Casey Handmer on Energy, Solar, and the Future of Civilization
On this episode, Casey Handmer (Caltech PhD, former NASA JPL, founder & CEO of Terraform Industries) walks me through how we can pull it off, and why he thinks a major part of this energy singularity will be powered by solar. His views are contrarian, but he came armed to defend them.
These are a few sections that stood out to me:
The fundamental energy transformation process
All conventional power generation follows the same path: find something in the earth that’s out of chemical equilibrium with the atmosphere, burn it to make heat, boil water, run it through a mechanical contrivance to create motion, twist a magnet to generate an electrical field, push electrons through wires, and ultimately approximate thinking through computational gates.
The key bottleneck is converting heat into electricity efficiently. Whether nuclear, gas, combined cycle, or coal plants, they all use the Brayton cycle – the same thermodynamic process that powers jet engines. Any Brayton cycle with Inconel spinning at high speed costs serious money.
Why China’s 20x solar manufacturing advantage isn’t determinative
China faces fundamental vulnerabilities: surrounded by hostile neighbors, dependent on Middle East oil via undefended shipping routes. The US has superior automation, financial capacity, and business environment. Solar manufacturing could be rapidly scaled domestically within 2-3 years if treated as a national priority.
Solar’s relentless learning curve
Solar has a 43% Wright’s Law coefficient – every time cumulative production doubles, costs drop 43%. Not only are adoption, production, and price decreases continuing, they’re accelerating. And the rate of acceleration itself is accelerating.
Why hyperscalers currently choose gas
Two reasons: they need immediate deployment (everything before 2030 is already spoken for), and they’re power availability sensitive rather than power cost sensitive.
The scale of solar infrastructure needed
For 5 gigawatts of AI capacity: 50,000 acres. This sounds massive but is comparable to Manhattan Project sites (Oak Ridge: 60,000 acres, Hanford: 100,000 acres). Technical specs: 10 acres per megawatt at four nines uptime, including solar overbuilding for reliability.
Batteries as grid replacement
Batteries do the same job as the grid – the grid transports power spatially, batteries transport power temporally. The grid suffers from Baumol cost disease while battery utilization runs much higher, making distributed battery systems more economical.
Environmental regulations as the binding constraint
Environmental regulations actively prevent renewable deployment in the US. This is why Texas is winning – out deploying California 10 to 1. NEPA reviews take 4+ years and generate more environmental impact than the solar projects themselves.
GDP becomes misleading in an AI economy
GDP will see massive deflation because the variable cost of running AI will be cheap compared to paying human wages. In the long run, it makes more sense to measure civilization’s size by raw energy use rather than GDP.
The ultimate vision: silicon consciousness
One human brain can be simulated in roughly a square meter of silicon floating in space. Integrated solar-computation wafers can fly as solar sails, adjust orientation with LCD panels, and operate indefinitely in space.
We’re seeing the beginning stages of a collapse back towards the simplest possible thermodynamic-to-cognition stack. The most efficient way to convert energy into usable cognition is silicon.
Timeline predictions
- 2027: Majority of new data centers breaking ground will be mostly solar-powered
- 2030s: Multiple 5-10 gigawatt AI facilities operational
- Long-term: Transition from current complex industrial economy to simplified silicon-based energy-to-cognition conversion
The overarching thesis: we’re approaching an “energy singularity” where solar becomes so cheap and abundant that it fundamentally reshapes both AI development and human civilization itself.