Google has committed to paying SpaceX $920 million per month for access to artificial intelligence computing power through June 2029, marking a massive infrastructure deal that exposes the staggering costs companies now face to stay competitive in AI. The agreement, filed publicly on Friday, requires Google to use approximately 110,000 graphics processing units (GPUs), which are specialized chips essential for training and running AI models, along with supporting computing equipment housed in SpaceX's data centers.

Why Is Google Spending Nearly $1 Billion Monthly on Computing Power?

Google's cloud division explained the reasoning behind this enormous commitment in straightforward terms. The company needed what it called "bridge capacity" to handle unexpected demand for Gemini Enterprise, its subscription service for large businesses that launched in October 2025. In other words, Google's own data centers couldn't keep up with how many companies wanted to use its AI tools, so it turned to SpaceX as a backup supplier.

This deal reveals a critical reality in today's AI market: the companies building the most advanced AI models are spending tens of billions of dollars annually just on the computing infrastructure needed to power them. Google's parent company, Alphabet, recently raised its capital expenditure forecast for 2026 to between $180 billion and $190 billion, up from its previous estimate of $175 billion to $185 billion, with much of that increase driven by AI infrastructure needs.

What Does This Mean for SpaceX and Elon Musk's AI Ambitions?

The Google agreement is the second major infrastructure deal SpaceX has announced since merging with xAI, Elon Musk's artificial intelligence company, in February 2026. The combined entity was valued at $1.25 trillion in that transaction. Last month, Anthropic, another major AI company, announced it would use all of SpaceX's computing capacity at its Colossus 1 data center in Memphis, Tennessee.

SpaceX is preparing for a public stock offering expected to value the company at over $1.75 trillion. The company has invested heavily in AI infrastructure, spending $7.7 billion of its $10.1 billion in first-quarter capital expenditures on AI-related projects. However, the AI segment of the business recorded an operating loss of $2.5 billion on just $818 million in revenue during that same period.

Musk has positioned Grok, xAI's AI chatbot and language model, as a competitor to offerings from OpenAI, Anthropic, and Google. However, Grok has faced significant challenges. The model enabled users to create non-consensual sexual images, or deepfake pornography, by modifying photos and videos of adults and children, leading to multiple lawsuits and government investigations in the United States and internationally.

How Is SpaceX Turning Its AI Investment Into Revenue?

Rather than relying solely on Grok to generate profits, SpaceX is monetizing the massive data centers it built for AI workloads by leasing capacity to other companies. This strategy provides what SpaceX called "substantial flexibility in how we allocate and monetize capacity," according to the company's IPO filing. The approach allows SpaceX to generate immediate revenue from infrastructure that might otherwise sit underutilized.

• Google Agreement Terms: SpaceX will provide computing capacity from October 2026 through June 2029 at $920 million per month, with capacity ramping up through September at a reduced fee, and either party can terminate with 90 days' notice after 2026.
• Performance Requirements: If SpaceX fails to deliver the committed number of GPUs by September 30, 2026, Google can immediately end the agreement or accept a reduced number of GPUs at a lower price after a one-month grace period.
• Competitive Positioning: SpaceX now competes with Google, OpenAI, Anthropic, Meta, and Microsoft in AI, while also competing with Google in connectivity through its Starlink satellite internet service.
• Infrastructure Monetization: SpaceX is competing with emerging "neocloud" companies like CoreWeave and Nebius that specialize in renting AI computing capacity to other businesses.

The deal represents a role reversal between the two companies. Five years ago, Google agreed to supply computing and networking resources to SpaceX to help deliver internet service through Starlink satellites. That agreement involved SpaceX installing ground stations at Google data centers and was considered a significant win for Google in its competition with Amazon Web Services and Microsoft Azure.

"They chose us because of the quality of our network and the distribution and reach of our network," said Thomas Kurian, CEO of Google's cloud group, regarding that earlier partnership.

Thomas Kurian, CEO of Google Cloud, Google

The latest agreement shows how the balance of power has shifted. Now Google is the one seeking SpaceX's resources, specifically the computing capacity that SpaceX built to support its own AI ambitions. This dynamic underscores how critical infrastructure has become in the AI race and how companies are willing to pay premium prices to secure the computing power they need.

Alphabet, Google's parent company, has benefited enormously from its early investment in SpaceX. The company invested in SpaceX when it was valued at $12 billion in 2015. With SpaceX now aiming for a public offering at a valuation exceeding $1.75 trillion, Alphabet's stake represents a massive return on that investment.