On Thursday, prominent AI researcher Yann LeCun confirmed the launch of his new startup, an open secret in technology circles, while clarifying he will not serve as its CEO. The company, named Advanced Machine Intelligence (AMI), has appointed Alex LeBrun as its chief executive. LeBrun is the co-founder and outgoing CEO of Nabla, a well-regarded AI startup focused on medical transcription.

Nabla announced LeBrun's move in a press release, and LeCun verified the details in a short LinkedIn post. "Yes, AMI Labs is my new startup. I’m the Executive Chairman. And Alex LeBrun is transitioning from CEO of Nabla to CEO of AMI Labs," he wrote.

According to a report, AMI Labs is already seeking to raise €500 million (approximately $586 million) at a €3 billion valuation (around $3.5 billion), even before its official launch. In the current climate of substantial venture capital investment in AI companies founded by renowned scientists, this ambitious funding target is not exceptionally unusual. For comparison, the startup Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, secured a $12 billion valuation in its seed round last year—and Murati's industry standing is generally considered less established than LeCun's.

LeCun, a professor at New York University and former VP and Chief AI Scientist at Meta, is a recipient of the prestigious A.M. Turing Award for his contributions to reinforcement learning. The press release also confirmed widespread speculation: AMI Labs is developing "world model" AI. This approach presents an alternative to large language models (LLMs) by aiming to build AI that comprehends its environment to simulate cause-and-effect and predict outcomes through hypothetical scenarios.

Proponents of world models believe this architecture could solve the persistent "hallucination" problem inherent to LLMs, whose non-deterministic, creative nature makes them prone to generating fabricated information. Other leading labs and startups, such as Google DeepMind and Fei-Fei Li's World Labs, are also actively pursuing world model research.

This competitive landscape makes AMI's reported fundraising goals appear particularly bold. When World Labs launched, it raised $230 million at a $1 billion valuation—a significant sum at the time, though that was back in August 2024, a period that now feels distant in the fast-moving AI sector.

Meanwhile, Nabla stated it will begin a search for a new CEO. In the interim, the company will be led by co-founder and COO Delphine Groll, though this arrangement is not yet permanent. Nabla also announced a partnership to integrate AMI's models as they become available. To date, Nabla has raised $120 million from a notable group of investors, including a $70 million Series C round this past June. Its backers include LeCun, Tony Fadell’s Build Collective, HV Capital, Highland Europe, and Cathay Innovation.

LeBrun's background makes him a compelling choice to lead AMI. He has been working on multimodal AI since before the term was widely used, including an early-2010s tenure at Nuance Communications, the original technology power behind Apple's Siri. Following the acquisition of Nuance by Microsoft, LeBrun founded and sold several natural language startups, one to Facebook, and later led Facebook's AI division before founding Nabla in 2018.

In his departure announcement, LeBrun noted that Nabla, a standout in the Paris AI scene, continues to perform strongly. "We have more than tripled our live ARR this year. Up next to $1B," he wrote. He will remain with Nabla as chairman and chief AI scientist. Nabla declined to provide additional comment, and AMI did not immediately respond to a request for comment.