SpaceX & the Sentient Sun

Elon Musk's SpaceX pay package reads like a fever dream: no salary bump unless the company hits a $7.5 trillion valuation with a million-person Mars colony, or builds space-based data centers drawing 100 terawatts, more than a thousand times what every data center on Earth currently uses combined. Miss both, and Musk keeps taking home the $54,080 he's been paid since 2019. The board signed off on this because Musk's track record on direction, if not always timing, has been eerily consistent. He said a private company could fly astronauts to orbit; SpaceX now does it routinely for NASA. He said reusable orbital rockets were possible when the industry treated boosters as trash; Falcon 9 has landed and reflown hundreds of times. He said satellite internet could be a real business after decades of bankruptcies; Starlink pulled in $11.4 billion in 2025.

The influence isn't hidden, either. SpaceX's drone ships, the ocean barges that catch returning boosters, are named after sentient AI starships from Iain M. Banks's Culture novels: Of Course I Still Love You, Just Read the Instructions, A Shortfall of Gravitas. At the 2023 UK AI Safety Summit, Musk called the Culture books the closest thing anyone's written to a genuinely utopian AI future. He's been telegraphing the blueprint on the sides of his landing pads for years.

Work backward from that blueprint and the current SpaceX roadmap starts making sense as a supply chain rather than a scattershot of projects. A self-sustaining Mars city needs thousands of Starship flights, each Earth-Mars transfer window only a few weeks wide and opening once every 26 months. A lunar base comes first as rehearsal, with Musk floating factories on the Moon that use mass drivers, electromagnetic slingshots exploiting one-sixth gravity, to fling satellites built from lunar regolith (roughly 20% silicon, 10% aluminum) straight into deep space. Then come the orbital data centers themselves: solar panels in space generate four to ten times more power than the same panels on Earth because there's no atmosphere, no clouds, no night. Musk claims that within five years SpaceX will launch more AI compute annually than the entire installed base currently sitting on Earth, which is presumably why SpaceX merged with xAI back in February.```

None of it works without Starship driving launch costs down toward $100-500 per kilogram, a staggering drop from the roughly $18,500 NASA historically paid. And none of it gets funded without Starlink, which now serves 10.3 million subscribers across 164 countries and runs at north of 60% EBITDA margins, or without Falcon 9, which alone launched 83% of all mass sent to orbit from Earth in 2025.

The origin story undercuts the mythology a little, in a good way. Musk didn't set out to build a rocket empire; he wanted to send a greenhouse to Mars to reignite public interest in space. When Russian arms dealers refused to sell him a cheap ICBM (one spat on him during negotiations), he pulled out a spreadsheet on the flight home showing that raw materials were only 2% of a rocket's quoted price, and decided to just build the thing himself. That spreadsheet logic became SpaceX's operating principle, later nicknamed the 'idiot index': if a part costs way more than its raw materials, you're either an idiot or working with idiots. An early engineer once built a $120,000 actuator for $3,900 following that exact rule.