"AI Godfather" Yann LeCun: Musk's xAI "Has Already Failed," AI Labs Face Bubble Burst Risk
0xBroomberg
Turing Award winner Yann LeCun publicly called xAI 'a failure of sorts' at Paris VivaTech — all 11 co-founders have left, and its data centers now rent compute to rivals; he warned the broader AI industry risks a massive bubble burst if labs don't cut costs.
Why does LeCun call xAI a 'failure'?
All 11 co-founders departed over the past year, including Jimmy Ba, Igor Babuschkin, Christian Szegedy, and Tony Wu.
LeCun blamed Musk's management directly: the way he treated the founding team "was not very nice," making it now extremely hard to recruit top AI talent.
This means → xAI's problem isn't a wrong technical bet — it's a hollowed-out core team. Without key researchers, even massive compute is an empty shell.
A $1.25 trillion valuation — where is the money going?
In February, Musk merged SpaceX with xAI in an all-stock deal valuing xAI at $1.25 trillion. Yet the quarter ending March 31 showed the AI unit posted an operating loss of $2.5 billion.
xAI operates two mega-scale data centers in Memphis — Colossus 1 and Colossus 2 — but can only rent out compute to Google and Anthropic to recoup costs.
In plain terms = the company built billion-dollar server farms it can't fully use itself, so it leases capacity to its own competitors. LeCun's summary: "Because that's the only way he can get a return on investment."
What does the bubble warning actually mean?
LeCun's logic is straightforward: AI service prices are rising, but operating costs aren't falling nearly fast enough, and most AI companies are still losing money.
He warned that without cost cuts and pricing discipline, the industry faces a "massive bubble burst."
This means → the problem extends far beyond xAI — the entire sector is burning cash for scale while revenue can't cover costs. In his words: "This can't go on for very long."
Why does he challenge large language models themselves?
LeCun argues that LLMs — large language models that predict the next word by learning text patterns — lack understanding of the physical world and cannot perform causal reasoning or long-term planning.
His position: "I personally don't think we can have general-purpose, reliable agentic systems until they are grounded in world models."
This reflects a fundamental split with the mainstream AI roadmap — and he has put capital behind that conviction. He left Meta in late 2025 to found AMI Labs, which closed a $1.03 billion seed round in March 2026 at a $3.5 billion pre-money valuation, backed by Nvidia, Samsung, and Jeff Bezos.
Frankly, xAI is a failure of sorts, because the founding team has left… Elon is now in a very, very difficult position.
Yann LeCun
Turing Award winner, former Meta Chief AI Scientist, founder of AMI Labs
(June 17, 2026, VivaTech conference, Paris)
Content is for reference only, not financial advice.