SpaceX has reportedly signed a multi-year agreement with Google worth approximately $920 million per month to provide AI compute capacity, marking one of the largest AI infrastructure deals announced to date. The agreement is expected to begin in October 2026 and run through June 2029, covering access to roughly 110,000 NVIDIA GPUs alongside the CPUs, memory, networking, and supporting infrastructure required for large-scale AI workloads.
According to reports, the deal will be rolled out in phases, with Google paying reduced fees initially as SpaceX gradually brings additional server capacity online through September 2027.
As per reports, this agreement, SpaceX is expected to build out AI infrastructure capable of supporting approximately 110,000 NVIDIA GPUs. Rather than delivering the full capacity immediately, the company will deploy additional racks over time, allowing Google to scale its AI workloads as infrastructure becomes available.
Reports suggest that Google could seek reductions in fees or potentially terminate portions of the agreement if SpaceX fails to meet deployment milestones by late 2027. Both companies are also said to retain options to exit the arrangement after December 2027 under specific conditions. The deal highlights the growing importance of dedicated AI compute infrastructure as technology companies race to secure GPU capacity for training and serving increasingly complex AI models.
Follows Other Musk-Linked AI Infrastructure Deals
The Google agreement follows another major AI infrastructure commitment reportedly secured by SpaceX earlier this year through Anthropic.
Anthropic is said to be using computing capacity from Elon Musk's xAI Colossus supercomputer infrastructure for AI workloads. Colossus gained industry attention for the speed at which it was deployed, although reports indicate the facility contains a mix of NVIDIA H100, H200, and GB200 GPUs.
AI Compute Could Become a New Growth Area for SpaceX
Historically, SpaceX has been associated with rocket launches, satellite deployment, and Starlink broadband services. However, the scale of its recent AI compute agreements suggests the company is rapidly expanding into infrastructure services for artificial intelligence.
Industry estimates indicate that the combined value of the Google and Anthropic agreements could generate more than $25 billion annually, potentially making AI infrastructure one of SpaceX's largest business segments if deployment targets are achieved. The development reflects a broader industry trend where access to compute power is becoming as strategically important as access to cloud services or network infrastructure.
AI Compute Capacity Becoming the Industry's New Battleground
The reported Google-SpaceX agreement underscores how demand for AI infrastructure continues to reshape the technology industry. While previous cloud expansion cycles focused on storage and networking, the current AI race is increasingly centred around securing access to GPUs, power, cooling, and large-scale data centre capacity.
For companies developing foundation models and agentic AI systems, access to compute has become a competitive advantage in its own right. As a result, infrastructure providers capable of deploying GPU clusters at scale are emerging as some of the most important players in the AI ecosystem. The deal also illustrates how technology companies are diversifying their infrastructure strategies. Rather than relying solely on internal data centres, firms are increasingly partnering with specialised compute providers to meet rapidly growing AI demand.
If fully realised, the Google agreement would position SpaceX as a significant new player in large-scale AI infrastructure, expanding its role far beyond aerospace and satellite connectivity into one of the fastest-growing segments of the technology industry.
