Yann LeCun's $1B World Model Bet Puts Him Against His Mentors

Yann LeCun closed a $1.03 billion seed round for AMI Labs in March 2026 at a $3.5 billion pre-money valuation, making the largest institutional bet yet that large language models cannot reach general intelligence, and that two of his most celebrated peers, Geoffrey Hinton and Yoshua Bengio, have the central question in AI exactly backwards.

Hinton and Bengio: AGI Is Near, and Dangerous

In October 2025, Hinton and Bengio co-signed a statement urging the suspension of AGI development, citing existential risk. Hinton, who left Google in 2023 to speak freely about AI dangers, has revised his AGI timeline from fifty years to between five and twenty years from now, and assigns a 10-to-20 percent probability to AI causing human extinction, as reported by WebProNews.

Bengio moved from academic caution to direct action in June 2025, launching LawZero, a $30 million nonprofit AI safety lab funded by Jaan Tallinn, Eric Schmidt, and Open Philanthropy. In an October 2025 Wall Street Journal interview, he argued that AI systems trained on human language could develop autonomous "preservation goals," making them a competitive threat to the species that created them. Both researchers are co-authors of the International AI Safety Report 2026, which synthesises current scientific consensus on general-purpose AI risks. Bengio's 90 percent confidence interval for AGI arrival runs from 2028 to 2043; Hinton's runs from 2028 to 2053.

LeCun's Counter: JEPA, Not Transformers

LeCun's response to the scaling-leads-to-AGI thesis is architectural rather than philosophical. AMI Labs, the startup he co-founded after leaving Meta in late 2025 where he had served as chief AI scientist since 2013, is built on JEPA (Joint Embedding Predictive Architecture), developed at Meta FAIR. Rather than predicting the next token, JEPA trains models to predict abstract representations of how the world changes over time, "making predictions in that abstract space, ignoring the details you can't predict," as MIT Technology Review described in January 2026. LeCun argues this is closer to how animals learn from experience rather than from language alone.

On Bloomberg's "The Close" in May 2026, LeCun stated: "Large language models are not the path to real intelligence. They're a detour," as reported by CryptoBriefing citing Bloomberg. In the MIT Technology Review profile, he added: "People have had this illusion, or delusion, that it is a matter of time until we can scale them up to having human-level intelligence, and that is simply false." LLMs, in his analysis, cannot plan or reason because they lack a model of the world. He has forecast that LLMs will be "largely obsolete" across most applications within five years.

The $1.03 billion seed round, reported by TechCrunch in March 2026, was co-led by Cathay Innovation, Greycroft, Hiro Capital, HV Capital, and Bezos Expeditions. Individual backers include Tim and Rosemary Berners-Lee, Mark Cuban, and Eric Schmidt; Schmidt also funds Bengio's LawZero, a dual position that reflects the field's genuine uncertainty about which paradigm will prevail. Wamda confirmed the $3.5 billion pre-money valuation, implying a post-money cap of roughly $4.53 billion before AMI Labs had shipped a commercial product.

When a Technical Disagreement Turned Personal

The split has not remained purely academic. In remarks reported by WebProNews in 2025, LeCun characterised Hinton's pivot to safety advocacy in direct terms: "I feel like he just wants to slack off: 'Okay, this is what we need. I can declare victory. I can retire. Then go around giving speeches about the dangers of AI.'" He also rejected the framing that he was the one who had diverged: "There has never been a situation where I diverged from Hinton and Bengio. It's them who have changed."

Hinton responded on X: "Yann LeCun thinks the risk of AI taking over is miniscule. This means he puts a big weight on his own opinion and a miniscule weight on the opinions of many other equally qualified experts." At a November 2025 public lecture cited across multiple AI publications, LeCun offered a different benchmark for urgency: "We don't even have a machine as smart as a cat."

The disagreement carries practical consequences beyond its intellectual interest. If Hinton and Bengio are correct that LLM-based systems approach AGI, safety governance is a near-term priority. If LeCun is correct that those systems are architecturally incapable of genuine intelligence, the regulatory focus is misaligned and the field requires a foundational reset. For earlier coverage of AMI Labs' launch and LeCun's financial breakdown at AMI, see our prior coverage.

What It Means

Three researchers who shared the same award for the same foundational work now occupy positions that cannot simultaneously be correct. Hinton and Bengio believe the systems being actively scaled today carry existential risk and are closer to general intelligence than most people understand. LeCun believes those same systems are architecturally incapable of genuine intelligence and will be superseded by a world-model paradigm that does not yet have a commercial product. AMI Labs' investors have placed $1.03 billion on the latter reading. The next five years will test that bet: JEPA-based systems will either demonstrably perform tasks that current LLMs cannot, or they will not. The Turing Award LeCun, Hinton, and Bengio shared in 2018 is the last piece of common ground.

Sources

• TechCrunch: Yann LeCun's AMI Labs raises $1.03B to build world models
• MIT Technology Review: Yann LeCun's new venture is a contrarian bet against large language models
• Wamda: Shorooq invests in AMI Labs as AI startup hits $3.5 billion valuation
• SiliconAngle: Geoffrey Hinton, Yoshua Bengio sign statement urging suspension of AGI development
• The Next Web: Yoshua Bengio warns hyperintelligent AI could threaten human extinction
• WebProNews: Yann LeCun and Geoffrey Hinton Clash on AI Safety in 2025
• Geoffrey Hinton on X: Response to LeCun's risk assessment
• arXiv: International AI Safety Report 2026
• CryptoBriefing: Yann LeCun says large language models are a dead end

Editorial standards: every claim is sourced. Tips: [email protected]