Multiple outlets reported Monday that SpaceX has acquired xAI, with TechCrunch claiming the deal is "official" and valued at $1.25 trillion. The real picture is murkier.
Sources familiar with the discussions say merger talks are advancing but not finalized. Two Nevada entities formed January 21 to facilitate a potential combination, and Musk appeared to confirm exploration via an X post. Cathie Wood told Bloomberg February 2 the merger is "becoming more and more likely," citing orbital data centers as the rationale.
What we know: xAI (valued around $200B in September 2025) burns roughly $1B monthly. SpaceX (valued around $800B in late 2025) generates 80% of revenue from launching its own Starlink satellites. Both are private. Both are controlled by Musk, who previously merged Tesla-SolarCity in 2016 and folded xAI into X in 2025.
The space data center concept deserves skepticism. Musk's memo claims orbital facilities solve power and cooling constraints, but the physics are challenging. Latency to ground stations via free-space optical links adds milliseconds versus terrestrial fiber. Launch costs per rack remain orders of magnitude higher than building in Virginia. Thermal management in vacuum requires radiators, not fans. And machine learning inference latency - critical for real-time AI applications - gets worse when your compute is 550km up.
Google and AWS have experimented with edge AI on satellites for specific workloads like Earth observation, where data generation happens in orbit. The business case for general-purpose cloud computing in space remains unproven. xAI's $200M DoD contract for Grok suggests defense applications might justify premium costs, but commercial viability is another question.
History suggests Musk's timelines are optimistic and his technical claims need verification. The merger - if it closes - would be significant for concentration of AI and launch capability. The orbital data center pitch? We'll see.
Three things to watch: actual deal closure with terms, SpaceX IPO timeline impact, and whether any hardware actually ships to orbit.