LeCun calls xAI a failure as AI losses hit $6.4 billion

Yann LeCun, the AI pioneer behind modern deep learning, says Elon Musk's xAI has failed and the broader industry is heading for a "big bubble explosion."

Yann LeCun, the AI researcher dubbed a "godfather of AI," declared Elon Musk's xAI a failure and warned that OpenAI and Anthropic face a "big bubble explosion" unless they cut costs or raise prices.

"xAI is kind of a failure, frankly, because the founding team has departed," LeCun, founder of AMI Labs and former chief AI scientist at Meta, said in a CNBC interview.

The numbers back his pessimism. SpaceX's AI segment, which absorbed xAI in February, posted a $2.5 billion operating loss in the quarter ended March 31 and a $6.4 billion loss for the full 2025 fiscal year, according to the SpaceX IPO prospectus filed with the SEC. The segment generated $3.2 billion in revenue, but the bulk came from X's advertising business — estimated at roughly $2.26 billion by eMarketer — and subscriptions, not from Grok or AI products. Just $465 million of the revenue came from AI solutions and infrastructure, including $365 million from X and Grok subscriptions and $88 million from data licensing.

The critique strikes at the heart of a $1.75 trillion valuation for SpaceX — which trades at more than 100 times 2025 sales — and raises questions about whether the AI industry's investor-funded growth model can sustain itself. Goldman Sachs, the lead underwriter on the SpaceX IPO, has projected the AI business could generate $322 billion in revenue by 2030, a forecast that assumes the composition of today's revenue changes beyond recognition.

A Talent Exodus and a $2.5 Billion Quarterly Loss

Over the past year, several of xAI's co-founders have left the organization, and LeCun argued that Musk's behavior has made it nearly impossible to recruit top AI talent. "Elon is now in a position that is very, very difficult for him to hire top people in AI, because he has not behaved in very good ways toward the previous team," LeCun said.

The talent drain comes as xAI burns cash at an accelerating rate. Its Colossus 1 and Colossus 2 data centers in Memphis — among the largest AI training facilities in the world — are being rented to competitors including Google and Anthropic, a move LeCun described as Musk's only way to recoup costs. "That's the only way he can recoup the cost," LeCun said.

The Cost Crisis Behind the Bubble Warning

LeCun's broader warning targets the entire AI industry's business model. Enterprise spending on AI has come under scrutiny as the technology proves more expensive to run than expected. OpenAI CEO Sam Altman said this month that AI costs are a "huge issue," with companies increasingly discussing how much they spend on the technology.

"The prices are going up of those AI services, but the cost of running them is going down, but not nearly fast enough," LeCun said. "All of those companies are losing money, and basically, the use for most people is funded by the investors. That can't go on for a very long time."

LeCun said labs like OpenAI and Anthropic face three options: raise prices, cut costs, or face a "big bubble explosion."

The warning carries weight given LeCun's track record. His early work on convolutional neural networks laid the foundation for modern AI, and his current venture, AMI Labs, raised $1 billion in March at a $3.5 billion pre-money valuation to pursue "world models" — an alternative to the large language models that power today's leading AI products. LeCun argues that LLMs, while useful for coding and math, cannot deliver reliable autonomous agents without a fundamental architectural shift. "I personally don't think we're going to have generalized reliable agentic systems until they're based on world models," he said.

For investors, the implications cut across the AI industry. SpaceX trades at more than 100 times sales on the promise that its AI segment transforms from an ad-dependent business into a genuine AI powerhouse. If LeCun is right about the cost structure and the limitations of current technology, the re-rating could be severe. Anthropic and OpenAI, both private, face similar scrutiny on their path to profitability.

This article is for informational purposes only and does not constitute investment advice.