Why did multiple members of xAI resign on the same day?

On May 9th of this year, Zhuang Juntang, the pre training manager of xAI, Xiuyu Li, a technical member, and Fraster Cook, a front-end engineer, announced their resignation on the same day. This is not the first time that xAI has experienced talent turbulence: in February, co founders John Woo and Jimmy Ba left within 48 hours; At the end of March, the last founding member Ross Nordeen resigned, marking the departure of all members of Musk's "AI Dream Team" formed in 2023. The departure of multiple core members on the same day is not accidental, but the inevitable result of the concentrated outbreak of four contradictions: the loss of strategic independence, high-pressure management conflicts, technological route differences, and cost and commercialization difficulties, reflecting the deep cracks in Musk's AI empire.

1、 End of Independent Identity: The 'Fatal Strike' of Being Acquired by SpaceX

On May 7, 2026, Musk announced that xAI would no longer exist as an independent company and would merge with SpaceX to become "SpaceXAI". In February, SpaceX completed the acquisition of xAI with a valuation of $1.25 trillion, with the core objective of paving the way for SpaceX's IPO. This integration became the last straw that crushed the core team.

For top scientists, the original intention of xAI was to establish an independent AI laboratory that "understands the essence of the universe and explores the truth", rather than an affiliated department of aerospace giants. After the acquisition, xAI needs to fully comply with SpaceX's strategy, tilt its resources towards aerospace AI, and completely dilute its research independence. More importantly, SpaceX announced on the same day that it will supply all 220000 NVIDIA GPUs of supercomputing resources to its competitor Anthropic, directly cutting off the lifeline of xAI's core computing power and making the technical team unable to see the future, accelerating their collective departure.

2、 The Tragedy of High Voltage Management: The Fundamental Conflict between Musk's Model and Scientific Research Laws

The deep root of the resignation wave of xAI is the serious opposition between Musk's management style of "extreme high pressure, speed first" and the basic research laws of AI. Musk requires employees to work 80-120 hours per week, implementing a "battle style" R&D rhythm, and making continuous overtime and staying up late to catch up with work the norm. Former CFO resigned after 102 days of employment, openly stating that he experienced a 'hell mode'; Executive Toby Pohlen's farewell letter stated that the priority is to get eight hours of sleep.

This model may have short-term effects at Tesla and SpaceX, but it is incompatible with the free exploration, long-term deep cultivation, and stable environment required for AI research. Core architect Greg Yang fell ill with Lyme disease due to long-term high-intensity work and retired from daily work after experiencing a health breakdown. Top scholars pursue "slow and deep" basic research, while Musk strongly promotes "fast and fierce" commercial iteration. The fundamental opposition of values forces talents to only choose "voting with their feet".

3、 Technical route implosion: dual pressure of power override and directional divergence

The core members resigned on the same day, and the direct trigger was the internal power restructuring and technological roadmap tearing. Before leaving, xAI initiated a large-scale restructuring, with core businesses such as post training and reasoning under the responsibility of John Woo being split, and the discourse power of core technologies being sidelined. Jimmy Ba and other founders have been marginalized, from core executives to "marginalized people," completely losing their dignity and sense of value.

At a deeper level, there is a divergence in the technological roadmap: the team tends to focus on long-term areas such as mathematical reasoning and basic models, while Musk strongly promotes aerospace AI, engineering implementation, and commercial applications, and even denies past achievements, stating that "xAI was not built well from the beginning" and demanding underlying restructuring. This practice of constantly changing orders and ignoring professional opinions has completely disappointed researchers and accelerated collective resignations.

4、 Cost out of control and lack of commercialization: difficult to fill the black hole of burning money

Since its establishment, xAI has been continuously burning money, consuming over $1 billion per month and burning $7.8 billion in the first nine months of 2025. Despite significant investment, commercialization is severely lagging behind: revenue in 2025 is only $500 million, far below OpenAI's level during the same period. The Grok model performed averagely in benchmark tests, lagging behind competitors such as GPT-5.5 and Claude Opus 4, indicating insufficient technological competitiveness.

Meanwhile, xAI overly relies on X platform data and has weak generalization ability; The user base is small, the download volume of independent applications is less than 10 million, and the business prospects are bleak. The core members failed to see technological breakthroughs and commercial returns, coupled with a high-pressure environment and strategic confusion, and ultimately chose to leave.

The resignation of multiple members of xAI on the same day is a concentrated outbreak of multiple crises in strategy, management, technology, and business, exposing the serious limitations of Musk's "personal centralization+high-pressure sprint" model in the field of AI basic research. The core of AI competition is the competition for talent, not the accumulation of computing power or capital. What top scholars need is an environment that respects the laws of scientific research, ensures independent exploration, and balances work and life, rather than endless overtime and power intervention.