The departure of a key figure in xAI deals a major blow to Musk's AI ambitions.

Edited by: Jingyu

Elon Musk, who was preparing to merge SpaceX and xAI to create a $1.25 trillion AI behemoth, didn't expect that his huge pie wouldn't be big enough for everyone.

On February 10, 2026, local time, Tony Wu, co-founder of xAI, announced his departure from Elon Musk's AI company.

This marks the second co-founder xAI has lost, following Igor Babuschkin's departure last August . Wu was responsible for AI inference capabilities—a key technology considered by the industry to be the core competitiveness of next-generation AI systems.

It's unusual in Silicon Valley for an AI company that's only been around for a little over two years to lose two co-founders. More importantly, this is happening right now, when AI competition is at its fiercest and talent is at its most scarce.

With founders leaving one after another, can Musk's AI ambitions continue?

01. The reasoning expert leaves.

Tony Wu's role at xAI is far more important than it appears on the surface.

As the technical lead responsible for inference capabilities, Wu reports directly to Musk. In the current stage of AI development, inference capability is considered a crucial bridge between large models like GPT-4 and Claude and true "general artificial intelligence."

Simply put, Wu's responsibility is to enable AI to "think," rather than just "remember and imitate."

Losing Wu at this time would be a fatal blow to xAI.

Tony Wu posted his resignation on X | Image source: X

From a technical perspective, breakthroughs in AI reasoning capabilities require long-term accumulation and continuous iteration. When a reasoning expert leaves, they take with them not only their personal skills, but also a complete set of technical ideas, experimental data, and judgments on future research directions. In the rapidly iterating AI industry, where iterations are measured in months, losing a key technical leader often means at least six months of research and development stagnation.

What's even more worrying is the timing. OpenAI has just released a new code model, achieving a significant breakthrough in AI coding; Anthropic's Claude is also performing increasingly well in inference tasks. Losing a key figure in the inference team at this time could very well cause xAI to fall behind in the most crucial technological race.

One developer bluntly stated on X: "Losing Tony Wu is like Tesla losing its head of battery technology. On the surface, the company is still operating, but its core competitiveness has been impacted."

Tony Wu is not the only one. In fact, in the past year, 5 out of the 12 founding members of xAI have left, a turnover rate of nearly half, comparable to the efficiency when Jack Ma was drastically cutting jobs at Twitter.

Why are top AI talents unwilling to follow Musk's AI vision?

02. The "Side Effects" of Horse-Style Management

The departure of two co-founders in succession forces us to re-examine what exactly happened within xAI.

Although the official reason for his departure has not been disclosed, judging from Musk's management style at Twitter, Tesla, and SpaceX, the problem may not be related to salary, but rather to a clash of management philosophies.

Musk has always been known for his "extremely demanding" management style.

During Twitter's transformation, he had employees sleep in the office and implemented large-scale layoffs using a "go all out or leave" approach. This management style might be effective in manufacturing or relatively mature technology products, but AI research and development requires creative thinking and long-term focus, not simply execution efficiency.

A former OpenAI researcher said in an interview: "AI research has its own rhythm. Sometimes an algorithmic breakthrough requires months of quiet reflection, and sometimes it requires repeated trial and error. If management is always urging 'faster, faster,' it is easy for researchers to feel frustrated."

More importantly, there is a divergence in technological approaches.

Musk has publicly stated that xAI aims to achieve "maximum realism" and "understand the universe." This grand vision is inspiring, but in terms of specific technological implementation, a more pragmatic approach is often required.

When the CEO's vision conflicts with the technology team's judgment, who has the final say?

In traditional AI research institutions, technical experts typically have more say. But in Musk's companies, the final decision-making power often rests with him.

03. The "Talent War" in AI

Looking at xAI's talent drain in a broader context, it is actually a microcosm of the "talent war" in the entire AI industry.

The current AI industry faces a scarcity of top talent comparable to that of nuclear physicists in the last century.

A top AI researcher might receive offers from OpenAI, Anthropic, and Google DeepMind simultaneously, with an annual salary easily exceeding $500,000 and equity value reaching astronomical figures.

In this environment, the key to retaining talent is not just money, but also the platform and atmosphere. Researchers prefer to go to places where they can focus on technology, have a clear research and development path, and are not frequently disturbed by management.

From this perspective, OpenAI and Anthropic do have advantages.

Both companies are led by AI researchers, and the technical teams have significant say in key decisions. In contrast, xAI is more like a "CEO-driven" company—Musk's personal will often overrides the judgment of the technical team.

This is not to say that Musk's approach is wrong, but rather that in the specific industry of AI, this management style may not be the optimal solution.

One Reddit user aptly pointed out: "Musk excels at engineering and productization, but the first half of AI research is more like scientific research, requiring patience and room for trial and error."

The question now is, how much time does xAI have left to fine-tune?

In the "winner-takes-all" game of AI, falling behind by six months can mean being completely eliminated. Losing two co-founders may come at a higher cost than imagined for an AI company still searching for a technological breakthrough.

After all, in this AI arms race, the scarcest resource has never been money, but rather those who truly understand how to make machines "think".