xAI Lost 8 Co-Founders. SpaceX Got the Engineers.
What launched / what broke
xAI launched in 2023 as Elon Musk's independent bid to build a truth-seeking AI company. In early 2026 SpaceX merged with it. Eight of the original co-founders departed in the following months, according to reporting by NewsBytesApp and Fortune India. SpaceX's Michael Nicholls, described by Fortune India as a SpaceX executive, became xAI president. Three new leaders now run model training, post-training/RLHF, and reinforcement-learning operations. SpaceX reported a $5 billion loss in 2025 on $18.5 billion revenue that now folds in xAI, per RTE News. An IPO for the combined entity is targeted for June 30 2026 with market cap expectations above $1.4 trillion, per Fortune India. The original thesis broke: Musk sold the public on xAI as a rival to OpenAI that would pursue truth without corporate safety filters. In practice the company could not close the capability gap fast enough on its own balance sheet. The merger reveals that xAI's burn rate and progress were unsustainable as a standalone. The departures are not random attrition; they are the predictable result of folding a startup into a much larger, cash-flow-negative aerospace company whose first priority is Starship, Starlink cash flow, and an IPO. The steelman for the merger is real: Anthropic has Google's balance sheet, OpenAI has Microsoft's, and xAI as a standalone was burning cash against two companies with essentially unlimited compute backing. Folding into SpaceX may have been the only viable path to staying in the compute race at all. But that argument and the mission statement cannot both be true at the same time.
The pitch was a pure AI company chasing cosmic understanding with elite talent. The reality is an internal engineering division whose primary product will be whatever improves SpaceX's valuation multiple.
What Nobody at the Company Can Say
The departed co-founders thought they were building an AI company. They were building a feature for SpaceX's IPO story. The quiet exits and radio silence suggest strong NDAs.
The Engineer Who Quit
The typical engineer who quit joined xAI believing they would outpace OpenAI and Anthropic on both capability and honesty. They discovered the best talent was being asked to optimize for SpaceX's launch cadence and regulatory arbitrage rather than frontier reasoning. After watching the mission collapse into making the numbers work for the IPO, they took the severance, signed the NDA, and left.
Who Pays
Early xAI employees and co-founders
Immediate; locked in by NDAs and restricted equity
Crushed equity expectations and lost years on a mission that no longer exists; they traded high-risk startup equity for SpaceX paper whose value depends on an IPO that may price in heavy AI losses
AI safety advocates and independent researchers
Slow-burn over 12-24 months as the independent lab count shrinks
Consolidation reduces the number of independent labs willing to criticize regulatory capture or maintain alternative safety frameworks
Dead Pool Watch
The merged entity will survive but its character has changed permanently. Watch for any remaining senior researchers who joined for the original mission statement. If two or more additional high-profile technical departures leak in the next quarter, it confirms the restructuring is not finished.
In 6 Months
SpaceX-xAI files IPO paperwork with xAI positioned as AI infrastructure, not an independent frontier lab
Signal S-1 filing emphasizes Starlink growth and Starship with xAI described only in an infrastructure appendix
Incremental model releases tuned for internal SpaceX use cases; no Grok successor that leads public benchmarks
Signal Grok's public leaderboard position drops relative to OpenAI and Anthropic by Q3 2026
What Would Change This
The bottom line changes only if xAI's next model demonstrably leaps past current leaders on independent benchmarks while simultaneously improving SpaceX's unit economics in a visible way. That would prove the merger created synergies instead of subordination. Short of that evidence, this is the story of one ambitious AI effort being absorbed and diluted to protect a larger company's balance sheet and IPO timeline.
Prediction Markets
Prices as of 2026-04-12 — the analysis was written against these odds
Related
EFF Logged Off X After 20 Years. It Didn't Say Goodbye.
moneyOpenAI Wants Regular People to Buy Its IPO. That's Not Generosity.
dramaSomeone Threw a Firebomb at Sam Altman's House. He Responded With a Blog Post.
dramaAnthropic Is 'Exploring' Custom Chips. That Is the Most Expensive Way to Say No.