Ross Nordeen helped build xAI from the ground up. Now he’s gone.
The departure of one of Elon Musk’s co-founders at his artificial intelligence venture, first reported by Business Insider, is the latest in a string of high-profile exits from xAI — a company barely two years old that has already burned through billions of dollars and absorbed an entire social media platform. Nordeen, who previously worked at Musk’s Neuralink before joining xAI at its founding in 2023, left the company in recent weeks, according to people familiar with the matter. He didn’t make a public statement. Neither did xAI.
The silence is telling.
Musk’s AI company has lost at least five of its original twelve co-founders since its inception, a turnover rate that would raise alarms at any startup — let alone one valued at $80 billion after its latest funding round. Among those who’ve departed: Greg Yang, a theoretical AI researcher who left in 2024 to start his own company; Jimmy Ba, a University of Toronto professor who returned to academia; and several others who quietly moved on without fanfare. The pattern suggests something more than the typical startup churn that characterizes early-stage companies in competitive fields.
What makes Nordeen’s exit particularly notable is its timing. It comes as Musk is orchestrating what may be one of the most consequential financial events of 2025 and 2026: the potential initial public offering of SpaceX, his rocket company, which could value the enterprise at $350 billion or more. Musk is simultaneously managing Tesla through a turbulent period of declining sales in key markets, running a federal cost-cutting operation in Washington that has generated enormous political controversy, and attempting to make xAI a credible competitor to OpenAI, Google DeepMind, and Anthropic.
That’s a lot of plates spinning. And some are wobbling.
xAI’s flagship product, the Grok chatbot, is integrated into X, the social media platform Musk acquired in 2022 for $44 billion. The strategy has always been straightforward: use X’s massive data firehose to train AI models, then distribute those models back through X to hundreds of millions of users. In theory, it’s a virtuous cycle. In practice, the execution has been rockier than Musk’s public pronouncements suggest. Grok has faced criticism for generating politically charged and sometimes inaccurate responses, and its market share among AI assistants remains a fraction of what ChatGPT commands.
The $80 billion valuation xAI secured in its most recent funding round — reported widely, including by Business Insider — places it among the most richly valued private companies in the world. But valuations and viability are different things. The AI industry is consuming capital at a staggering rate, and investors are growing more discerning about which companies can convert massive compute expenditures into sustainable revenue. xAI’s Colossus supercomputer cluster in Memphis, Tennessee — reportedly one of the largest single AI training installations ever built — represents an enormous fixed-cost commitment that demands commensurate returns.
Co-founder departures at this stage aren’t just HR issues. They’re signals.
When founding team members leave a company during its hypergrowth phase, it typically means one of three things: disagreements over strategic direction, frustration with management style, or the calculation that personal equity is better preserved by departing before things get harder. In Musk’s companies, there’s often a fourth factor — the sheer physical and psychological demands of working for someone who expects 80-hour weeks as a baseline and treats urgency as a permanent condition. Former employees across Musk’s ventures have described a culture that is exhilarating and exhausting in roughly equal measure.
The SpaceX IPO and Musk’s Expanding Financial Architecture
While xAI churns through co-founders, the bigger financial story may be what’s happening at SpaceX. Musk has been laying groundwork for a potential public offering of his rocket company, a move that would be transformative not just for SpaceX but for Musk’s entire constellation of enterprises. A SpaceX IPO at a $350 billion valuation would make it one of the largest public offerings in history and would crystallize an enormous amount of paper wealth for Musk and early investors.
The implications extend well beyond SpaceX itself. Musk has used cross-pollination between his companies — sharing talent, technology concepts, and occasionally infrastructure — as a strategic advantage for years. A publicly traded SpaceX would introduce new governance constraints and disclosure requirements that could limit some of that flexibility. But it would also generate liquidity that could be redirected into xAI, Neuralink, The Boring Company, or whatever Musk decides needs capital next.
SpaceX’s Starlink satellite internet division has been the primary revenue engine making an IPO feasible. Starlink surpassed $6.6 billion in annual revenue in 2024, according to estimates from financial analysts, and its subscriber base continues to grow rapidly across both consumer and enterprise segments. The unit economics have improved dramatically as SpaceX has scaled its satellite constellation past 6,000 active units in low Earth orbit. For institutional investors who’ve long wanted exposure to SpaceX but couldn’t access private shares, an IPO would be the event they’ve been waiting for.
But here’s the tension. Musk’s attention is fractured in ways it has never been before. His role leading the Department of Government Efficiency — the federal spending review initiative operating under the Trump administration — has consumed significant bandwidth since early 2025. Tesla’s stock, while still trading at elevated multiples relative to traditional automakers, has come under pressure as European sales declined sharply and Chinese competitors like BYD continue to gain ground. The reputational damage from Musk’s political activities has been measurable: Tesla brand perception scores have dropped in multiple surveys, and some institutional investors have grown uncomfortable with the concentration of controversy around a single individual.
Against this backdrop, xAI needs to demonstrate that it can compete technically with organizations that have deeper AI research benches and longer track records. OpenAI, despite its own internal dramas, continues to push the frontier with successive model generations. Google DeepMind has the advantage of virtually unlimited compute resources and decades of accumulated research talent. Anthropic has positioned itself as the safety-conscious alternative and recently closed a massive funding round of its own.
xAI’s pitch to investors has centered on speed. The company went from founding to launching a competitive large language model in roughly a year — a timeline that impressed even skeptics. Musk’s ability to recruit top-tier engineers and researchers by offering equity in a Musk-affiliated venture, combined with the promise of working on frontier AI without the bureaucratic overhead of a large corporation, proved potent in the early days. Whether that recruiting advantage persists as co-founders depart is an open question.
There’s also the data advantage to consider. X generates billions of posts, images, and interactions daily, and xAI has access to that corpus for training purposes. This is a genuine competitive asset — one that other AI companies can’t easily replicate. Reddit signed a $60 million annual deal with Google for training data access. X’s data, funneled directly into a sister company, comes without that kind of external price tag. The question is whether social media data, with all its noise, bias, and toxicity, produces models that are actually better for the applications that generate revenue.
So far, the evidence is mixed.
Grok has developed a niche following among users who appreciate its more irreverent, less filtered personality compared to ChatGPT or Claude. But niche followings don’t justify $80 billion valuations. Enterprise AI — the segment where real money is being made — requires reliability, consistency, and the kind of institutional trust that takes years to build. xAI has made some moves in this direction, but it remains far behind Anthropic and OpenAI in enterprise adoption.
Nordeen’s departure won’t, by itself, alter xAI’s trajectory. No single co-founder exit does. But the cumulative effect of losing nearly half the founding team within two years creates a narrative problem at minimum and a talent retention problem at worst. The engineers and researchers who remain will inevitably wonder whether the departures reflect something they should be concerned about. Prospective hires will ask the same question.
Musk has navigated these dynamics before. Tesla’s early years were marked by executive turnover that would have sunk most companies. SpaceX lost senior engineers regularly during its first decade. In both cases, the companies not only survived but thrived, in large part because Musk’s vision and capital-raising ability proved more durable than any individual contributor’s presence. The question for xAI is whether the AI industry, with its unique combination of talent scarcity and capital intensity, follows the same pattern — or whether the rules are different this time.
The next twelve months will be clarifying. If SpaceX goes public and the offering succeeds, Musk will have an enormous new pool of resources and credibility to deploy across his portfolio. If xAI can ship a model generation that genuinely competes with GPT-5 or whatever Google releases next, the co-founder departures will be footnotes. But if the technical progress stalls while the founding team continues to thin out, the $80 billion valuation will start to look less like a floor and more like a ceiling.
For now, the man who insists he can run six companies simultaneously is about to find out whether the market agrees.