Reid Hoffman Calls xAI 'a Complete Train Wreck'

A Blunt Tour of the AI Boom

Few people are better placed to survey the artificial intelligence industry than Reid Hoffman, and few have been as willing to say what they really think. The LinkedIn co-founder, prolific venture investor and longtime Silicon Valley fixture used a podcast appearance to deliver a string of unsparing verdicts on the companies and personalities driving the AI race, according to Fortune, which reported his remarks on June 24, 2026.

The headline jab was aimed at Elon Musk's xAI. Speaking on the "Pioneers of AI" podcast, Hoffman branded the venture "a complete train wreck" when it comes to the hard work of building foundational models. Coming from an investor who has backed and advised across the sector, the dismissal was striking for its bluntness rather than its mere existence as criticism.

Why xAI Draws His Scorn

Hoffman's skepticism does not float free of evidence. Fortune notes that all 11 of xAI's original co-founders had left the company by May 2026, a remarkable exodus for a firm still trying to establish itself against deeper-pocketed rivals. High founder turnover is often read in venture circles as a signal of internal disorder or strategic drift, and Hoffman's verdict appears to lean on exactly that kind of churn.

The comment also fits a broader pattern. Hoffman and Musk have clashed publicly over politics and the direction of AI, so a harsh assessment of Musk's pet project carries personal as well as analytical weight.

SpaceX, Cursor and the "IAC of AI"

Hoffman next turned to SpaceX, which went public on June 12 and has been acquiring AI assets, including the coding tool Cursor. He rejected the idea that this makes it an AI company at heart. "SpaceX isn't an AI company," he said, likening its approach to a roll-up and quipping that "you could almost think of it as the IAC of AI," a reference to the conglomerate known for assembling internet brands rather than building core technology itself. The implication: bolting on AI products is not the same as developing frontier capability.

Room for Both OpenAI and Anthropic

On the rivalry that dominates AI watercooler talk, Hoffman declined to crown a single victor. He waved away the zero-sum framing of OpenAI versus Anthropic, insisting "there's a lot of room for both of them to win incredibly." His reasoning, consistent with how many investors view fast-growing markets, is that demand is expanding quickly enough to sustain more than one dominant player.

He saved sharper words for Washington. Addressing the U.S. export controls placed on Anthropic's Fable and Mythos models on June 11, Hoffman called the policy "autocratic willy-nilly" and faulted officials for failing to provide a "principled" rationale for the restrictions. The critique reflects a wider industry anxiety that abrupt government action could hamper American competitiveness without clear strategic logic.

A Verdict From the Inside

What gives the remarks their bite is the source. Hoffman is stepping back from Microsoft's board, where Fortune says he will stay through the end of the year, to concentrate on Manas AI, a drug-discovery startup he is building. That positions him as an active operator rather than a retired observer lobbing opinions from the sidelines.

For an industry awash in hype, valuations and breathless launch events, a candid scorecard from one of its most respected insiders is a rare thing. Whether or not his calls prove correct, Hoffman's willingness to separate genuine model-building from marketing and acquisition strategy offers a useful, if uncomfortable, lens on where the AI boom may be overpromising.