SpaceX plans to expand its compute capacity more than fivefold by end of 2027 to approach ten gigawatts, requiring over two million Nvidia Rubin GPUs if the entire buildout uses that platform.
The buildout follows xAI's February 2026 merger with SpaceX (a deal that valued SpaceX at $1 trillion and xAI at $250 billion) and reflects rapid growth in SpaceX's AI segment, which leases Colossus compute capacity to cloud customers—including a single unnamed customer accounting for roughly $1.52 billion in Q2 2026 revenue—though the segment remains unprofitable despite tripling revenue quarter-over-quarter.
What happened
SpaceX aims to grow its compute capacity more than fivefold by the end of 2027, reaching above two gigawatts of power (approaching ten gigawatts by Elon Musk's latest guidance), up from its current 1.4 gigawatts. The expansion will run on Nvidia's upcoming Vera Rubin platform; if the entire buildout uses Rubin, the GPU count could exceed two million units.
Why it matters
SpaceX's AI segment, which includes xAI after its February 2026 merger, generated $2.56 billion in revenue during Q2 2026 but posted an operating loss of $1.26 billion. A single unnamed AI customer (likely Anthropic) accounted for roughly $1.52 billion of that revenue from cloud contracts leasing Colossus compute capacity. The massive GPU acquisition signals SpaceX is betting heavily on becoming a major compute provider, competing against rivals like Anthropic and OpenAI that diversify across AWS, Google, and AMD infrastructure.
What to watch
SpaceX has not disclosed how much of its planned expansion will actually be Vera Rubin GPUs versus other Nvidia chips. Current filings show approximately 540,000 total GPUs in the existing and planned Colossus buildout (roughly 100,000 H100, 110,000 GB200, 110,000 GB300, plus at least 220,000 more GB300 units). Musk has previously stated long-term ambitions to run data centers in orbit.
SpaceX announced ambitious compute expansion targets during its Q2 2026 earnings call, with Elon Musk stating the company wants to grow its capacity more than fivefold by the end of 2027. Currently at 1.4 gigawatts, SpaceX intends to finish 2027 above two gigawatts of power, though Musk suggested capacity could climb closer to ten gigawatts than five by year-end. The entire buildout will run on Nvidia chips, centered on the upcoming Vera Rubin platform, which Musk called the best architecture available. The company's existing Colossus clusters already run on Nvidia's H100, GB200, and GB300 systems.
SpaceX's IPO filings detail the current and planned Colossus buildout at roughly 100,000 H100, 110,000 GB200, and 110,000 GB300 processors, plus at least 220,000 more planned GB300 units—totaling about 540,000 GPUs. To hit the new capacity targets, the company would need well over one million Rubin GPUs; if the entire expansion used Vera Rubin, the number could exceed two million. SpaceX has not yet disclosed how much of the buildout will actually be Rubin.
This expansion strategy accelerated after xAI merged into SpaceX in February 2026 through a stock-funded deal that valued SpaceX at $1 trillion and xAI at $250 billion, for a combined $1.25 trillion. SpaceX's AI segment, which now includes both xAI and X, posted $2.56 billion in revenue during Q2 2026 alongside an operating loss of $1.26 billion—a significant improvement in revenue from Q1's $818 million but a sharp increase in losses from Q1's $2.47 billion operating loss. Most of the revenue growth came not from Grok but from new cloud contracts leasing out Colossus compute capacity. A single unnamed AI customer accounted for roughly $1.52 billion of that revenue; internal analysis suggests this customer is likely Anthropic. By contrast, competitors spread their compute across multiple providers: Anthropic uses AWS Trainium, Nvidia GPUs, and Google TPUs to train and run Claude, backed by up to five gigawatts of additional AWS infrastructure and about 3.5 gigawatts of extra TPU capacity, while OpenAI has announced six gigawatts of AMD capacity, at least ten gigawatts of Nvidia capacity, and around two gigawatts of AWS Trainium capacity. Longer term, Musk has expressed interest in running data centers in orbit.
SpaceX's compute expansion plan marks a sharp pivot toward becoming a large-scale AI infrastructure provider. The February 2026 merger of xAI into SpaceX (valued at $250 billion for xAI and $1 trillion for SpaceX) created a unified entity positioned to leverage SpaceX's capital and manufacturing capabilities for accelerated GPU procurement. The Q2 2026 results show this strategy is already generating revenue: the AI segment's $2.56 billion in revenue—driven largely by a single major customer's cloud compute leases—demonstrates immediate demand for Colossus capacity, even though the segment remains deeply unprofitable at $1.26 billion in operating losses.
The scale of SpaceX's ambition contrasts sharply with how rivals structure their compute. Anthropic diversifies across AWS Trainium, Nvidia GPUs, and Google TPUs (backing up to five gigawatts of AWS infrastructure and about 3.5 gigawatts of extra TPU capacity), while OpenAI splits between AMD (six gigawatts), Nvidia (at least ten gigawatts), and AWS Trainium (around two gigawatts). SpaceX's single-vendor commitment to Nvidia—specifically Vera Rubin for future expansion—creates both opportunity and risk: tighter integration and performance optimization are possible, but dependency on one supplier's roadmap and availability may constrain flexibility. The article does not specify pricing or terms of the compute contracts, so the margin profile of SpaceX's cloud business remains unclear.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Silicon Motion announced a private placement of US$1 billion in aggregate principal amount of 0% convertible s…
Sundar Pichai, CEO of Alphabet and Google, announced on August 11 that the Gemini app's monthly active users (…
A new platform called frontier.fast has launched an open competition where anyone can submit code patches to m…
An AI system generated a research draft that strengthened a mathematical bound related to the Riemann hypothes…
Researchers published findings showing AMIE (Video), a Gemini-based AI system, performed at or above the level…
After former lead writer Stella Sacco posted on Bluesky that Saber replaced her with ChatGPT midway through de…
The AI news that matters, in one minute each morning.
Sign up free