The chief scientist at Meta is said to be in early financing talks for his new startup, valued at $585 million. That’s the word from a report Thursday by the Financial Times, which said a funding round in this ballpark would put Yann LeCun’s startup, Advanced Machine Intelligence (AMI) Labs, at an estimated $3 billion valuation pre-launch.
LeCun, who last month said he planned to leave Meta and return to the private sector, has recruited Alexandre LeBrun, founder of French health tech startup Nabla, as CEO of the startup, according to people with knowledge of the matter quoted in the report.
The startup will work on constructing so-called “world models” to create a new generation of superintelligent artificial intelligence systems, the report says. These are A.I. models that supposedly have a working understanding of the physical world, and applications that span everything from robotics to transportation.
AMI Labs will expand on LeCun’s work at Meta by helping develop a new AI architecture that can learn about the world from videos and spatial data (not just language), has persistent memory, and can reason and rearrange elements to plan complex sequences of actions.
LeCun is one of an increasing number of AI scientists taking on entrepreneurial roles, including Mira Murati, the former chief technology officer at OpenAI, who set up Thinking Machines Lab in August and is valued at about $10 billion.
Another OpenAI exec, Co-founder Ilya Sutskever, raised $2 billion for his Safe Superintelligence startup just in April, putting the company’s valuation at $32 billion.
As AiVolution reported last month, LeCun’s decision to form a new company comes amid turmoil in his former company’s artificial intelligence (AI) businesses. In October, Meta announced it would cut 600 jobs from its AI wing to move more nimbly, part of a push to create a singular TBD Lab to merge product and research roles.
Investors have also questioned Meta’s AI spending. The company reported a 32 percent increase in total costs and expenses to $30.7 billion from the third quarter of last year. Capital expenditures totaled $19.4 billion, a record quarterly high. Meta also said it expected full-year spending of $116 billion to $118 billion and capex of $70 billion to $ 72 billion.
That loss of the scientist who helped establish much of Meta’s early credibility in the artificial intelligence realm would only intensify investor skepticism, according to the report.
“Meanwhile, Meta’s AI aspirations are fraught with execution risks,” Aivolution noted. “CEO Mark Zuckerberg has described the company’s mission as ‘building personal superintelligence for everyone,’ while we see the company continuing to invest in compute, data centers, and AI talent with thin near-term monetization.”
