Anthropic has signed an agreement with SpaceX to use all of the compute capacity at Colossus 1, the company’s data centre in Memphis, Tennessee. The deal, announced on 6 May 2026, adds more than 300 megawatts and over 220,000 NVIDIA GPUs to Anthropic’s available infrastructure — and takes effect within the month.
The agreement arrives as Anthropic moves to ease persistent capacity pressure on its paid Claude users. Alongside the compute deal, the company made three immediate changes to rate limits across its Claude plans — including doubling Claude Code’s five-hour usage caps and removing peak-hour throttling for Pro and Max accounts.
On the same day, Elon Musk confirmed on X that xAI would be dissolved as a separate company, with all AI products moving under SpaceX as “SpaceXAI.” SpaceX had acquired xAI in an all-stock transaction in February 2026.
Colossus 1 Is Now Claude’s Engine
Anthropic’s deal with SpaceX hands it one of the largest GPU clusters on Earth — and today’s users feel it immediately.
What the Deal Covers
Colossus 1 sits in a former Electrolux factory at 3231 Riverport Road, Memphis, Tennessee. As of late 2025, the cluster houses a mix of over 150,000 NVIDIA H100, 50,000 H200, and 30,000 GB200 GPUs — built in phases over roughly 122 days from its initial deployment. Anthropic gains access to the full capacity of this cluster.
SpaceX acquired xAI in February 2026 in an all-stock deal. Elon Musk announced on 6 May 2026 that xAI would be dissolved as a standalone entity, with all AI products — including Grok — operating under “SpaceXAI,” a division of SpaceX. Grok development has already moved to Colossus 2, making Colossus 1 available for Anthropic’s use.
Last month, Anthropic said demand for Claude had created “inevitable strain” on its infrastructure, affecting reliability during peak hours for paid users. The other large compute deals — with Amazon, Google, and Microsoft — are not expected to come fully online until late 2026 or into 2027. The SpaceX deal provides capacity now.
On X, Musk said he spent time with senior Anthropic staff in the days before the announcement and was “impressed.” He also noted SpaceX would “reserve the right to reclaim the compute if their AI engages in actions that harm humanity.” This came after months of public criticism of Anthropic from Musk, including referring to the company as “misanthropic.”
Rate Limit Changes — Effective 6 May 2026
Three changes took effect immediately. Tap each tab to explore what changed for your plan.
Claude Code is Anthropic’s agentic coding tool. Learn how it compares to other AI coding platforms making headline deals.
For context, the OpenAI–Microsoft amended partnership and its revised revenue terms were covered separately.
Anthropic’s Full Compute Portfolio
The SpaceX deal is the only one with immediate effect. The others come online across 2026 and 2027. Tap any row for details.
“Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good.”— Elon Musk, post on X, 6 May 2026
The Road to Colossus 1
Beyond Earth
Orbital AI Compute — Expressed Interest, Not a Signed Contract
As part of the agreement, Anthropic has “expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity.” The SpaceXAI announcement describes this as an engineering programme rather than a research concept, citing SpaceX’s launch cadence and constellation operations as key reasons. However, no public milestones, financing plan, launch schedule, or deployment timeline have been disclosed. Space-based compute is presented as a long-term direction to address power and cooling constraints on Earth, not an imminent product.
The Wider Picture
In March 2026, the Pentagon declared Anthropic a supply chain risk, blocking it from US military contracts after negotiations over model usage terms failed. Anthropic sued the Trump administration in two courts to reverse that decision. That litigation is ongoing. Meanwhile, the Defense Department has been using competing AI models from other providers.
Anthropic is currently in talks with investors about raising capital at a reported valuation of $900 billion. The company was founded in 2021 by researchers who left OpenAI, and is best known for its Claude model family. The compute expansion across multiple partnerships has been covered as part of Anthropic’s broader infrastructure build-out during this fundraising period.
Enterprise customers in financial services, healthcare, and government increasingly need in-region infrastructure to meet data residency and compliance requirements. Anthropic’s deal with Amazon includes additional inference capacity in Asia and Europe. The company stated it will partner only with democratic countries whose legal and regulatory frameworks support investments of this scale.
Anthropic has committed to covering any consumer electricity price increases caused by its US data centres. As the company expands internationally, it is exploring ways to extend that commitment to new jurisdictions and work with local community leaders in areas that host its facilities.
The Anthropic–SpaceX agreement was covered in terms of Colossus 1 access, the GPU count, the power capacity, and the immediate rate limit changes for Claude users. The xAI-to-SpaceXAI rebrand, confirmed by Elon Musk on the same day, was discussed as part of the same announcement. Orbital compute ambitions were noted as expressed interest, with no confirmed deployment schedule. Anthropic’s broader compute portfolio — involving Amazon, Google, Broadcom, Microsoft, NVIDIA, and Fluidstack — was summarised alongside the ongoing Pentagon litigation and the company’s international expansion plans.
