Yann LeCun, a leading figure often called the "Godfather of AI," has publicly criticized Elon Musk's xAI, labeling it a "failure," and warned that the broader AI industry risks a "bubble burst" if it does not swiftly cut costs and raise prices.

LeCun, founder of AMI Labs, stated in a CNBC interview that xAI's core founding team has largely departed, making it extremely difficult for Musk to recruit top AI talent now. He argued this leaves xAI unable to compete with leaders like OpenAI and Anthropic in cutting-edge AI. Furthermore, he highlighted that the operational costs for AI services are not falling as rapidly as needed. He pointed out that major labs are currently subsidizing user access with investor capital, a model he views as unsustainable.

LeCun's comments have reignited market skepticism about the valuations of leading AI firms and cast a shadow over the industry's fervent investment climate. His own venture, AMI Labs, secured $1 billion in funding this past March at a pre-money valuation of $3.5 billion. It focuses on developing "world models," which LeCun believes represent the next phase of AI.

Assessing xAI's "Failure": Team Exodus and Talent Challenges

LeCun's critique of xAI centers on its human capital. "xAI is, to be honest, a bit of a failure already because the founding team is gone," he remarked.

Over the past year, several of xAI's co-founders have left the company.

LeCun suggested that Musk's past treatment of teams has significantly diminished his appeal within the elite AI talent pool. "Elon is in a very, very difficult position now to hire top AI people because of the way he treated the previous team... it was not great."

Commercially, xAI also faces pressure.

In February, Musk merged SpaceX with xAI in a deal that valued xAI at a staggering $1.25 trillion. However, financial data shows that the AI business segment (including xAI) under SpaceX reported an operating loss of $2.5 billion for the quarter ending March 31.

LeCun also noted that xAI operates two large data centers in Memphis, Colossus 1 and Colossus 2, and rents out computing power to companies like Google and Anthropic, "because that's the only way he [Musk] can recoup costs."

He expressed that he is "not optimistic" about xAI's future prospects.

The Unsustainable Gap Between Costs and Revenue

LeCun raised deeper concerns about the business models prevalent across the AI industry.

He observed that while prices for AI services are increasing, the pace of decline in operational costs is insufficient, leading to ongoing losses where investors effectively fund user consumption. "That cannot last for very long, right?" he said.

He warned that top labs like OpenAI and Anthropic face three choices: raise prices, cut costs, or face a "bubble burst."

This assessment echoes internal industry worries. According to CNBC, OpenAI CEO Sam Altman stated in a company livestream this month that enterprise clients are increasingly discussing AI spending, calling AI costs "a huge problem."

Diverging Technical Paths: LLM Limits vs. Betting on "World Models"

LeCun maintains reservations about the current dominant AI technical approach. He has long criticized the limitations of large language models (LLMs), which predict the next word by learning language patterns and excel in reasoning and coding. He believes this architecture cannot underpin truly general-purpose AI systems.

The "world model" approach he champions is fundamentally different. It aims to enable AI to develop an understanding of how the real or simulated world operates, encompassing objects, causality, and the logic of actions. "I personally believe that we will not have general-purpose, reliable agent systems until we have them based on world models," LeCun stated.

Currently, major AI companies from Anthropic to OpenAI are heavily investing in AI agents—systems capable of autonomously performing complex tasks. LeCun acknowledges that LLMs have practical value in areas like programming and mathematics but stresses that "the cost to run these systems at that level of performance is much higher than what people are willing to pay." He identifies this mismatch as central to the industry's bubble risk.

The $1 billion funding round completed by AMI Labs in March represents a major bet by LeCun on the world model path and serves as a clear departure from the current mainstream AI narrative.