packs and charge/discharge ~5x-10x slower. Pack and cell architectures are evolving to much larger cell sizes than are practical in EVs (which need smaller cells in series to reach practical voltages). Larger cells mean fewer installation steps and connections, along with less non-productive mass. Cells with thicker electrodes could push these improvements even further.
More power electronics are being integrated at the factory rather than in the field, reducing time and labor. Installed pack costs, not including interconnection, should easily fall well below $100/kWh everywhere, and this is already the case in China. The battery component of the levelized cost of storage would fall below $50/MWh at $100/kWh installed if we extrapolate the 2023 Lazard methodology.
Industrial customers might buy electricity from co-located solar and batteries to save money, especially if they can reduce electricity demand on the cloudiest days of the year to limit overbuilding of solar capacity. A titanium processor in West Virginia and a steel micromill in California are already planning to use solar plus storage for most of their needs.
Ten-hour to twenty-hour batteries should also be feasible. A recent Volta report suggests LFP cell prices could soon be $30/kWh in China, even in smaller form factors. And that price does not consider chemistries like LMFP that might trade better energy density and cost for lower cycle life. Many thousands of cycles aren't necessary for these applications. Batteries with more capacity could help solar PV breach 50% market share and possibly 80%-90% in most electricity markets.
Any electricity source can charge vehicles or stationary batteries, but solar is likely to benefit the most, given its cost and generation profile.
Hydrogen/Chemical Feedstocks
Hydrogen, carbon dioxide, and nitrogen can be the low-level feedstocks for most petrochemical production. Hydrogen is the most challenging because it must be unbearably cheap to compete with fossil hydrocarbons. Alkaline electrolyzers appear to be making the most progress in scale and cost so far, but there are a few other possible options.
Heat + Thermal Storage
Thermal storage could provide 20%-35% of global energy use for process heat, district heating, and replacing fuel in thermal power plants. Fossil fuels are thermal storage, so the incentives were not strong before solar PV was cheap.
The case for continuing to use heat in industry is strong. Thermochemical techniques dominate many chemical processes because they have cost benefits from scale and often have much better selectivity than biological or electrochemical processes. Cost-effective solar-to-heat technology is important in maintaining (or expanding) high standards of living.
Electric heating elements and storage materials can be inexpensive, making them much more convenient than redesigning processes that are inflexible. In practice, it is challenging to make such low costs a reality.
I am the CEO of Standard Thermal. Our focus is on delivering thermal storage costs low enough to make months' worth of energy storage possible. I wrote a companion blog post to this one detailing our strategy and thermal storage more generally.
How Cheap Can Solar Get?
Cheap solar PV is critical for these applications to be practical. The most relevant cost metric is direct current (DC) cost since many applications might be fully or partially off-grid. There are several strategies to reduce this cost:
-
Adapt Solar Farm Design to Low-cost Panels
Technologies like trackers maximize the output of each panel. As costs fall, it becomes cheaper to “lay the panels flat” on much simpler racks. Each panel produces less electricity but also costs less to install. Shading is not a concern if the panels are flat, and density can increase 2x-3x compared to arrays with single-axis tracking. Non-panel DC costs can decrease below $100/kW.
-
Increase Panel Efficiency
Panel efficiency gains directly impact DC costs because high-wattage panels increase the electricity output for inputs like labor and racking. Today’s mass market panels are ~21% efficient but could rise to ~26% with silicon technology in the pipeline, providing a 20% cost reduction holding other inputs constant. Technology like perovskite tandem cells could push efficiency well into the 30s and further depress both panel and non-panel costs.
-
Continue Small Improvements and Automation
Larger panel form factors, automation of panel cleaning, simplified installation technology, robots, and other incremental improvements could continue to reduce non-panel costs. Using cheaper substitutes for aluminum, glass, and silver is another promising path.
These improvements could result in arrays that cost $100/kW DC under the most optimistic assumptions. The levelized cost of electricity could be near $5/MWh at an 8% discount rate. An energy cost of $5/MWh at customer sites would be disruptive. Energy costs that low are only available near fossil fuel mines/fields, and then only as heat. Costs like that available almost anywhere would be a gift for most of the world.
Opening the Next Era of Growth
The world has arguably been energy-constrained for fifty years, with most economic growth coming in regions with significant coal reserves. Oil has been expensive and a brake on growth. The oil and gas industry has only grown its reserves with high prices. It cannot maintain price stability like in the pre-1973 era. Advancements like shale gas have so far been regional phenomena rather than global forces due to high transportation costs for LNG. The tyranny of oil supply could ease as the manufacturing capacity for solar PV approaches 1 TW/year, with no real constraints to further supply. Deployments at that scale will quickly saturate electric grids and stall solar growth without the development of complementary technologies. The age of solar PV is still young, and it could power humanity’s next leap in living standards.
Expanding the Universal Marginal Energy Source
2025 August 18 Twitter Substack See all postsSolar PV is poised to change the global energy landscape.
The Marginal Energy Source's Past
Humanity typically has one "marginal" energy source that determines how much work we can do and where it can happen. Availability and portability are key attributes. It usually becomes expensive because usage grows until supply is tight. Other sources might be important when available, but it is the marginal source that drives much of humanity's growth.
As authors like Vaclav Smil have documented, humans started with foraging in the hunter-gatherer phase. Supplies were limited and inconsistent. Next, agriculture provided humanity with grain, allowing the storage of food and the domestication of more animals. These animals, and the ability to feed them outside of wild forage, pushed energy, industry, and transportation beyond human muscle across a wide domain.
There were secondary sources of energy. Wood was mostly limited to heating. Wind and water power were sometimes available, but extremely context-dependent, and only applicable to oceans or rivers. Low productivity and the enormous land and labor inputs required for pre-industrial grain production limited growth and muscle power (human or beast).
Coal was next, and came of age during the Industrial Revolution. Steel, steam engines, railroads, chemicals, and electricity increased the productivity of a human by orders of magnitude. Its weaknesses were transportation cost and the inability to scale down in transportation applications. Today, almost every country with decent coal reserves is at least middle-income, but energy-poor countries often cannot afford the price of imported coal.
The Age of Oil
We undeniably live in the age of oil. Oil became increasingly important in the 20th century, but dominated after World War II. Cars, highways, semi-trucks, airliners, and plastic all grew at astounding rates in places like the United States. On the supply side, American prospectors found absurdly large reserves in the Middle East and elsewhere. Oil's transportation costs are minimal. It is simple to pump and has excellent energy density. Cheap energy could spread everywhere. That portability meant oil was not just transforming transportation and materials but powering basic industrial processes and fueling marginal electricity production.
Market Share of various ammonia feedstocks over time.
Unfortunately, oil's supply hit a wall in the 1970s. Prices post-1973 have been higher and more volatile. Oil remains the universal marginal unit of energy, but it can't meet the demand it would with a better supply. The global economy has reorganized to limit oil use outside of transportation, sapping labor and capital. Oil usage fell in process heat, electricity generation, and petrochemicals in favor of coal and natural gas. Only the most energy-poor places, like Hawaii, regularly burn oil for electricity due to the extreme cost. I find it jarring to see heavy fuel oil as the default energy comparison in 1950s and 1960s engineering reports.
Oil prices aren't the same.
It is also a part of the "What happened in the early 1970s?" meme. Did we collectively give up at once, or did the scarcity of our marginal energy source force us to reorganize around less productive energy sources?
The limits of oil have encouraged the development of alternative resources, though none have taken the crown. It is challenging to beat oil’s energy density, cost of transportation, ability to scale up and down, and flexibility.
The Importance of Complementary Technologies
Humans devise technologies to harness new energy sources for heating, transportation, work, and agriculture in a positive feedback loop. Animal husbandry, plows, and plant domestication are a few classic examples from the age of grain. Coal is of little utility without steam engines and blast furnaces. Humans mostly burned pools of oil in jars before turbines, petrochemicals, and the internal combustion engine brought oil into dominance.
A feature of complementary technologies is that inventors often optimize them for a specific energy source. Coal cannot easily power a car or aircraft. The gap between discovering a new energy source and how to use it most effectively can take decades (or centuries). Typically, the production technology for raw energy improves by serving an early market. Coal had heating, and oil had lamps. Lower raw energy costs and scale allow other markets to become viable.
Solar PV as a Universal Marginal Energy Source
Solar PV has features that could make it the universal marginal source of energy.
Solar Panels are Lightweight
The active material in a solar panel is only a few pounds, and the entire panel is light enough for one or two humans to carry. One modern panel produces the same amount of electricity in 30 years as 10 tons of coal. That is 300x more energy per mass. Even places with terrible infrastructure can cost-effectively import energy by purchasing solar panels.
Source
The mass of silicon per watt has fallen precipitously over time, down to 2 g/watt. A paper analyzing modern panels finds that the energy return on investment would be near 30:1 at a 12% capacity factor. Many of the largest producers hope to shrink wafer thickness again because emerging technologies like TOPCon do not lose efficiency at thinner wafer sizes. The grams per watt could easily halve. Items like aluminum frames will become the largest source of embodied energy.
Solar PV is Modular and Scalable
Energy use cases can range from charging a cordless electric drill up to powering a gigawatt-scale aluminum smelter when paired with the appropriate complementary technologies.
Sunlight is Widely Distributed
The concentration of fossil fuels presents issues because it can be too expensive to transport, and reserves are often in unfavorable places. The sun shines virtually everywhere on Earth, and even a gloomy place like Scotland receives half the sunlight that sunnier places get. Energy access can be universal.
The Need to Move Beyond the Grid
The electric grid is a premium service that is too expensive and inflexible to provide the marginal unit of energy for many applications. Hooking up to the grid adds several hundred dollars per kilowatt in power plant capital costs, can tack on years to the project timeline, and charges ruinous transmission and distribution costs. Wholesale electricity prices are now less than half of what customers pay.
Distribution and transmission are expensive.
Industry and materials are the most cost-challenged by current electricity prices; grid costs are a non-starter, and they already avoid using it. Aluminum smelters and other large electricity consumers typically use "captive" power plants or have subsidized pricing. Transportation is replacing expensive liquid fuels, but truck charging points are likely to incorporate microgrids to reduce demand charges.
There is talk of relying on negative electricity prices for low-cost applications, but trends do not favor this strategy. Users must still pay transmission and distribution fees or build infrastructure to tap into high-voltage power, even if wholesale prices are free. Subsidies and inflexible thermal power plants staying online for operational reasons often drive negative prices. Subsidies will not last forever, and batteries now have enough capacity in markets like California to handle evening ramps, reducing the need for gas power plants to be online during the day.
Solar will need to directly power many applications and processes without the electric grid to evolve.
Solar's Enablers
A pool of oil has little utility for anybody. But, the internal combustion engine, gas turbine, catalytic cracker, and oil-fired boiler powered massive economic growth. Similarly, a solar panel alone is limited in impact without complementary technologies.
It was not clear that solar PV could be inexpensive enough to be a base unit of energy until recently. A thoughtful book, “Taming the Sun,” written in 2017, proposed using sunlight more directly through photocatalysts or solar concentrators. The rapid drop in solar PV costs means technologies can hook into arrays of cheap solar panels instead of developing more complicated and less mature technologies. Storage or variable output is still required to solve intermittency.
There are four primary enabler categories.
Vehicles and Vehicle Batteries
Batteries fill the role of the internal combustion engine in a solar-dominated world. They turn raw electricity from solar power into work to power cars and other vehicles. Batteries have different tradeoffs than internal combustion engines. "Fuel" costs are lower even at grid prices, there is less powertrain maintenance, fewer emissions, and convenient nightly charging. But, energy density is less than that of liquid fuels, even after adjusting for efficiency.
Lower energy density means engineers must design vehicles for batteries, and the economic incentives to do this are strong. A car manufacturer bears the cost of adding efficiency features to gasoline-powered cars, but only the customer pays for fuel. There are rapidly diminishing returns to further efficiency. The math is different with battery-powered vehicles because the battery pack is much more expensive than a fuel tank. Efficiency gains pay off because they reduce pack size and manufacturing cost. The best-selling EVs are much more aerodynamic, shed wasteful mass, have low rolling resistance tires, and have features like heat pumps to maximize efficiency.
Autonomous driving will accelerate the trend to electric vehicles as most fleet operators favor EVs.
Stationary Storage
Stationary storage batteries solve most of the "the sun doesn't shine at night" issues and improve the utilization of transmission and distribution assets. They are ~100x larger than car battery packs and charge/discharge ~5x-10x slower. Pack and cell architectures are evolving to much larger cell sizes than are practical in EVs (which need smaller cells in series to reach practical voltages). Larger cells mean fewer installation steps and connections, along with less non-productive mass. Cells with thicker electrodes could push these improvements even further.
More power electronics are being integrated at the factory rather than in the field, reducing time and labor. Installed pack costs, not including interconnection, should easily fall well below $100/kWh everywhere, and this is already the case in China. The battery component of the levelized cost of storage would fall below $50/MWh at $100/kWh installed if we extrapolate the 2023 Lazard methodology.
Industrial customers might buy electricity from co-located solar and batteries to save money, especially if they can reduce electricity demand on the cloudiest days of the year to limit overbuilding of solar capacity. A titanium processor in West Virginia and a steel micromill in California are already planning to use solar plus storage for most of their needs.
Ten-hour to twenty-hour batteries should also be feasible. A recent Volta report suggests LFP cell prices could soon be $30/kWh in China, even in smaller form factors. And that price does not consider chemistries like LMFP that might trade better energy density and cost for lower cycle life. Many thousands of cycles aren't necessary for these applications. Batteries with more capacity could help solar PV breach 50% market share and possibly 80%-90% in most electricity markets.
Any electricity source can charge vehicles or stationary batteries, but solar is likely to benefit the most, given its cost and generation profile.
Hydrogen/Chemical Feedstocks
Hydrogen, carbon dioxide, and nitrogen can be the low-level feedstocks for most petrochemical production. Hydrogen is the most challenging because it must be unbearably cheap to compete with fossil hydrocarbons. Alkaline electrolyzers appear to be making the most progress in scale and cost so far, but there are a few other possible options.
Heat + Thermal Storage
Thermal storage could provide 20%-35% of global energy use for process heat, district heating, and replacing fuel in thermal power plants. Fossil fuels are thermal storage, so the incentives were not strong before solar PV was cheap.
The case for continuing to use heat in industry is strong. Thermochemical techniques dominate many chemical processes because they have cost benefits from scale and often have much better selectivity than biological or electrochemical processes. Cost-effective solar-to-heat technology is important in maintaining (or expanding) high standards of living.
Electric heating elements and storage materials can be inexpensive, making them much more convenient than redesigning processes that are inflexible. In practice, it is challenging to make such low costs a reality.
I am the CEO of Standard Thermal. Our focus is on delivering thermal storage costs low enough to make months' worth of energy storage possible. I wrote a companion blog post to this one detailing our strategy and thermal storage more generally.
How Cheap Can Solar Get?
Cheap solar PV is critical for these applications to be practical. The most relevant cost metric is direct current (DC) cost since many applications might be fully or partially off-grid. There are several strategies to reduce this cost:
Adapt Solar Farm Design to Low-cost Panels
Technologies like trackers maximize the output of each panel. As costs fall, it becomes cheaper to “lay the panels flat” on much simpler racks. Each panel produces less electricity but also costs less to install. Shading is not a concern if the panels are flat, and density can increase 2x-3x compared to arrays with single-axis tracking. Non-panel DC costs can decrease below $100/kW.
Increase Panel Efficiency
Panel efficiency gains directly impact DC costs because high-wattage panels increase the electricity output for inputs like labor and racking. Today’s mass market panels are ~21% efficient but could rise to ~26% with silicon technology in the pipeline, providing a 20% cost reduction holding other inputs constant. Technology like perovskite tandem cells could push efficiency well into the 30s and further depress both panel and non-panel costs.
Continue Small Improvements and Automation
Larger panel form factors, automation of panel cleaning, simplified installation technology, robots, and other incremental improvements could continue to reduce non-panel costs. Using cheaper substitutes for aluminum, glass, and silver is another promising path.
These improvements could result in arrays that cost $100/kW DC under the most optimistic assumptions. The levelized cost of electricity could be near $5/MWh at an 8% discount rate. An energy cost of $5/MWh at customer sites would be disruptive. Energy costs that low are only available near fossil fuel mines/fields, and then only as heat. Costs like that available almost anywhere would be a gift for most of the world.
Opening the Next Era of Growth
The world has arguably been energy-constrained for fifty years, with most economic growth coming in regions with significant coal reserves. Oil has been expensive and a brake on growth. The oil and gas industry has only grown its reserves with high prices. It cannot maintain price stability like in the pre-1973 era. Advancements like shale gas have so far been regional phenomena rather than global forces due to high transportation costs for LNG. The tyranny of oil supply could ease as the manufacturing capacity for solar PV approaches 1 TW/year, with no real constraints to further supply. Deployments at that scale will quickly saturate electric grids and stall solar growth without the development of complementary technologies. The age of solar PV is still young, and it could power humanity’s next leap in living standards.