Yann LeCun and Elon Musk: The AI "Godfather" just called Musk's xAI "kind of a failure."The usual case against xAI runs through spreadsheets—the operating losses, the compute bill, the benchmarks where it trails OpenAI and Google. Yann LeCun has a blunter theory, and it has nothing to do with the math. The people who built the company left, nearly all of them, and now the good ones won't come. The AMI Labs founder and former Meta chief AI scientist told CNBC that Elon Musk's lab is "kind of a failure," and the reason he gave was almost personal: Musk, in his telling, treated his own founding team badly enough that the talent pool noticed, and a frontier lab that can't hire is a frontier lab in trouble.LeCun, one of three researchers commonly tagged a "Godfather of AI," didn't reach for diplomacy. xAI is a failure "frankly, because the founding team has departed," he said. The exits are the mechanism that does the damage. "Elon is now in a position that is very, very difficult for him to kind of hire top people in AI, because he's kind of, you know, not behaved in sort of very good ways toward the previous team." In a discipline where a few dozen names actually move the frontier, that's the entire argument compressed into a sentence.The whole founding bench walked, and Musk is hiring from the rubblexAI launched in 2023 with a roster of researchers poached from DeepMind, OpenAI and Google—the kind of names that signal a serious lab to everyone else's recruiters. Three years on, every non-Musk cofounder is gone. Ross Nordeen, the last holdout, walked in March 2026, reportedly cut off from internal systems and dropped from a company group chat on his way out the door.LeCun's argument hangs on what that does to the next hire. Reputations are sticky in a field this small, where the people worth recruiting all know each other and compare notes. Treat the team that built your lab badly enough and the story reaches the candidate before your offer does—so Musk is now trying to staff a frontier lab from a thinner pool, against rivals who haven't given anyone a reason to hesitate.xAI rents out its Colossus data centres to recoup costs, says LeCunHe didn't spare the economics either. xAI built "huge infrastructure," LeCun said, "which he rents to other people, because that's the only way he can recoup the cost." That's the Colossus complex in Memphis, and the tenants aren't fringe operators—Google reportedly pays SpaceX in the neighbourhood of Rs 7,650 crore a month for compute, and Anthropic rents capacity there too. xAI now sits inside SpaceX, which went public last week, and the combined AI unit booked a $2.5 billion operating loss last quarter. Asked point-blank whether xAI could ever catch OpenAI and Anthropic on the frontier, LeCun gave a one-word verdict: "No, I don't."Yann LeCun's and Elon Musk's feud is a years-long AI rivalryThis is well-worn ground for the two of them. Back in 2024, LeCun posted a long thread crediting Musk's cars, rockets and solar work while accusing him of pushing "blatantly false" predictions and amplifying conspiracy theories. Musk's reply has barely changed in tone since—LeCun is "out of touch with AI," and, in one memorable jab, "He thinks if he can't do it, no one can." The feud flared again early this year over humanoid robots, when LeCun argued that no company, Tesla included, knows how to make its machines smart enough to be useful, and Musk waved it off.What sharpens this round is what LeCun is selling. In March he raised roughly $1 billion for AMI Labs, built on the thesis that the large language models OpenAI, Anthropic and xAI all sell are a dead end, and that the future runs through "world models" instead. "I personally don't think we're going to have generalized reliable agentic systems until they're based on world models," he told CNBC. So the takedown of xAI and the warning he tacked on—that the whole industry faces a "big bubble explosion" unless prices rise or costs fall—double neatly as a pitch for his own alternative. The unit-economics problem he describes is real and widely conceded. It just happens to be described, this time, by the person with the most to gain from everyone believing it.