In an unexpected collaboration, SpaceX—now merged with xAI since earlier this year—is partnering with Anthropic to enhance Claude’s capabilities. This alliance is noteworthy as it unites companies that previously found themselves at odds during President Donald Trump’s administration.
On Wednesday, both entities revealed that Anthropic will gain access to Colossus 1, the extensive AI supercomputer cluster created by xAI for training and operating sophisticated AI models. This deal promises Anthropic over 300 megawatts of additional computing power, aiming to boost Claude Pro and Claude Max services.
“We’ll begin scaling Claude inference on Colossus shortly. Excited to collaborate with SpaceX,” noted Anthropic co-founder Tom Brown on X. “Given the immense demand for AI, moving vast amounts quickly is crucial, and there’s no better at it than SpaceX (on or off Earth).”
Elon Musk further explained in a subsequent post on X that dialogues with senior Anthropic officials reassured him of their responsible approach to AI development.
“I spent considerable time last week engaging with the senior team at Anthropic to learn about their efforts to ensure Claude benefits humanity, and I was impressed,” Musk shared. “Everyone demonstrated high competence and a strong commitment to ethical practices. No alarms were triggered regarding any malicious intent. So long as they maintain rigorous self-examination, Claude is likely to be beneficial.”
Musk also mentioned that “SpaceXAI” has already transitioned training to Colossus 2, an upgraded AI supercomputer cluster designed to surpass the approximately 220,000 Nvidia GPUs of Colossus 1 and accommodate future Grok models.
This partnership highlights differing stances within the AI sector: Musk aligns with Trump and U.S. government departments like Justice and Defense, whereas Anthropic advocates for stringent AI safety standards. This has led to conflicts, notably when Anthropic resisted Pentagon demands for unrestricted military use of Claude due to ethical safeguards against mass surveillance or autonomous weaponry. Following this, Trump directed federal agencies to phase out Anthropic’s tech, while Pentagon officials insisted on allowing any lawful government use.
The collaboration underscores the evolving alliances in Silicon Valley driven by the quest for computing power. In February, Musk integrated xAI into SpaceX, citing limitations in Earth-bound AI data centers due to electricity and cooling needs. On Wednesday, Musk confirmed that xAI would be dissolved, with parts of its operations continuing under “SpaceXAI.”
In March, Musk criticized Anthropic as “misanthropic” after it secured $30 billion at a valuation of $380 billion. Despite past tensions, Anthropic added this to their growing list of infrastructure partnerships that include Amazon, Google, Broadcom, Microsoft, and Nvidia.
Anthropic expressed interest in further collaborations with SpaceX for developing significant gigawatts of orbital AI computing capacity. While space-based data centers are still experimental and face engineering challenges, SpaceX indicated that existing terrestrial resources may not suffice for future AI demands, suggesting space as a potential source for sustainable energy supplies.
“If these technical hurdles can be addressed, space-based compute could provide nearly unlimited sustainable power with minimal Earth impact,” the company noted.