News listSpaceXAI Talent Exodus: 50 Employees Lost in Three Months, All 11 xAI Co-founders Have Departed
動區 BlockTempo2026-05-15 01:53:55

SpaceXAI Talent Exodus: 50 Employees Lost in Three Months, All 11 xAI Co-founders Have Departed

ORIGINALSpaceXAI 人才潰堤:三個月流失 50 名員工,xAI 11 位共同創辦人全數離職
AI Impact AnalysisGrok analyzing...
📄Full Article· Automatically extracted by trafilaturaGemini 翻譯1449 words
In February 2026, after SpaceX completed its acquisition of xAI, the merged company lost more than 50 employees within three months, with broader estimates putting the figure above 80. More critically, all 11 original co-founders of xAI have departed. (Background: Musk announced that xAI is no longer an independent company, renamed "SpaceXAI" and positioned as an AI product under SpaceX) (Context: Musk's SpaceX obtained an acquisition option for Cursor: $60 billion to buy the AI development tool or a $10 billion breakup fee) This is the pace of talent loss at SpaceXAI since completing the merger in February 2026, with broader estimates exceeding 80 people. Among them, the most striking fact is that all 11 original co-founders of xAI have departed. 0 founders remaining—three months and they all said goodbye. The last two, Manuel Kroiss and Ross Nordeen, quietly left in late March, closing the chapter on this founding team. They didn't leave because they couldn't find opportunities; they actively chose to walk away. In February 2026, SpaceX officially acquired xAI, and earlier this month Musk announced that the originally independently operated xAI has been formally renamed SpaceXAI, becoming an AI division under SpaceX. In the eyes of outsiders, this is Musk's strategic move to integrate his various resources and counter OpenAI. But for xAI's researchers and engineers, integration first brought performance pressure and cultural conflict. Those who left were not peripheral staff. Department heads for coding, world models, Grok voice, and several other key areas have all departed. These are not roles that can be quickly backfilled—after all, what they possess is judgment over model direction, along with the unique research context accumulated during their time at xAI. The logic of integration was originally: SpaceX's engineering discipline plus xAI's research capabilities equals a more powerful AI organization. But reality has gone the other way—research-oriented talent is leaving under the high-pressure engineering culture faster than expected, and those leaving are precisely the hardest to replace. The biggest beneficiary of this exodus is made clear by the numbers. Meta has hired at least 11 researchers and engineers from xAI since February. For Meta, every senior researcher who leaves SpaceXAI represents engineering knowledge accumulated during the Grok model training process, transferred along with the person themselves. Thinking Machines Lab, founded by Mira Murati, has also taken in at least 7 former xAI employees. Murati is the former Chief Technology Officer of OpenAI, and this new company she founded after leaving in 2024 is rapidly accumulating top-tier research talent (though Thinking Machines Lab itself also faces talent turnover: its 5 founders have successively moved to Meta). Anthropic has also absorbed at least 2 former xAI employees this year. Broader destinations also include OpenAI, Google DeepMind, and the option of starting their own ventures. This flow of talent is, in a sense, a microcosm of the 2026 AI competitive landscape: the metric for measuring a company's competitiveness isn't just model benchmarks, but also whether researchers are willing to stay. Musk tried to reshape xAI using the SpaceX model: high pressure, rapid iteration, with surpassing the strongest competitor as the sole objective. This approach works in rocket engineering, because engineering metrics are clear and success or failure is immediately apparent. But the rhythm and logic of AI research may be different: breakthroughs in model capabilities often come from long-term directional bets rather than sprints for short-term benchmark scores; researchers' creativity also cannot endure the pressure of repeated benchmark competitions. In the AI race, compute and capital are the entry ticket, but retaining researchers is the true moat. The problems SpaceXAI now faces, beyond whether Grok's benchmark scores are high enough, also include who will decide which direction the next version should take.
Data Status✓ Full text extractedRead Original (動區 BlockTempo)
🔍Historical Similar Events· Keyword + Asset Matching6 items
💡 Currently matching via keywords + symbols (MVP) · Will be upgraded to embedding semantic search later
Raw Information
ID:ae9cf0f5f9
Source:動區 BlockTempo
Published:2026-05-15 01:53:55
Category:zh_news · Export Category zh
Symbols:Unspecified
Community Votes:+0 /0 · ⭐ 0 Important · 💬 0 Comments