SpaceX 2 GW to 10 GW AI Compute Build-Out Raises $1T Revenue Stakes

SpaceX is no longer just a rocket company—its valuation and momentum increasingly reflect a build-out of AI compute capacity that could rival cloud hyperscalers. For the space and defense sector, the key metric isn't launch cadence but megawatts: Musk told investors the company will exit 2026 at 2 gigawatts and end 2027 'closer to 10 gigawatts than 5 gigawatts.' That 5x expansion will test SpaceX's manufacturing, power, and data center deployment advantages in a market where speed to capacity equals revenue share.

SpaceX's market debut earlier this year has turned into one of the most volatile and consequential narratives in tech and finance. Priced at $135 in its IPO, shares spiked to $225 within days and then shed more than half their value in subsequent weeks, settling around $150 by mid-September 2026. That kind of swing is unusual even for a high-profile growth stock, and it reflects a market that has not yet decided whether SpaceX should be valued as an aerospace and launch company, a satellite broadband operator, or—most aggressively—as a hyperscale AI infrastructure provider. In his second-quarter earnings call, Elon Musk told investors the company updated its revenue timeline and now expects to reach $1 trillion in annual revenue by 2030, up from just over $20 billion in the trailing four quarters. That target, if achieved, would be one of the fastest large-company revenue expansions in corporate history.

“

Priced at $135 in its IPO, shares spiked to $225 within days and then shed more than half their value in subsequent weeks, settling around $150 by mid-September 2026.

The mechanism for that projected ramp is not Starship or Falcon manifest growth alone; it is SpaceX's AI compute segment. Management said the company is renting out compute capacity and has signed deals with Anthropic, Alphabet, Reflection AI, and an additional undisclosed customer. The company expects to end 2026 with 2 gigawatts of compute capacity and to exit 2027 with 'closer to 10 gigawatts than 5 gigawatts,' according to Musk. For context, a single gigawatt of data-center IT load is an enormous amount of power, comparable to the consumption of a mid-sized city or a major cloud availability zone. A move to 10 GW in roughly 12 to 18 months would represent one of the fastest compute build-outs in the industry, placing SpaceX in direct competition with established hyperscalers and AI infrastructure specialists. This capacity expansion also explains why the stock trades at a price-to-sales ratio close to 100—a multiple that implies a market capitalization around $2 trillion, based on the trailing revenue base. Investors are not paying for what SpaceX earns today, but for the optionality on AI compute demand over the rest of the decade.