Anthropic compute – Anthropic is expanding Claude Code capacity by tapping SpaceX’s Colossus One data center as demand strains its current compute limits.
Anthropic is scrambling for more AI compute power, and it is looking to SpaceX to help it keep up.
With demand rising for its AI coding service, Claude Code, Misryoum reports that Anthropic will route additional workloads to SpaceX’s Colossus One data center in Memphis. The move is tied to infrastructure capacity, as Anthropic addresses the strain created by higher usage among developers.
In parallel, Anthropic said it is also raising developer limits.. Misryoum notes that the company is doubling the current 5-hour rate limit for developers on Pro. Max. Team. and seat-based Enterprise plans.. It is also increasing rate limits for developers using the Claude Opus model’s API. allowing more input and output per minute.
This matters because compute is often the bottleneck for fast-growing AI tools. Even when models and software are ready, companies can only scale outputs as quickly as their access to specialized chips allows.
Anthropic’s decision to use SpaceX’s infrastructure comes at a time when the wider AI industry is racing to secure capacity. Misryoum reports that SpaceX’s Colossus One hosts tens of thousands of advanced Nvidia chips used for compute-intensive AI work.
The deal points to a broader shift in how AI companies find room to grow.. Rather than relying solely on internal capacity. Misryoum says Anthropic is sourcing additional chip time and processing capability from large infrastructure operators.. In this context, the SpaceX arrangement fits into a pattern already underway with other hyperscaling players.
Misryoum also highlights that Anthropic expects the partnership to provide more than 300 megawatts of computing capacity within the month, spanning more than 220,000 Nvidia GPUs. The company described the effort as a response to already strained computing resources.
For SpaceX and its AI-linked ecosystem, compute sales can turn existing infrastructure into a revenue stream. And for AI developers, higher limits can mean fewer interruptions as they build and test with Anthropic’s tools.
Misryoum’s takeaway: this kind of capacity deal is becoming as important as model releases. When competition is about speed and reliability, access to large-scale compute can determine how quickly companies can satisfy users and expand services.