Only Elon Musk remains! The 11 co-founders of xAI have all left.

When it comes to stock trading, just look at analysis reports from the Golden Qilin analysts—authoritative, professional, timely, comprehensive. Help you uncover high-potential themes and opportunities!

（Source: XinzhiYuan）

XinzhiYuan report

Editor: Rhino

【XinzhiYuan’s Guide】Eleven co-founders cleared out within three years—leaving only Musk. Behind xAI’s “once-great team disbanding,” lies the most brutal talent war of the AI era and the fracture lines of an empire.

On March 28, Ross Nordeen quietly removed his xAI employee certification badge on the X platform.

He posted a photo—“Touching some grass.”

No long farewell letter, no melodramatic monologue about the next journey, and not even a single sentence thanking his former employer.

Just like that—quietly, he left.

But this seemingly calm move sent a shockwave through Silicon Valley.

Because Ross Nordeen wasn’t a typical employee—he was the last co-founder still staying at xAI besides Musk.

His departure means a cold reality: the “xAI founding supergroup” that Musk personally assembled in 2023—11 co-founders—by the end of March 2026, will have cleared out completely, with not one remaining.

Musk really became a “lone commander.”

The weight of the compute chief

To understand the significance of Nordeen’s exit, you first need to clarify what this 36-year-old Michigan Tech graduate actually did at xAI.

In simple terms: he was Musk’s “right hand.”

In xAI’s organizational setup, Nordeen reported directly to Musk, responsible for coordinating the company’s operational priorities and driving execution across departments.

In the words of insiders—he was the person “who makes things happen.”

If Musk was the general issuing orders, then Nordeen was the chief of staff ensuring every order reached the very last soldier precisely.

But Nordeen’s role went far beyond “operations caretaker.”

His real killer advantage was compute power.

Before joining xAI, Nordeen served as a technical project manager on Tesla’s Autopilot team, specifically responsible for building the data center that trains Tesla’s fully autonomous driving system.

Put another way, he was the person who knew how to turn tens of thousands of GPU cards into a supercomputer that could actually run.

After coming to xAI, he personally led the company’s data center build-out from the ground up—covering everything from underlying hardware to the software stack—laying the physical foundation for xAI’s supercomputing cluster Colossus, which it claims is meant to “understand the true essence of the universe.”

How powerful is Colossus?

More than 200,000 Nvidia H100 GPUs—one of the largest AI training clusters in the world at the time.

And much of this beast’s skeleton was welded together by Nordeen’s team.

In addition, Nordeen’s relationship with Musk wasn’t ordinary—just boss versus subordinate.

According to Walter Isaacson’s Musk biography, Nordeen was old friends with Musk’s cousin, James Musk.

During the major cleanup after Musk acquired Twitter in 2022, Nordeen was one of dozens of engineers who were temporarily pulled from Tesla and SpaceX to coordinate the layoffs on Musk’s orders.

Now that these “inner circle of the inner circle” have all left, you can tell how serious it is.

An 11-person exodus

Nordeen wasn’t the first to leave.

He was the last.

Let’s lay out the full resignation list—this may be the most “glamorous” mass exodus in the history of the AI industry:

In February 2025, Christian Szegedy left. This former Google researcher was the first to send signals, and at the time it didn’t cause much of a stir.

In August 2025, chief engineer Igor Babuschkin left. He came from Google DeepMind and was one of the soul figures behind xAI’s technical architecture.

After that, Kyle Kosic also quietly stepped away.

On February 10, 2026, Tony Wu announced his resignation. He was the head of the reasoning team and one of the most core co-founders on xAI’s operations side. On X, he wrote: “It’s time to begin my next chapter.”

Within 24 hours, Jimmy Ba followed and resigned.

This Toronto University associate professor isn’t an ordinary figure—he was a co-author of the 2014 Adam optimizer paper, which has been cited more than 95,000 times, making it the most cited paper across the entire AI field.

Reports say his departure was related to Musk’s harsh demands for improvements in model performance.

Soon after, Greg Yang announced he would step down due to treatment for Lyme disease.

Toby Pohlen left within a few weeks of being appointed to lead the Macrohard project (xAI’s AI agent initiative).

In mid-March, Zihang Dai and Guodong Zhang departed in succession—reports say Musk blamed Zhang for falling behind on programming tools.

By mid-March, only Manuel Kroiss, who was responsible for pretraining, and Nordeen, the “right-hand operations officer,” remained.

In the final week of March, Kroiss told colleagues that he was about to leave.

On Friday, Nordeen left.

At this point—eleven people. Over three years. All zeroed out.

$250 billion—couldn’t support a research team

If this were just normal employee turnover at a typical startup, there would be no need to make such a big deal.

In startups, the company’s machinery stays running while people cycle in and out—it’s completely normal.

But xAI is different.

First, this wasn’t “employee movement.” It was “wipeout.”

Eleven co-founders, 100% departure rate. This kind of complete reset exodus is extremely rare in Silicon Valley’s startup history.

Even in the most chaotic startups, there’s always at least one or two diehards who stay behind.

Second, these weren’t random hires off the street.

They came from Google DeepMind, OpenAI, Microsoft Research, Google Brain—an aggregation of the very top talent in the AI industry.

It was already an extraordinary achievement that Musk managed to pull these people together in 2023.

But now, he hasn’t kept a single one.

Third—and most crucially—when these people left, xAI’s valuation was $25 billion.

On February 2, 2026, SpaceX acquired xAI via an all-stock transaction. SpaceX was valued at $100 billion, xAI at $25 billion. After the merger, the combined enterprise value was $125 billion, setting the largest corporate acquisition valuation record in history.

One month earlier, Tesla had just invested $2 billion into xAI’s Series E financing.

In other words, this was a company that had just received an astronomical valuation, backed by some of the wealthiest corporations in the world—but its founding team ran away like they were escaping a sinking ship.

Money couldn’t hold people; authority couldn’t hold people.

So where exactly is the problem?

The collapse behind a plain truth

The answer was said by Musk himself.

On March 13—on the eve of the departures of the last few co-founders—Musk posted an extremely rare moment of candor on X: “xAI was not built right first time around, so is being rebuilt from the foundations up.” (xAI didn’t get it right the first time, so it’s being rebuilt from the ground up.)

That same day, at the Abundance conference, he also publicly admitted that xAI’s AI programming tool had no real competitiveness in comparison with Anthropic’s Claude Code and OpenAI’s Codex.

“Grok is currently behind in programming,” he said.

For a company just valued at $25 billion, with the founders saying outright that the product isn’t good and needs to be rebuilt—this is the kind of “legendary moment” you rarely see in CEO remarks in Silicon Valley.

And Tesla’s shareholders likely couldn’t sit still after hearing this: you just used our $2 billion to fund that company, and now the founder himself says it needs to be rebuilt?

No wonder Tesla shareholders have already filed lawsuits, accusing Musk of violating fiduciary duties and diverting shareholders’ money into his own private business.

But Musk’s statement also explains why the founders left.

Imagine you’re a world-class AI scientist. You give up Google DeepMind or OpenAI’s million-dollar salary and a stable environment, and you follow Musk to build from scratch for three years.

Then the boss comes out and says what we did before doesn’t work—we need to tear it down and start over.

What would you think?

Staying means your work from the previous three years is being否定, and you have to start over inside a new organization that has already been merged with SpaceX and whose leadership has undergone a major shakeup.

Leaving, then?

The 2026 AI talent market is the hottest job market in the history of tech—reports say Meta offered top AI researchers a compensation package worth $300 million over four years to keep them.

OpenAI, Google DeepMind, and Anthropic are all aggressively expanding their research teams.

Scientists aren’t soldiers. They have no obligation to keep sailing on a ship that’s being dismantled and reassembled under heavy pressure.

Musk: King of hardware, stuck in software

The clearing out of xAI’s founding team wasn’t an isolated event.

If you place it within Musk’s management blueprint, you’ll see a clear pattern.

After Musk acquired Twitter in 2022, 80% of the company’s employees left or were laid off within a few months—almost all top executives left as well.

Tesla’s executive ranks have also been steadily bleeding talent in recent years, especially after Musk began managing six companies at the same time, spreading his focus severely.

There’s a deeper contradiction here: Musk’s management style—extremely adventurous, extremely high-pressure, extremely fast-paced—wins hands down in the hardware engineering domain.

SpaceX can make rockets reusable; Tesla can sell electric cars and make them #1 globally. That’s all driven by this ironclad execution mindset of “you have to make the impossible happen.”

But AI research is a different story.

The core competitive advantage in AI is people.

Not GPU cards. Not data centers. Not funding—although all of those matter.

Ultimately, it’s the scientists who can put their own names in the first-author position on papers.

These people have countless options, tolerate instability extremely poorly, and their results require time, freedom, and a research culture.

Musk can use willpower to have engineers build the world’s largest supercomputing cluster, Colossus, in 122 days.

But he can’t use willpower to get a group of top scientists to keep producing breakthrough research under high pressure.

xAI’s former founding team joined in part because they were attracted by Musk’s resources and ambition—money and opportunities weren’t lacking.

But once that choice became “continue working in an environment of constant reorganization, leadership turmoil, and where the product is being denied by the founders themselves,” they also had the right to choose to leave.

What else does xAI have?

Saying “xAI is over” would probably be premature.

Musk still has cards in hand.

First is Colossus. The supercomputing cluster with more than 200,000 H100 GPUs remains one of the strongest AI training infrastructure bases in the world.

Second is distribution channels. Grok, xAI’s AI chatbot, is embedded in X’s massive user base.

Although Grok lags behind ChatGPT and Claude in technical performance, the simple fact that it’s “available somewhere” is something that many AI startups dream of.

Third is the halo effect of SpaceX. SpaceX’s planned IPO may value it as high as $1.75 trillion, with the potential to become the largest initial public offering in history.

As a wholly owned subsidiary of SpaceX, xAI can leverage SpaceX’s capital, infrastructure, and engineering talent—resources that are out of reach for most AI companies.

Musk is also actively “replenishing blood.”

He recruited two executives from the AI programming tools company Cursor—Andrew Milich and Jason Ginsberg—who together oversee Cursor’s product engineering.

Musk also personally reviewed job applications that had previously been rejected with his colleague Baris Akis, trying to uncover talent that had been missed—and even publicly apologized, saying “Sorry I didn’t give you the opportunity before.”

But the problem is: can the newcomers fill the holes left by those 11 people?

The core competitive strength of a research institution is never just the sum of individual abilities. It’s the tacit understanding, culture, and methodology formed from long-term collaboration.

Starting from 2023, xAI’s founding team spent three years refining their work together. Their coordination, their understanding of the Grok architecture, and their grasp of training strategies—these things can’t be restored quickly just by hiring a few new people.

And besides, none of xAI’s competitors are waiting for Musk.

OpenAI is sprinting. Anthropic is accelerating. Google DeepMind is expanding its ranks. Meta is throwing money at it.

Competition among large language models has entered a “heat level measured in months” phase, and xAI doesn’t have time to rebuild calmly.

A story about “people”

Looking back, the zeroing out of xAI’s founding team is not fundamentally a story about technology, nor about capital.

It’s a story about “people.”

Musk has always believed that as long as there is enough funding, enough compute power, and enough determination, there’s nothing that can’t be done.

The success of SpaceX and Tesla repeatedly validated this belief.

But the AI field taught him a different lesson: in this industry, the scarcest resources aren’t GPUs or dollars—it’s those brains that can create breakthroughs.

These brains have their own will, their own choice, and buyers everywhere competing to recruit them.

Eleven co-founders cast votes with their feet, leaving cleanly and completely.

Where they go—whether to join OpenAI, Anthropic, Google DeepMind, or to start their own companies—will reshape the AI industry’s talent landscape again.

And what about Musk?

He still sits there, beside the world’s largest supercomputing cluster, with a $25 billion valuation, and a grand plan he promises to “rebuild from the ground up.”

Throughout history, there are no shortages of cases where people succeed after tearing everything down and starting over. Tesla’s early days also went through near-death experiences, and all of SpaceX’s first three rocket launches failed.

What Musk is least lacking in is the courage to stand back up on ruins.

But this time, he faces not physical laws, but people’s hearts.

And people’s hearts are the hardest thing in this universe to reverse-engineer.

The story of xAI is far from over.

But the story of the founding team, on the evening of March 28, 2026, in Ross Nordeen’s photo, officially came to an end.

References:

Massive information, precise interpretation—available in the Sina Finance app