Google just made a huge bet on SpaceX’s AI computing power. Starting October 2026, Google will pay SpaceX $920 million every month. This deal runs through June 2029. Over the full term, Google will spend close to $30 billion for access to powerful AI hardware.

The contract gives Google access to about 110,000 Nvidia GPUs, along with CPUs, memory, and other key components. These GPUs are needed to train and run Google’s Gemini AI models, which power its AI subscription called Gemini Enterprise. Demand for this service has surged beyond Google’s existing data centers.

Why would Google pay SpaceX for extra computing capacity? Google already owns a massive global cloud network. The answer is simple: this deal acts as a bridge. It covers the sudden spike in AI workloads that Google’s own centers can’t handle yet.

The deal includes a ramp-up phase through September 2026. During this time, SpaceX will gradually provide access to the GPUs at a reduced fee. If SpaceX fails to deliver the full hardware by the end of September, Google can cancel or accept fewer GPUs with a lower monthly bill. After December 2026, either side can end the contract with 90 days’ notice.

SpaceX’s Growing Role in AI Cloud Computing

SpaceX isn’t just a rocket company anymore. Its data centers, originally built for Elon Musk’s AI startup xAI, are now a hot commodity. xAI’s Colossus 1 data center in Memphis, Tennessee, houses over 220,000 Nvidia GPUs. It’s a massive facility designed for heavy AI workloads.

Google isn’t the only big player renting from SpaceX. Anthropic, another AI company, has a separate deal to use Colossus 1’s full computing power. Anthropic pays SpaceX $1.25 billion every month.

Between Google and Anthropic, SpaceX expects to pull in nearly $2.2 billion monthly in AI compute contracts. That’s a massive revenue stream just before its planned IPO on June 12, 2026. SpaceX aims to raise $75 billion by offering shares on the stock market, hoping to become the biggest IPO ever.

Why This Matters for AI and SpaceX

This deal shows how much AI depends on hardware. AI models like Gemini need huge numbers of GPUs to train and run smoothly. These chips are expensive and in short supply. SpaceX’s facilities offer a rare resource in this tight market.

For Google, locking in this capacity reduces risk. It keeps their AI projects moving even as demand grows fast. It also diversifies where Google gets its computing power, which is smart. Relying on just one cloud infrastructure can be risky in such a competitive field.

For SpaceX, this deal validates its data center strategy. It’s no longer just about rockets and satellites. SpaceX is positioning itself as a key player in AI infrastructure. The money from these contracts can help fund its starship programs and plans to expand satellite internet with Starlink.

There’s also a futuristic angle here. Google is exploring “off-planet” AI processing using satellite networks. That’s part of an internal plan called Project Suncatcher. SpaceX’s control over space tech and cloud computing could make this a reality.

Google keeps full ownership of its data and AI models on SpaceX’s systems. The contract ensures SpaceX cannot use Google’s AI training data to improve its own AI ventures. This separation protects Google’s competitive edge.

Overall, this deal signals a shift. SpaceX is leveraging its space and tech assets beyond launches. It’s becoming a heavyweight in cloud computing for AI. Meanwhile, Google is securing the hardware muscle it needs to stay competitive in the AI race.