Mastery over energy

A Paper by Jason Crawford

The bold, ambitious future is going to require a tremendous amount of energy.

Most people have been taught to believe that energy usage is at best a necessary evil—or perhaps that it’s not even all that necessary. Journalist Cleo Abrams reports that when she first heard a compelling case for energy abundance, it “put my brain through the BLENDER,” because she had always seen energy usage as “shameful.”³⁹ Calling attention to a technology’s energy use is a quick and easy way to criticize it, no further explanation needed: energy is bad, therefore jet travel or AI is bad. But from a techno-humanist point of view, this logic is backwards: planes, AI, and the rest of the industrial economy are good, and therefore energy, the most fundamental enabler of the economy, is tremendously good.

Energy usage per capita, however, flatlined in the US and Europe in the early 1970s.⁴⁰ While we have squeezed out some efficiency gains, creating marginally more value with the same amount of energy,⁴¹ stagnation in energy is overall a sign of stagnation in economic growth. We would be better off with more energy. J. Storrs Hall argues that lack of progress in energy density explains which technologies dreamed of in the 1960s actually arrived (such as the Internet) and which didn’t (such as the space economy).⁴²

He suggests, in short, that energy density is why we got progress in bits but not in atoms.

With sufficient energy, we could solve all of our water problems, making drought a thing of the past. Water is, after all, enormously abundant in the oceans; the problem with ocean water is that it’s salty and it’s below sea level—we want it to be fresh and elevated, so it can flow through our pipes to our faucets. Normally we rely on the natural water cycle to do this: an inefficient, unreliable use of solar energy. With enough industrial energy, we can control the process ourselves, desalinating water and pumping it to any elevation.⁴³

With sufficient energy, we could solve almost all other material resource problems, too. The materials are out there; the challenge is always to gather them from dilute sources and to isolate and purify them—whether extracting metals from ore, bromine from salt water, or gasoline from crude oil. Seawater contains gigatons of metals such as lithium, nickel, copper, vanadium, molybdenum, and uranium.⁴⁴ With sufficient energy, we could extract elements from common clays instead of high-grade ores, or even from landfill.⁴⁵ There never need be a shortage of any resource.

Energy not only extracts material resources, it also shapes them. “A long-running story of our material world,” writes Ian MacKay, “is the constant march to more embodied energy in everything we use”: as we advanced from wood and copper, to bronze, to cast iron, to steel, to titanium, to semiconductors, the energy used per kilogram to process our materials has increased by several orders of magnitude.⁴⁶ By using more energy, we’re able to make lighter and tougher materials and to tap more abundant resources. It stands to reason that the materials of the future will continue this trend.

Credit: Ian McKay, “The future is made of energy”

We need more energy for supersonic flight, which uses around 3x more fuel per passenger-mile than conventional subsonic jets.⁴⁷ Cargo could also travel faster if we could afford to spend more energy on it.⁴⁸

We need more energy for the rapid growth of AI. Although each query to an LLM uses only a small amount of energy (maybe a few watt-hours depending on the length), total AI energy usage is projected to consume on the order of 10% of US electricity by 2030.⁴⁹ If the US were to reach the 100-to-1 ratio of AI to humans mentioned earlier, using an NVIDIA H100 chip as a stand-in for one human’s worth of compute, those chips alone would draw around 12 TW—more than half the entire world economy today.⁵⁰

We need more energy for all the robots, too. By one estimate, there could be more than a billion humanoid robots in use by 2050, which would consume about 10% of current world electricity production.⁵¹

The need for more and more energy is ultimately the best reason to transition off of fossil fuels in the 21st century: there just aren’t enough of them. There are an estimated 13,600 TW-years of coal, oil and gas remaining in the ground; if we were to 10x world energy usage, that would only represent 68 years of energy.⁵² Only solar, nuclear, and geothermal have a hope of powering the superabundant energy future