On May 6, SpaceX submitted two“SpaceXAI” wordmark applications to the U.S. Patent and Trademark Office. The filings were made the same day Musk announced xAI would be dissolved as a standalone company and folded into SpaceX as a new division.
The first trademark covers the core vision: “satellite-based data center services and orbital computing infrastructure,” along with AI software-as-a-service for data processing and AI workload management on satellite constellations. The second filing is broader: “global positioning system using satellite constellations; internet server, namely, computer network server; telecommunication hardware,” plus internet service provider services, cloud storage, and social networking. The social networking inclusion reflects a reality: xAI acquired X last year, meaning SpaceX now technically controls what used to be Twitter.
Elon Musk also confirmed SpaceXAI will have a new logo. xAI won‘t exist as a separate company anymore.
11 Founders Walked Out. Musk Is Rebuilding.
xAI had 12 original co-founders. By May 2026, all 11 non-Musk founders had left. The departures began after SpaceX acquired xAI in February 2026. They covered every key function—pre-training, reasoning, code and image generation. Pre-training lead Zhuang Juntang announced his departure on May 9, after another pre-training lead, Manuel Kroiss, had left in mid-March.
Musk himself admitted in an X post: “xAI was not built right first time around, so is being rebuilt from the foundations up.”
Reid Hoffman summed it up this way: xAI is“a complete train wreck,” and SpaceX is“not an AI company” but“buying your way into relevance” with its market cap.
Colossus 1 Is Now Anthropic’s Compute
On May 6, SpaceX opened Colossus 1—the Memphis supercomputer with over 220,000 Nvidia GPUs and 300+ megawatts of capacity—to Anthropic. Anthropic is paying $1.25 billion per month through 2029 for full access. The contract includes a 90-day cancellation clause.
For Anthropic, the deal was critical. Claude Code had been hitting usage limits due to compute shortages. With Colossus 1, Anthropic doubled Claude Code‘s rate limits, removed peak-hour restrictions, and raised API limits for Claude Opus.
For SpaceX, the deal monetized a resource it wasn’t fully using. According to The Information, xAI was only utilizing about 11 percent of its total GPU capacity before the merger. Musk didn‘t kill xAI because he gave up on AI. He killed it because the standalone model didn’t work. The infrastructure, however, was still valuable—just not for his own models.
Orbital Data Centers Are Part of the Plan
The trademark filings also protect something that doesn‘t yet exist: orbital data centers. SpaceX has already filed an application to launch up to 1 million satellites to build distributed AI data centers in orbit. The satellites would operate at 500 to 2,000 kilometers, using solar panels for power and radiation cooling to replace water or air cooling.
Elon Musk has shown a concept satellite called “AI Sat Mini,” featuring large solar panels and about 100 square meters of radiating panels, with a size exceeding the Starship rocket itself. The company is also pursuing a chip development plan called Terafab, with a Texas-based chip factory focused on producing space-grade AI chips designed to operate in high-temperature and radiation environments. SpaceX estimates the orbital system could add about 100 GW of AI computing capacity per year if fully deployed.
P.S. xAI lost 11 founders, $400 million per month in training costs, and any real shot at catching OpenAI. Reid Hoffman called it a“complete train wreck.” But Colossus 1 is still running—just for Anthropic. The name is gone. The infrastructure isn‘t. Musk isn’t trying to beat Anthropic anymore. He‘s trying to bill them.
