In an unusually forward appraisal of an AI competitor, AMI Labs\u2019 founder and former head of AI at Meta Yann Lecun has lambasted xAI. The latter, now part of Elon Musk\u2019s SpaceX, has been left without its founders, which LeCun sees as caused by Musk\u2019s purportedly bad behavior.<\/p>\n
LeCun has fairly regularly been at odds with Musk, citing the world\u2019s first trillionaire\u2019s social media comments and his stance on AI. Musk\u2019s treatment of former xAI staffers is also making life difficult for him, LeCun believes. \u201cElon is now in a position that is very, very difficult for him to kind of hire top people in AI,\u201d he said. \u201cBecause he\u2019s kind of, you know, not behaved in sort of very good ways toward the previous team.\u201d<\/p>\n
No OpenAI-Anthropic fight expected<\/p>\n
LeCun does not expect xAI to compete with OpenAI or Anthropic at the frontier. Grok, the AI model series built by xAI, has not yet been widely adopted by enterprises, which has been a particularly successful discipline for Anthropic\u2019s Claude LLMs.<\/p>\n
xAI merged with SpaceX in February in a deal that valued the combined company at $1.25 trillion. Rather than becoming a leading model developer, xAI now rents out compute capacity from its Colossus 1 and Colossus 2 data centers in Memphis, Texas to other companies, including Google and Anthropic. Google is paying SpaceX $920 million per month<\/a> for GPU compute capacity under a deal running to 2029. SpaceX\u2019s AI segment posted a $2.5 billion operating loss in Q1 2026.<\/p>\n LeCun\u2019s efforts<\/p>\n Yann LeCun is among a rare group of individuals known as a \u2018godfather\u2019 of AI. It seems media headlines pick whichever figure is at hand to label them as such, but Geoffrey Hinton and Yoshua Bengio tend to be the others. They and LeCun were all recognized with the 2018 Turing Award (known as the effective \u201cNobel Prize of computing<\/a>\u201d as The Verge highlights). Go back in time far enough and Alan Turing is often listed as another, more primordial AI forefather.<\/p>\n At any rate, LeCun has mostly earned this title through his work on Convolutional Neural Networks (CNNs) in the late 1990s. \u201cLeNet\u201d has been used extensively, really just about anywhere structured data can be organized and divided based on patterns, either prescribed or discovered by the CNN.<\/p>\n