Space compute and the price of permission: Terraform's Casey Handmer
Key points
Key takeaways from Relentless interview with Terraform Industries founder Casey Handmer (September 2026):
Lunar mass driver economics. Handmer, who spent years on Hyperloop-style magnetically accelerated levitation, argues the Moon's lower gravity and airlessness let you launch mass into orbit without a rocket: a sub-1,000 ft electromagnetic shuttle firing a few kilograms roughly once a second could move 100 tonnes a day. It only makes sense, he is blunt, if the orbital AI constellation demands more mass than cheap Earth launches can supply — Starship should push launch costs to a few hundred dollars per kilogram.
Space compute within three years. He backs Elon Musk's claim on the Dwarkesh and John Collison podcast that in three years it will be cheaper to launch space compute than build it on the ground: Starship enables gigawatt-scale orbital deployment, while terrestrial mega data centres face cost headwinds and jurisdictions now in a price-discovery phase over their cut.
Terraform's solar-plus-AI co-location. Terraform's 1.8 MW solar array, built earlier this year, runs its synthetic-fuel plant by day and a continuous, smaller load — an AI compute or data-centre rack backed by batteries — at the same time without the two users conflicting, more than doubling the output of fixed solar infrastructure, the dominant land cost.
Solar, batteries and bureaucracy. Over the next decade almost all AI compute converges on solar plus batteries on Earth, or a Starlink-style satellite with a radio and GPU in orbit; the deciding variable is whether batteries are cheaper than launch. On the ground the remaining inputs are land — Nevada alone has ~65 million acres of it, much still federal — and permission.
Large projects are becoming illegal. Fifty years of regulation, Handmer argues, have made big infrastructure effectively impossible: approval processes are so slow and costly that the people needed to run them lose interest. He cites Grok's value-of-statistical-life maths putting the economic loss of California's failure to repair its subsided aqueduct (design capacity ~4.5 million acre-feet a year, actual ~50%) at roughly 10,000 lives.
Speed is the progressive strategy. Henry Kaiser — Hoover Dam, Liberty ships — built his Lake Tahoe estate in 22 days before an injunction could land, and WWII shipbuilding cut labour hours 10x while lifting output 50x. Moving quickly cuts labour cost and, Handmer says, benefits poorer households most, since the wealthy can always buy around failures.
Terraform outgrowing Burbank. Even in heavily industrially zoned Burbank, restrictions block expansion: neighbouring blocks were redeveloped as light industrial, and metal refining is effectively illegal across California — no PCB traces, only grandfathered legacy plants — so Terraform must move miles away to a larger industrial area despite the city's goodwill.
Terafab and the national-security lane. On Elon Musk's plan to build the world's largest building in Texas within five years, Handmer notes such a project would get stuck in the normal lane anywhere else; Musk stays on the critical path for chip onshoring — the "do you want chips or do you want to lose the next war to China" call — to win it and cut corners.
Read more: Relentless
