Big quote: Yann LeCun isn't buying the current AI boom – or at least not the way it's unfolding. In a recent interview with CNBC, one of the "Godfathers of AI" and AMI Labs founder took aim at both the business model and the underlying technology of today's leading AI companies, suggesting the industry could be headed for a correction. Along the way, he singled out Elon Musk's xAI as a company facing particular trouble.

LeCun, who previously served as Meta's chief AI scientist, didn't mince words. "xAI is kind of a failure, frankly, because the founding team has" departed, he said, pointing to a steady stream of exits over the past year. Several co-founders have left the company since it launched, leaving open questions about how xAI maintains momentum in an increasingly crowded talent market.

That turnover, he argued, will make it harder for Musk to rebuild. "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," LeCun said.

The criticism lands even as xAI has scaled aggressively. Earlier this year, Musk merged the company with SpaceX in a deal that valued the combined operation at $1.25 trillion. Central to that strategy has been heavy investment in computing infrastructure, including the Colossus 1 and Colossus 2 data centers in Memphis. The facilities were built to support large-scale AI training, but they're increasingly doing double duty as a revenue source.

LeCun pointed to that shift as telling. xAI has "huge infrastructure" that it rents out to other companies, he said, "because that's the only way he [Musk] can recoup the cost." Google and Anthropic have both tapped into that capacity – a sign of just how expensive, and in demand, AI compute has become.

Credit: App Economy Insights

Still, the financial strain is hard to miss. In the first quarter, SpaceX's AI segment, which includes xAI, posted a $2.5 billion operating loss. That kind of deficit isn't unique to xAI, but it points to a broader problem: the cost of building and running advanced AI systems remains extremely high, even as companies race to deploy them.

LeCun believes that imbalance is becoming harder to ignore. "The prices are going up of those AI services, but the cost of running them is going down, but not nearly fast enough. And so 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 right?" he said.

If that dynamic continues, he expects a reckoning. "Labs like OpenAI and Anthropic are going to have to increase prices, they're going to have to cut costs, or there's going to be a big bubble explosion."

Beyond the financial concerns, LeCun's critique cuts to the core of how AI is built today. Most leading systems rely on large language models, which excel at generating text and handling tasks like coding and structured reasoning. But he argues the approach has limits – especially when it comes to building systems that can reliably operate in the real world.

His alternative is what he calls "world models," systems designed to understand how environments actually function: capturing cause and effect, physical interactions, and context in a more grounded way. "I personally don't think we're going to have generalized reliable agentic systems until they're based on world models," he said.

That puts him somewhat at odds with the current direction of the industry, where companies like OpenAI and Anthropic are pushing toward more capable AI agents built on LLM foundations. LeCun doesn't dismiss those systems outright, but he questions whether they can scale economically. He says that the expense of operating these high-performing systems remains far above what users are generally willing to pay.

AMI Labs is betting on the alternative path. The company raised about $1.03 billion earlier this year at a reported $3.5 billion pre-money valuation, with a focus on building world model-based systems.

For now, demand for AI systems and infrastructure remains strong. But LeCun's comments reflect a growing unease among some insiders – not just about who wins, but whether the current model of building and funding AI is sustainable at all.