In a move that signals a seismic shift in the artificial intelligence arms race, Alphabet has secured a massive $30 billion computing agreement with SpaceX. Under the terms of the deal, Google will pay approximately $920 million monthly starting in October 2026 through June 2029. This aggressive procurement of processing power underscores the escalating desperation among tech giants to secure the physical infrastructure necessary to run the next generation of AI models.
The agreement provides Google with access to 110,000 Nvidia GPU chips, including the high-specification H200 series, alongside necessary CPU and memory components. This vast array of hardware represents an estimated 100 megawatts of power—enough to supply 75,000 homes simultaneously. For Alphabet, this is a tactical bridge intended to stabilize its 'Gemini Enterprise' agent platform, which has seen customer demand surge well beyond the capacity of Google’s internal cloud infrastructure.
SpaceX is utilizing its AI subsidiary, xAI, to execute this pivot into the compute-rental business. By leveraging its data center hubs in Memphis and Mississippi, the company is positioning itself as a critical infrastructure provider rather than just a model developer. This strategy was a focal point of SpaceX’s recent IPO roadshow, framing the company as the backbone of the AI era, following similar high-profile compute-supply deals with startups like Anthropic and Cursor.
The relationship between the two entities remains uniquely complex, characterized by a 'frenemy' dynamic of deep integration and fierce competition. Alphabet currently holds an estimated 5% stake in SpaceX, a relationship that predates their current rivalry in AI development. Beyond terrestrial compute, the firms are reportedly discussing the launch of orbital data centers, a project internally known at Google as 'Project Suncatcher,' which could further entwine the destinies of these two tech behemoths.