Anthropic turns to SpaceX as Claude usage explodes
Anthropic has partnered with SpaceX for AI compute capacity while raising Claude usage limits amid surging enterprise demand.
Claude needs more compute, and they finally have a solution. Anthropic announced on 6 May 2026 that it has signed a compute partnership with SpaceX while simultaneously raising usage limits across Claude products.
The company said the additional infrastructure will improve reliability and availability for Claude users within weeks.
Claude users are getting higher limits immediately
The first visible changes are aimed at developers and heavy users. Anthropic said it has doubled Claude Code’s five-hour rate limits for Pro, Max, Team and seat-based Enterprise plans. The company has also removed peak-hour limit reductions for Pro and Max subscribers.
API capacity is increasing, too. Anthropic confirmed that API rate limits for Claude Opus models have been raised significantly to support larger coding, automation and agentic workloads.
The SpaceX partnership gives Anthropic massive compute capacity
The headline deal centres around compute access. Anthropic said the agreement provides access to the full compute footprint at SpaceX’s Colossus 1 data centre, which includes more than 300 megawatts of power capacity and over 2.2 lakh NVIDIA GPUs.
That infrastructure is expected to come online within the month. The partnership adds another major layer to Anthropic’s increasingly diversified compute strategy, which already spans AWS Trainium chips, Google TPUs and NVIDIA GPU clusters.
Why Anthropic is scrambling for more AI infrastructure
Demand for Claude has accelerated sharply this year. Anthropic said its annualised revenue run rate has crossed US$30 billion, compared to roughly US$9 billion at the end of 2025. That surge has increased pressure on compute availability across both consumer and enterprise products.
The SpaceX deal is designed as near-term relief. Anthropic said the additional capacity will help absorb usage spikes while larger infrastructure agreements ramp up over the next two years.
This is part of a much bigger infrastructure expansion
The SpaceX announcement is only one piece of Anthropic’s compute push. In April, the company expanded its partnership with Amazon Web Services for up to 5 gigawatts of AI infrastructure capacity, including nearly 1 gigawatt of Trainium compute expected by the end of 2026.
Anthropic has also signed a multi-gigawatt agreement with Google Cloud, and Broadcom is focused on next-generation TPU infrastructure expected to arrive from 2027 onwards.
Microsoft and FluidStack are also part of the roadmap
The company’s infrastructure pipeline now spans multiple continents and providers. Anthropic previously committed to purchasing US$30 billion worth of Azure compute capacity through partnerships involving Microsoft Azure and NVIDIA. At current exchange rates, that works out to roughly Rs 2.5 lakh crore.
Another large investment is already underway. Late last year, Anthropic announced a US$50 billion infrastructure partnership with Fluidstack to build AI data centres across Texas and New York, equivalent to around Rs 4.1 lakh crore.
AI infrastructure is becoming a global supply-chain race
This is no longer a normal cloud expansion cycle. AI labs are now competing for chips, electricity, cooling systems, land, and data-centre capacity at an industrial scale. Analysts say compute access is quickly becoming one of the most important strategic advantages in the AI market.
Anthropic appears determined to secure multiple layers of redundancy. The company said it is also expanding international inference infrastructure across Asia and Europe to meet local compliance, residency and governance requirements for sectors such as healthcare, financial services and government.
The electricity question is becoming impossible to ignore
AI growth is also colliding with power infrastructure realities. Anthropic said it plans to cover any increases in consumer electricity prices linked directly to its US data-centre operations. The company added that it is exploring ways to extend similar commitments to other regions as expansion continues.
That reflects a growing industry concern. As frontier AI systems become larger and more energy-intensive, infrastructure discussions are increasingly revolving around power availability as much as model capability.
Anthropic’s strategy is becoming clearer
The company is building for scale before demand fully arrives. For Anthropic CEO Dario Amodei, the SpaceX partnership offers immediate breathing room while longer-term infrastructure projects mature across Amazon, Google, Microsoft and other hardware partners.
The larger signal is hard to miss. AI companies are no longer competing only on models or benchmarks. They are competing on who can secure enough compute, energy and infrastructure to keep the next generation of AI systems running reliably at a global scale.