Anthropic Partners with SpaceX as the AI Competition Takes a Strange Turn

Anthropic and Elon Musk’s SpaceX announced on Wednesday that they have finalized an agreement for Anthropic to utilize computing resources from xAI’s data center located in Memphis, Tennessee. This development is part of an industry effort to secure sufficient computing power for complex AI software. Previously independent, SpaceX and xAI merged earlier this year into a single entity owned by Musk, now called SpaceXAI.

Anthropic executives revealed the news during the company’s annual developer conference in San Francisco. Additionally, SpaceXAI released a blog post with further details about the collaboration, which will enable Anthropic to leverage xAI’s Colossus 1 supercomputer.

This partnership is timely for SpaceXAI, which is looking to go public as early as next month. Collaborating with a prominent AI laboratory could enhance SpaceX’s appeal to investors interested in the lucrative potential of additional data centers, including those situated in space.

Notably, SpaceXAI mentioned in its blog that Anthropic has shown “interest” in partnering to develop “orbital AI compute capacity”—effectively, data centers in outer space. Having Anthropic as a potential client could instill confidence in investors regarding the viability of SpaceXAI’s costly supercomputing initiatives.

Earlier this year, Musk publicly criticized Anthropic’s AI models on X, labeling the company’s policies as “misanthropic” and “evil,” and questioning their fairness without providing evidence. However, his stance has shifted; he recently expressed on X that he “spent considerable time last week engaging with senior members of the Anthropic team to grasp their assurances that Claude is beneficial for humanity and was impressed.”

SpaceXAI began construction on Colossus 1 at a former Electrolux site in 2024. The company claims it to be one of the largest and fastest AI supercomputers globally, featuring around 220,000 Nvidia GPUs, from the H100 and H200 series to the newer GB200 chips. They proudly assert that the supercomputer was built in just 122 days.

However, emissions from gas turbines at the data center have led to multiple complaints from local residents. Last month, environmental protesters gathered at a nearby SpaceX investor event ahead of the IPO.

Anthropic intends to utilize this enhanced computing power to “directly improve capacity for Claude Pro and Claude Max subscribers,” as confirmed by both companies. The agreement grants Anthropic access to over 300 megawatts of new capacity, translating to approximately 220,000 Nvidia GPUs. Recently, software developers have increasingly reported issues with rate limits and outages affecting Claude Code, attributed to the limitations of existing computing resources.

As usage of services like OpenAI’s Codex and Anthropic’s Claude Code increases—often running coding tasks continuously for hours—the systems are becoming overloaded. According to Anthropic, the average developer now spends at least 20 hours weekly using Claude Code.

Yesterday, The Information reported that Anthropic has committed to spend $200 billion on Google’s AI cloud services and TPU chips. Since 2023, Anthropic has also utilized Amazon’s cloud computing services and AI chips. Last month, they pledged “more than $100 billion over the next ten years to [Amazon] technologies.” Contracts with Anthropic and OpenAI now constitute “more than half of the $2 trillion in backlogs at major cloud providers” such as Amazon, Microsoft, and Google.

Paresh Dave and Zoe Schiffer contributed to this report.