Google to Pay SpaceX $920 Million a Month for Compute Capacity

Google Strikes $920 Million Monthly Deal with SpaceX for Compute Capacity

In one of the most striking technology partnerships of 2026, Google has agreed to pay SpaceX a staggering $920 million per month for compute capacity. The deal, which covers the period from October 2026 through June 2029, was revealed in a SpaceX filing submitted to the U.S. Securities and Exchange Commission (SEC) on Friday. The announcement not only underscores the explosive demand for artificial intelligence infrastructure but also highlights SpaceX's rapidly expanding role as a major player in the enterprise cloud and compute market — a development few would have predicted even two years ago.

What the Deal Actually Covers

According to the SEC filing, the compute capacity Google is purchasing from SpaceX includes approximately 110,000 NVIDIA GPUs, CPUs, memory, and other related components. This is not a small-scale cloud agreement; it represents a massive, purpose-built hardware commitment designed to handle the intensive processing requirements of modern AI workloads at enterprise scale.

The contract also contains a notable flexibility clause: starting in 2027, either party may terminate the agreement with just 90 days' notice. This gives both Google and SpaceX a degree of operational freedom while still locking in a significant revenue stream in the near term. For SpaceX, this arrangement provides a highly predictable and lucrative source of income ahead of what is widely anticipated to be a record-setting IPO.

Why Google Needs SpaceX's Compute Power

Google confirmed the partnership in a statement to Business Insider, noting that the deal was necessary to meet surging demand for Gemini Enterprise, its agentic AI platform. As AI workloads grow more complex and more widespread across enterprise environments, even a company with the infrastructure resources of Google finds itself needing to source additional capacity from outside providers.

"Google Cloud and SpaceX are long-time partners," a Google Cloud spokesperson said, framing the agreement as a natural extension of an existing collaborative relationship rather than an entirely new direction. Still, the financial scale of the arrangement — nearly $11 billion per year — signals just how serious Google is about keeping Gemini Enterprise competitive in a market increasingly defined by who has the most compute at their disposal.

Gemini Enterprise is Google's flagship agentic AI offering, designed to automate complex, multi-step tasks across enterprise workflows. Powering such a platform at scale requires enormous computational resources, and the demand from businesses adopting AI agents has clearly outpaced what Google can serve from its own data centers alone.

SpaceX Is Becoming a Serious Compute Powerhouse

The Google deal is not SpaceX's first foray into the lucrative world of AI compute contracts. Just weeks earlier, SpaceX's S-1 IPO filing revealed that Anthropic — the AI safety company behind the Claude family of models — was already paying SpaceX $1.25 billion per month for compute resources. Together, these two contracts alone represent more than $2 billion in monthly recurring revenue from compute services, a business line that most people still primarily associate with rocket launches and satellite internet.

This pivot — or more accurately, this expansion — reflects SpaceX's broader strategy of leveraging its engineering capabilities and hardware procurement relationships to build out data center and compute infrastructure. The company's ability to negotiate access to massive quantities of NVIDIA GPUs at a time when those chips remain extraordinarily difficult to source gives it a powerful competitive advantage in the compute market.

The Broader Context: AI Infrastructure Is the New Space Race

The Google-SpaceX deal is a symptom of a much larger trend reshaping the technology industry. Access to AI compute has become one of the most strategically important resources in the world, with major technology companies, AI labs, governments, and now even aerospace firms all competing to secure GPU capacity. NVIDIA's H100 and successor chips have become the gold standard for AI training and inference, and any organization that can offer tens of thousands of them to paying customers holds enormous leverage.

For Google, paying nearly a billion dollars per month to an outside provider is a remarkable admission that even the world's most powerful cloud infrastructure company cannot always build fast enough to keep up with its own AI ambitions. It also raises important questions about the long-term economics of AI development: if the cost of compute continues to grow at this pace, how will that shape pricing for AI services, and which companies will ultimately be able to afford to compete?

What This Means for SpaceX's IPO

The timing of these compute contract disclosures is clearly not accidental. SpaceX is moving toward what analysts expect to be a landmark initial public offering, and the company has every incentive to demonstrate that its revenue base extends far beyond rocket launches and Starlink subscriptions. Landing marquee clients like Google and Anthropic — and doing so at a scale that generates billions of dollars per month — sends a powerful message to prospective investors about the company's diversification and growth trajectory.

Investors evaluating SpaceX's IPO will now need to assess not just the aerospace business, which remains exceptional, but also an emerging compute and infrastructure business that appears to be scaling with remarkable speed. The combination of reliable government and commercial launch contracts, a global satellite internet constellation, and now a fast-growing AI compute revenue stream makes SpaceX a genuinely unusual investment proposition.

Key Takeaways

• Google has agreed to pay SpaceX $920 million per month for compute capacity running from October 2026 to June 2029, covering approximately 110,000 NVIDIA GPUs and associated hardware.
• The compute power is intended to support Google's Gemini Enterprise agentic AI platform, which is experiencing rapid demand growth across enterprise customers.
• Either party can exit the agreement with 90 days' notice starting in 2027, providing flexibility for both sides as the AI landscape continues to evolve.
• SpaceX's SEC filing also revealed a $1.25 billion monthly compute contract with Anthropic, meaning the company now earns over $2 billion per month from AI infrastructure services alone.
• These deals significantly bolster SpaceX's financial profile ahead of its widely anticipated IPO, positioning the company as a diversified technology and infrastructure business rather than purely an aerospace firm.

The Bottom Line

The Google-SpaceX compute deal is more than a business transaction — it is a clear signal about where value is being created in the modern technology economy. Compute is infrastructure, and infrastructure is power. As AI continues to reshape industries and drive insatiable demand for processing capacity, the organizations that control large-scale GPU resources will increasingly dictate the pace and direction of technological progress. SpaceX, once known almost exclusively for its rockets, has quietly positioned itself at the center of that competition, and the numbers now speak for themselves.