Anthropic has signed a significant agreement to acquire SpaceX’s Colossus 1 data center, which boasts over 300 megawatts of power and 220,000 NVIDIA GPUs, with operations set to commence within the month. This acquisition is part of Anthropic’s broader strategy to secure available power sources amid a growing bottleneck in the construction of new data centers due to prolonged grid connection times and transformer shortages. Unlike many U.S. data centers planned for 2026, which face delays or cancellations, Anthropic is opting to lease existing powered facilities to expedite their capacity growth, marking a pivotal shift in the competitive landscape of AI infrastructure development.
Amazon: Amazon Web Services (AWS) is a dominant cloud platform offering Trainium chips optimized for AI model training. It acts as Anthropic’s primary cloud provider and strategic training partner. AWS has deepened its compute commitments to Anthropic to fuel Claude model development and deployment.463
Google: Google Cloud provides TPU accelerators tailored for large-scale AI training and inference. It partners with semiconductor firms to expand AI infrastructure capacity. Google supplies substantial compute to Anthropic through recent agreements supporting its diversified supply chain.477
NVIDIA: NVIDIA produces GPUs that power the majority of AI training and inference workloads globally. Its architecture enables massive parallel processing in supercomputers. NVIDIA GPUs equip the Colossus 1 cluster now fully available to Anthropic through SpaceX.23
SpaceX: SpaceX develops reusable rockets, Starlink satellite internet, and supports xAI’s supercomputing initiatives with facilities like Colossus powered for dense AI workloads. The company leverages its infrastructure expertise to enter the AI compute leasing market. SpaceX is providing Anthropic with full capacity at its Colossus 1 data center in Memphis, including interest in future orbital AI compute collaboration.2128
Broadcom: Broadcom specializes in semiconductors, including custom AI chips and networking solutions for data centers. It collaborates with cloud providers on AI compute initiatives. Broadcom contributes hardware capacity in Google-led deals benefiting Anthropic’s infrastructure expansion.48
Anthropic: Anthropic is an AI company focused on developing safe and reliable frontier models like Claude, which excels in coding and complex reasoning tasks. It employs a multi-vendor compute strategy across AWS, Google Cloud, Azure, and specialized providers to train and deploy its models amid surging demand. In this deal, Anthropic secures exclusive access to SpaceX’s entire Colossus 1 data center, which is going live soon to support expanded Claude usage limits.2045
Microsoft: Microsoft Azure delivers enterprise cloud services with strong AI integrations and GPU support. It teams with hardware leaders for high-performance compute. Azure provides dedicated capacity to Anthropic via multi-year infrastructure pacts.51
Fluidstack: Fluidstack offers GPU-centric cloud infrastructure for AI and high-performance computing. It specializes in scalable data center solutions. Fluidstack partners with Anthropic on custom infrastructure builds to enhance long-term compute availability.46
Power Bottleneck: Power grid constraints and transformer shortages have overtaken GPU availability as the main barrier for new AI data centers.35
Data Center Delays: Many U.S. data centers announced for 2026 face delays or cancellations due to prolonged grid connection times and equipment issues.38
Strategic Acquisition: AI developers are shifting to leasing powered existing facilities rather than constructing new ones to accelerate capacity growth.42