Anthropic Signs $1.25B-per-Month Colossus Compute Deal
Business Insider reports that Anthropic agreed to pay $1.25 billion a month to SpaceX, in a contract that extends through May 2029, according to SpaceX's S-1 filing. Business Insider says the agreement gives Anthropic access to SpaceX's Colossus and Colossus 2 data centers and includes a discounted rate for May and June, per the filing. Business Insider reports an Anthropic spokesperson confirmed the $1.25 billion per month figure to the outlet. The S-1 filing, as reported by Business Insider, states either side may terminate with 90 days' notice and projects more than $40 billion in revenue for SpaceX if the deal runs its full term. Editorial analysis: For infrastructure and ops teams, this underscores how hyperscale labs are securing multi-year, high-dollar compute capacity and the downstream implications for procurement and portability.
What happened
Business Insider reports that Anthropic agreed to pay $1.25 billion a month to SpaceX, under a contract disclosed in SpaceX's S-1 filing that runs through May 2029. Business Insider says the S-1 states the deal grants Anthropic access to SpaceX's Colossus and Colossus 2 data centers and includes a discounted rate for May and June. Business Insider reports an Anthropic spokesperson confirmed the $1.25 billion per month figure to the outlet. According to the S-1 filing cited by Business Insider, either party may terminate the agreement with 90 days' notice, and SpaceX projects more than $40 billion in revenue from Anthropic over the contract term if it stands. Business Insider also reports that Anthropic compute chief Tom Brown wrote earlier in May that the Colossus compute would be used for inference, and that SpaceX's prospectus states it expects to sign similar deals in the future.
Editorial analysis - technical context
Companies negotiating multi-year, high-dollar compute contracts typically trade price certainty for reduced portability and increased operational coupling to a single infrastructure provider. For practitioners, that pattern raises integration, data egress, and runtime portability questions, especially when inference workloads are routed to provider-managed hardware. Observed patterns in similar arrangements show teams often build abstraction layers, containerization, and model sharding strategies to limit vendor lock-in exposure.
Context and significance
Industry observers have treated large compute agreements as a central lever for scaling large-model inference and fine-tuning workloads. The sheer dollar magnitude reported by Business Insider, $1.25 billion per month and a potential $40 billion lifetime value, highlights how AI compute demand is reshaping capital flows into bespoke data-center capacity. The deal also exemplifies a broader trend where nontraditional cloud providers expand into AI infrastructure, altering supplier dynamics for labs and enterprises that need large, low-latency inference fleets.
What to watch
Monitor public filings and follow-on disclosures for contract scope details such as service-level commitments, network/topology descriptions, and data ingress/egress terms; Business Insider and SpaceX's S-1 are the primary sources for current details. Industry context: Observers should watch whether other AI labs announce comparable long-term arrangements, and whether common engineering practices emerge to manage portability and multi-provider inference deployments.
Scoring Rationale
A reported **$1.25B/month** compute contract between Anthropic and SpaceX is a major infrastructure development that materially affects procurement, vendor dynamics, and cost modeling for labs running large inference workloads.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems


