Meta Platforms Inc. is reportedly forming a new lab to develop superintelligence, a term for artificial intelligence models that can outperform humans at many tasks.

The New York Times detailed the initiative today. Alexandr Wang, the co-founder and chief executive of Scale AI Inc., is expected to join the lab. Several other employees of the startup will reportedly also take up roles at the new unit. 

According to The Information, Wang is joining Meta as part of a deal that will see the Facebook parent invest $14.8 billion in Scale AI for a 49% stake. That implies a valuation of about $30 billion. Scale AI was worth $13.8 billion following its most recent funding round last year, which included the participation of Meta, Nvidia Corp. and other tech giants. 

The steep valuation increase is a reflection of the company’s rapid revenue growth. Scale AI reportedly expects to end the year with sales of more than $2 billion, up from $870 million in 2024.

San Francisco-based Scale AI provides a platform through which organizations source training data for their AI projects. The company relies on a network of more than 100,000 professionals to produce that data. One of Scale AI’s main specialties is providing annotations, text snippets that describe the items in an AI training dataset. Such explainers make it easier for AI models to learn patterns.

Scale AI also competes in the AI tooling market. It provides software that companies can use to prevent their AI models from hallucinating and equip them with RAG, or retrieval-augmented generation, features.

It’s unclear what role Wang will hold at Meta’s new superintelligence lab. According to the Wall Street Journal, Meta recently informed employees that it was “finalizing details on its plans for AI leadership.”

Scale AI is not the only AI company from which Meta is recruiting experts for the lab. The Times’ sources said that the Facebook parent has also extended offers to dozens of researchers at OpenAI, Google LLC and other market players. It’s believed that several of the researchers have already agreed to join.

The recruiting drive reportedly began after Meta’s in-house AI development efforts encountered challenges. In April, it previewed a new large language model called Llama 4 Behemoth that features more than 2 trillion parameters. According to the Journal, the company has decided to delay the LLM’s release over concerns about its output quality.

The reported terms of Meta’s Scale AI investment are similar to two recent transactions inked by fellow tech giants. In August, Google hired the founders of Character.AI Inc. and agreed to pay the AI startup $2.7 billion. Earlier, Microsoft Corp. recruited two co-founders of LLM developer Inflection AI Inc. as part of a deal reportedly worth $650 million.

Photo: Wikimedia Commons