Yann LeCun, the former chief AI scientist at Meta, has raised $1.03 billion for his startup Advanced Machine Intelligence (AMI), marking one of the largest funding rounds for a company pursuing alternatives to today’s dominant large language models.
The round gives AMI a $3.5 billion pre-money valuation and reflects rising investor interest in alternative approaches to artificial intelligence beyond today’s dominant large language models.
Key Takeaways
• Advanced Machine Intelligence (AMI) raised $1.03 billion at a $3.5 billion pre-money valuation
• The startup is building AI systems based on reasoning, planning and “world models” rather than traditional large language models
• The funding highlights growing competition over the future architecture of artificial intelligence
Advanced Machine Intelligence said the financing was backed by Cathay Innovation, Greycroft, Hiro Capital, HV Capital and Bezos Expeditions, the investment firm linked to Amazon founder Jeff Bezos.
LeCun founded the company after leaving Meta Platforms at the end of 2025. He joined Meta in 2013 to establish Facebook AI Research (FAIR) and became one of the company’s most prominent AI researchers.
AMI is developing AI systems designed to reason, plan and interact with complex environments, rather than relying primarily on predicting the next word or pixel — the technique used by many current AI models.
In an interview with Reuters, LeCun said AI systems based solely on large language models are unlikely to produce broadly capable intelligent agents.
The company’s initial focus will be organisations operating complex systems, including manufacturers, automakers, aerospace companies, biomedical firms and pharmaceutical groups.
These sectors increasingly require AI tools capable of managing complex processes and physical environments, rather than simply generating text or images.
Over time, the technology could also support consumer applications. LeCun said domestic robots, for example, would require a level of common-sense understanding of the physical world to operate effectively inside homes.
He also said discussions are underway with Meta about potential integration of the technology into Ray-Ban Meta smart glasses, which could represent an earlier commercial application.
For business leaders, the funding underscores how quickly the AI investment landscape is evolving, with venture capital increasingly backing competing approaches to building more capable AI systems.
Many companies have already begun deploying generative AI tools based on language models for tasks such as content creation, software development and customer support. However, sectors including manufacturing, aerospace and pharmaceuticals may require AI systems capable of interacting with complex physical environments and industrial processes.
If approaches such as AMI’s “world model” architecture prove viable, they could expand artificial intelligence into areas such as robotics, industrial automation and advanced scientific research, significantly broadening enterprise applications.
LeCun has long argued that scaling up large language models alone will not produce advanced artificial intelligence. His view contrasts with strategies pursued by companies such as OpenAI, Google and Anthropic, which have focused heavily on expanding model size and training data.
At the same time, Meta has been intensifying its AI ambitions. In June 2025 the company reorganised its efforts under a new division called Meta Superintelligence Labs, led by former Scale AI chief executive Alexandr Wang.
Executives and investors will now be watching whether AMI can convert its research approach into commercially viable technology. Early enterprise partnerships, product deployments and platform integrations could determine whether alternative AI architectures reshape the next phase of enterprise AI investment and industry competition.
