One of the most respected names in computer science has given a blunt assessment of Elon Musk’s company in the field of artificial intelligence, suggesting it has fallen badly behind its rivals. He has also cautioned that several of the world’s leading firms in this area could be heading towards serious financial trouble unless they change course.
Yann LeCun spent years as the top scientist at Meta, helping shape the social media giant’s approach to building computer systems that can write, reason and respond like a person. He has since left to start his own company, AMI Labs. He said Mr Musk’s firm, xAI, was unlikely to compete on equal footing with better-known names such as OpenAI and Anthropic.
According to Mr LeCun, the company is “kind of a failure”, a verdict he linked to the loss of much of the team that originally built it. Several senior figures who helped found xAI have left over the past year, and Mr LeCun suggested this had made life harder for Mr Musk when it comes to attracting new talent. He pointed to the way Mr Musk had treated former colleagues, arguing it would put off other skilled researchers from wanting to work for him in future.
The comments are the latest in a long-running disagreement between the two men, who have clashed publicly before. Mr LeCun has previously accused Mr Musk of spreading conspiracy theories online, while Mr Musk has dismissed his rival as out of touch with recent developments in the field.
xAI has been through a period of considerable change. Earlier this year, Mr Musk combined the business with his rocket company, SpaceX, in a deal that valued the merged operation at $1.25 trillion. Despite that enormous figure, the part of SpaceX that includes xAI reported a $2.5 billion operating loss for the first three months of this year. Mr LeCun claimed the company rents out its large stock of computing hardware to other businesses simply to help cover its costs.
His own venture has taken a markedly smaller path. AMI Labs brought in $1 billion from investors in March, putting its value at $3.5 billion before that money was added — a fraction of the sums attached to the biggest players in the industry, though Mr LeCun argues his firm is pursuing a more sustainable route to building useful machines.
Beyond his criticism of Mr Musk personally, Mr LeCun raised a wider worry about the direction the entire industry is heading in.
Many companies, he said, are spending far more on running their systems than they currently earn from selling access to them, with everyday users largely subsidised by outside investors rather than by paying customers. Unless firms either raise their prices or bring down their running costs, he warned, the sector risks “a big bubble explosion.”
That concern follows separate remarks attributed to OpenAI’s chief executive, Sam Altman, who is reported to have told staff recently that the expense of running modern AI tools had become a serious problem for the business.
Mr LeCun also used the conversation to explain why he believes the dominant approach in the industry, building systems that predict the next word in a sense has limits.
He favours an alternative method, sometimes called a “world model,” which tries to give machines an understanding of how the physical world actually works: objects, cause and effect, and the likely results of different actions. Without that kind of grounding, he argued, computers will struggle to carry out complicated tasks reliably on their own, even though today’s tools remain genuinely useful for jobs such as writing software or working through mathematical problems.
The exchange comes at an uneasy moment for the wider industry, with spending on artificial intelligence tools by businesses around the world coming under growing scrutiny, as many discover that running the technology costs considerably more than they had originally budgeted for.
