Former Meta Platforms, Inc. chief AI scientist and AMI Labs founder Yann LeCun has stated in a recent interview that Elon Musk's artificial intelligence startup, xAI, is a "failure" and cannot compete at the cutting edge of AI. He also shared his views on what could lead to a potential "bubble burst" within the industry.
LeCun's comments have reignited his long-running debate with Musk and cast doubt on the valuations of some of the world's largest AI companies.
Over the past few years, LeCun and Musk have disagreed on numerous topics, including AI and what LeCun has termed the "conspiracy theories" spread by the Tesla CEO on social media. Musk, in turn, has accused LeCun of having "been clueless about AI for a long time."
In his latest interview, LeCun stated, "Frankly, xAI is kind of a failure because the founding team has disbanded."
He added, "Elon is in a very difficult position now, finding it hard to recruit top AI talent because of how he treated the previous team."
Over the past year, several of xAI's co-founders have left the company. In February, Musk merged SpaceX with xAI in a massive deal valuing the AI firm at up to $1.25 trillion.
In the first quarter, SpaceX's AI division, which includes xAI, reported an operating loss of $2.5 billion. Meanwhile, LeCun's AMI Labs raised $1 billion in a funding round in March to build a global model, achieving a pre-money valuation of $3.5 billion.
LeCun noted that xAI possesses "massive infrastructure" and is leasing it out to other companies "because that's the only way he [Musk] can recoup his costs."
The infrastructure LeCun referred to includes xAI's Colossus 1 and Colossus 2 data centers in Memphis, Tennessee. Companies like Google and Anthropic have leased computing power from xAI's data centers.
"I'm not very optimistic about xAI's prospects," LeCun said, further adding that he does not believe xAI can compete with giants like OpenAI and Anthropic.
Potential for a Major Bubble Burst
In recent months, corporate spending on AI has come under intense scrutiny as the technology's costs have far exceeded expectations. It was reported that OpenAI CEO Sam Altman stated in a company livestream this month that firms are currently discussing their AI expenditures, calling the cost of AI a "huge problem."
LeCun remarked, "The price of AI services is going up, and while running costs are coming down, they're not coming down fast enough. So, all these companies are losing money, and essentially, most people's usage is funded by investors. This can't last forever, right?"
He added that labs like OpenAI and Anthropic "either have to raise prices or cut costs, or there will be a huge bubble burst."
LeCun has long been critical of the limitations of large language models (LLMs), which underpin much of current machine learning. He advocates instead for "world models."
Addressing the AI Investment Challenge
LLMs learn language patterns to predict what comes next, making them well-suited for reasoning and programming. LeCun's world models take a different approach, focusing on building an understanding of how the real or simulated world works, involving objects, causality, and behavior.
"I personally believe that only agent systems based on world models can be truly general and reliable," LeCun said.
From Anthropic to OpenAI, AI companies are focusing on AI agents—systems capable of autonomously performing more complex tasks.
LeCun acknowledged that LLMs are very useful in areas like programming or mathematics. However, he pointed out that "the cost of running these high-performance systems is very high compared to what users are willing to pay."
Comments