Casey Handmer is the founder and CEO of Terraform Industries and a writer on energy, engineering and industrial scale. His work is relevant to AI through a concrete question: how to supply the electricity and physical infrastructure needed by large computing systems.[1][2]
Start with China is killing the US on energy. Does that mean they’ll win AGI? — Casey Handmer →
X 
GitHub Background
Handmer did his doctorate in Caltech’s theoretical-astrophysics group from 2010 to 2015, studying computational general relativity. Earlier research at the University of Sydney examined optical physics. His subsequent career included Hyperloop and NASA’s Jet Propulsion Laboratory before founding Terraform Industries.[1][2][5]
Contributions
Extract gravitational waves from numerical simulations
In work with Béla Szilágyi and Jeffrey Winicour, Handmer developed a method for extracting gravitational radiation from numerical simulations. The method propagates inner-boundary data outward, transforms the metric into an asymptotic inertial frame and treats the associated radiation fluxes uniformly. The computations were calibrated using the Spectral Einstein Code, connecting his later engineering writing with earlier work on numerical methods.[6]
Connect industrial production to abundant energy
Terraform Industries develops synthetic-fuel production using electricity. Handmer’s public work connects that industrial project with a broader argument about expanding energy supply. His AI discussion is chiefly about the physical conditions for scaling compute, rather than the design or training of models.[1][3]
Ideas
In his Dwarkesh interview, Handmer compares natural gas, solar power, storage and large data-center sites. He argues that learning curves and deployment speed matter alongside the cost of an individual generator. The discussion exposes assumptions about land, batteries, regulation and demand rather than reducing the problem to a single electricity price.[3]
Tools, writing & talks
Writing, research & talks
Start here
Dwarkesh Patel interviews Handmer about solar, storage and the power demands of AI infrastructure.
Watch the comparison of natural gas and solar at 8:28, then the data-center design discussion at 27:02. Track the assumptions about storage, land and deployment time; the linked transcript makes it easier to inspect the argument.
Interview guest[4]Original sourceResearch on extracting gravitational waveforms and radiation fluxes from computational simulations.
Co-author with Béla Szilágyi and Jeffrey Winicour[6]