Anthropic has announced an expansion of its infrastructure and product capacity through a major deal to use the full compute power of SpaceX’s Colossus 1 data center. The facility is reported to provide around 300 megawatts of AI compute capacity with more than 200,000 NVIDIA GPUs, including H100 and newer-generation chips. This setup is designed for large-scale AI training and high-volume inference workloads. Along with this, Anthropic is also increasing usage limits for Claude Code, doubling session caps and raising API rate limits for higher-tier users.
These developments seem like a two-layer expansion strategy, which includes securing massive backend compute resources while immediately easing constraints on developer-facing products. On the infrastructure side, access to a 300 MW-class cluster puts Anthropic in a small group of AI labs operating at hyperscale compute levels. Notably, systems of this size are usually used to train frontier large language models that need thousands of GPUs working together for weeks or even months at a time. The reported hardware setup – mainly NVIDIA’s H100 and H200 GPUs along with newer GB200-class chips – suggests the facility is built for both training advanced next-generation models and handling very high-volume inference workloads.
Colossus 1 is described as a high-density AI data center built specifically for large-scale machine learning workloads, rather than general-purpose cloud computing. It is engineered with advanced power systems, liquid cooling, and ultra-low-latency networking so that tens or even hundreds of thousands of GPUs can operate together as a unified compute system. And with around 300 megawatts of capacity, it ranks among the largest dedicated AI computing sites in the world. The involvement of SpaceX adds another layer of strategic significance. While SpaceX is primarily known for aerospace and satellite operations, it has increasingly been associated with large-scale compute infrastructure initiatives.
Meanwhile, on the product side, Anthropic’s changes to Claude Code directly address growing demand from developers using AI for software engineering tasks. Claude Code is designed for extended coding sessions, multi-file reasoning, debugging workflows, and agent-like programming assistance. Therefore, by doubling session limits for Pro, Max, Team, and Enterprise users, the AI firm is effectively allowing longer uninterrupted work cycles, which is especially important for complex development tasks that previously risked hitting usage caps mid-session.
Additionally, the removal of peak-hour throttling for higher-tier users improves consistency during high-traffic periods, reducing latency spikes and session interruptions that have historically affected AI coding tools. The company is also increasing API rate limits for advanced models like Claude Opus, which is particularly relevant for enterprise deployments where large volumes of requests are processed in real time.
The Tech Portal is published by Blue Box Media Private Limited. Our investors have no influence over our reporting. Read our full Ownership and Funding Disclosure →
Ashutosh is a Senior Writer at The Tech Portal, largely reporting on new tech, and intersection of technology and business. Ashutosh’s career spans across nearly a decade of technology writing across multiple platforms and languages.
