Last updated: May 2026
Quick answers
Why did all of xAI’s co-founders leave?
All 11 original xAI co-founders left between February and March 2026 after SpaceX acquired the company in an all-stock deal. The departures were triggered by a culture clash between xAI’s research-first approach and SpaceX’s aggressive, hardware-style execution model, which included 12-to-16-hour workdays and milestone-based performance metrics replacing open-ended research cycles.
What is SpaceXAI?
SpaceXAI is the rebranded AI division that replaced xAI as a standalone company. Musk announced on May 7, 2026 that xAI would be dissolved and all AI efforts, including the Grok chatbot, would operate under SpaceXAI within SpaceX. The rebrand positions AI as part of SpaceX’s broader offering ahead of a planned $1.75 trillion Nasdaq IPO in June 2026.
Is the xAI founder exodus normal after an acquisition?
Not at this scale. Research shows about 52% of founders leave within two years of being acquired, but a 100% co-founder departure within two months is virtually unprecedented. The speed and completeness of the xAI exodus signals a deeper structural failure in cultural integration, not typical post-acquisition turnover.
On February 10, 2026, Tony Wu logged off for the last time. The xAI co-founder, one of the most operationally central members of the team Elon Musk had assembled to build a rival to OpenAI, posted no public farewell. Within 24 hours, Jimmy Ba followed. Then Kyle Kosic. Then Greg Yang. Then Toby Pohlen. By the end of March, Manuel Kroiss and Ross Nordeen walked out too, making it official: every single co-founder Musk had recruited to build xAI was gone.
Eleven out of eleven. That doesn’t happen.
The xAI co-founder exodus is the most complete leadership departure in modern tech history. It wasn’t a slow bleed over years or a couple of disgruntled executives cashing out. It was an entire founding team evacuating a company within eight weeks of an acquisition. The pattern is familiar to anyone who’s studied how Musk’s risk-taking style can simultaneously build and break organizations. The story matters to every founder who has ever considered selling, merging, or building inside someone else’s organization, because it answers a question most acquirers never ask: what happens when two operating rhythms collide and nobody planned for the wreckage?
The xAI co-founder exodus refers to the departure of all 11 original co-founders of Elon Musk’s AI startup between February and March 2026, following SpaceX’s acquisition of the company in an all-stock deal that valued xAI at $250 billion.
What caused the xAI co-founder exodus?
The trigger was SpaceX’s acquisition, but the cause was cultural incompatibility that neither side prepared for. xAI was built as a research-oriented AI lab. The co-founders, recruited from Google DeepMind, Microsoft Research, and OpenAI, expected long experimental cycles, open-ended exploration, and the kind of patient iteration that produces breakthroughs in machine learning. SpaceX runs on a completely different clock.
When SpaceX’s audit teams arrived at xAI’s offices in early March 2026, the collision became physical. Internal documents described a transition from a “research-heavy AI lab” to a “modular, product-driven organization” aligned with SpaceX’s operational philosophy. One xAI engineer told Fortune that the exploratory nature of AI development was being replaced by the rigid, milestone-based metrics used in SpaceX’s Starship and Starlink programs.
The working conditions shifted overnight. Employees reported 12-to-16-hour workdays with an expectation of responding to messages within 30 minutes at any hour. Musk directed most product changes through a 300-person group chat on X rather than internal Slack channels. For researchers accustomed to deep-focus work and multi-week experimentation, the environment became untenable.
This wasn’t a personality conflict. It was a systems-level incompatibility. SpaceX’s “test, fail, fix, fly” methodology works brilliantly for hardware engineering where you can blow up a rocket prototype and learn from the debris. AI research doesn’t work that way. You can’t crash-test a large language model. The iteration cycles are different, the feedback loops are different, and the definition of progress is different. The same pattern has played out across the broader AI hardware race, where the companies winning are the ones that match their operating tempo to the actual nature of the technical work.
The OpenAI co-founder departures offer a useful comparison. Nine of OpenAI’s original co-founders have left the company since 2024, including CTO Mira Murati and chief scientist Ilya Sutskever. But those departures happened over two years across multiple triggering events. xAI compressed a similar outcome into eight weeks, which suggests the forcing function was far more acute.
Who were the xAI co-founders and where did they go?
The full departure timeline reveals how fast the exodus accelerated once the acquisition closed on February 2, 2026.
February 2026: Tony Wu and Jimmy Ba left within 24 hours of each other on February 10. Ba’s departure was reportedly tied to tensions over demands to improve Grok’s benchmark performance against ChatGPT and Claude. Kyle Kosic (formerly OpenAI), Greg Yang (formerly Microsoft Research), and Toby Pohlen (six years at Google DeepMind) followed within days.
Early March 2026: Guodong Zhang, who ran xAI’s Imagine team, and Zihang Dai departed as SpaceX audit teams conducted their first organizational review. Christian Szegedy, known for his work on inception neural networks at Google, also left during this window.
Late March 2026: Manuel Kroiss and Ross Nordeen became the final co-founders to leave on March 27, 2026. TechCrunch confirmed Nordeen’s departure as the last co-founder exit.
The co-founders weren’t alone. More than 80 AI engineers left alongside them, according to reporting from Metaintro and Fast Company. That’s not attrition. That’s a team choosing to leave together. In an industry where companies are using AI as cover for layoffs, the xAI situation inverts the script: these researchers left voluntarily, walking away from SpaceX equity that could be worth considerably more after the IPO. They chose autonomy over money. That tells you how bad the culture fit was.
How did the SpaceX acquisition of xAI actually work?
SpaceX acquired xAI in February 2026 through an all-stock deal that valued SpaceX at $1 trillion and xAI at $250 billion. The transaction made SpaceX the most valuable private company in the world, combining rocket engineering, satellite internet (Starlink), social media (X), and artificial intelligence under one roof.
The deal’s structure matters for founders thinking about M&A. It was all-stock, meaning xAI’s shareholders received SpaceX equity rather than cash. That creates a different incentive dynamic than a cash acquisition. Co-founders couldn’t simply take a payout and walk. Their compensation was tied to SpaceX’s future valuation, which in turn depended on the success of a June 2026 IPO targeting $1.75 trillion. The math should have kept people around. It didn’t.
Musk addressed the departures on X on March 12, writing that “xAI was not built right first time around, so is being rebuilt from the foundations up. Same thing happened with Tesla.” The comparison to Tesla was deliberate. Musk has used the “rebuild from scratch” narrative before, positioning organizational overhauls as necessary growing pains rather than failures. But Tesla’s early leadership changes happened over years, not weeks. And Tesla never lost 100% of its founding team simultaneously.
What does Grok’s user decline tell us about the fallout?
Product metrics tell a story that press releases don’t. Grok’s mobile daily active users fell from 13.9 million in March to 12.2 million in April 2026, a 12.5% month-over-month decline. In the U.S., usage dropped even harder, falling 15.6% from 1.4 million to 1.1 million daily users.
The market position shift was severe. Grok entered 2026 as the world’s #2 chatbot app behind ChatGPT. By April, it had fallen to fifth place, overtaken by Claude (from Anthropic), Gemini (Google), and DeepSeek. Web visits dropped from 10.5 million daily in March to 9.3 million in April.
For founders watching this from the outside, the Grok numbers illustrate a principle that’s easy to forget during M&A negotiations: product quality is a trailing indicator of team health. The co-founders started leaving in February. The user decline showed up in March and April. By the time the metrics flagged the problem, the people who could fix it were already gone.
Musk himself acknowledged the product issues, telling employees that xAI’s AI coding tools “simply did not work” and that the underlying system needed a complete rebuild. He brought in two executives from Cursor, the AI coding startup, to lead the new effort. But replacing an entire founding team’s institutional knowledge with new hires, however talented, takes time that a pre-IPO company may not have.
The competitive landscape compounds the problem. While Grok lost ground, Anthropic grew revenue 80x in a single quarter, and Claude surged 44% in monthly users during the same period Grok declined. Google’s Gemini also gained share. The AI chatbot market isn’t waiting for SpaceXAI to figure out its organizational chart.
Why does the xAI exodus matter for founders facing acquisitions?
About 52% of founders leave within two years of being acquired, according to research aggregated by SaaStr and Qubit Capital. But those departures typically happen gradually. A CEO stays for the earnout period, a CTO drifts away after 18 months, a co-founder takes a “strategic advisor” title and stops showing up. The xAI case broke that pattern completely.
The lesson isn’t “don’t get acquired.” Acquisitions create wealth and scale opportunities that organic growth can’t match. The lesson is that cultural due diligence is as important as financial due diligence, and almost nobody does it.
Three specific failures made the xAI exodus predictable in hindsight:
No operating rhythm mapping. xAI ran on research cycles measured in weeks and months. SpaceX runs on daily standups and 30-minute response windows. Neither side mapped these differences before the deal closed. A founder evaluating an acquisition should ask: how does the acquirer define “a productive day”? If the answer is incompatible with how your team works, the integration will break.
No cultural integration plan. SpaceX sent audit teams to xAI’s offices in March. That’s an oversight mechanism, not an integration strategy. Successful acquirers like Salesforce and Cisco have dedicated M&A integration teams that spend months aligning workflows before changing anything. SpaceX applied its standard operating procedures to a fundamentally different type of organization.
No founder retention structure. The all-stock deal should have included explicit retention mechanisms: protected research budgets, autonomy guarantees, or staged integration timelines. Instead, xAI was absorbed wholesale into SpaceX’s operational culture. The co-founders had financial incentive to stay (SpaceX equity ahead of a $1.75T IPO) but no structural reason to believe their work environment would remain viable.
The Daimler-Chrysler merger of 1998 failed for similar reasons. So did AOL-Time Warner. The pattern is consistent: when an execution-driven culture acquires a creative or research-driven culture without a deliberate integration plan, the creative talent leaves. Every time.
What is SpaceXAI and what happens next?
On May 7, 2026, Musk announced that xAI would cease to exist as an independent entity. All AI products, including Grok, would operate under SpaceXAI, a new division within SpaceX. The rebrand positions AI as one component of SpaceX’s broader technology portfolio alongside Starlink and the Starship rocket program.
The timing is strategic. SpaceX filed its confidential S-1 with the SEC on April 1, 2026, and plans to begin its IPO roadshow on June 8. The offering targets $75 billion raised at a $1.75 trillion valuation. Folding xAI into SpaceX before the IPO simplifies the corporate structure and eliminates the narrative of a troubled standalone AI company from the offering documents.
SpaceXAI’s filings describe services tied to satellite-based data centers, orbital computing, and AI workload management through satellite constellations. The company also signed a major compute deal with Anthropic, giving the Claude developer access to more than 300 megawatts of compute power from SpaceX’s Colossus 1 supercomputer in Memphis, Tennessee.
The question for founders isn’t whether Musk can rebuild. He’s done it before at Tesla, and SpaceX itself nearly died in 2008 before Falcon 1’s fourth flight succeeded. The question is whether SpaceXAI can attract top AI researchers when the story of the last batch is “they all left within two months.” Reputation travels fast in machine learning circles. The talent pool is small. And the researchers who left xAI are now at competitors, carrying institutional knowledge about Grok’s architecture, training data, and failure modes.
The xAI situation also intersects with the broader Musk-Altman saga over OpenAI’s direction. Musk originally founded xAI in 2023 partly because he believed OpenAI had betrayed its founding mission by shifting from nonprofit to for-profit. The irony is thick: Musk’s own AI company lost its entire founding team for reasons that mirror some of the same cultural tensions he criticized at OpenAI.
What founders should do before any acquisition
The xAI exodus offers a concrete framework for founders on either side of an acquisition:
Map operating rhythms before signing. Document how each team defines a productive day, how decisions get made, how priorities change, and how fast people are expected to respond. If there’s a 10x gap in tempo, that’s not a difference to manage. It’s a structural incompatibility to resolve or walk away from.
Negotiate cultural protections into the deal. Protected research budgets, guaranteed team autonomy windows, named reporting lines, and staged integration timelines belong in the term sheet, not in a handshake. Buffer’s transparent salary structure survived their acquisition because it was contractually protected. xAI’s research culture had no contractual protection at all.
Watch the product metrics early. If you’re acquiring a team, track the product’s quality indicators weekly during the first 90 days. User engagement, feature velocity, and internal NPS scores are leading indicators of whether the integration is working. Grok’s user decline started appearing within weeks of the co-founder departures. The data was there. Nobody was watching.
Plan for the worst-case scenario. What happens if every key person leaves? If the answer is “the product dies,” the acquisition isn’t buying a product. It’s buying people. And people leave when the environment changes. Structure the deal accordingly.
Before any merger, map the operating rhythm of both teams. Mismatched cadences kill culture faster than misaligned valuations. The xAI story will be studied in business schools for years as proof that the hardest part of any deal isn’t the cap table, the share price, or the product roadmap. It’s whether two groups of people can actually stand to work together every day. Eleven co-founders, $250 billion in valuation, and eight weeks. That’s how fast culture kills a deal when nobody plans for the collision.
