Meta Building New Superintelligence Unit

Mark Zuckerberg is personally assembling a new team of AI experts to achieve “superintelligence” or artificial general intelligence ahead of other tech companies. He is reportedly directly involved in the process and aims to hire around 50 people for the new team.

The New York Times reports that Meta has offered compensation packages ranging from seven to nine figures to AI researchers from top competitors. The micromanagement of the recruitment effort is said to be driven, in part, by frustration over the quality of Llama 4 – the latest version of Meta’s open LLM, which failed to impress.

As reported on Bloomberg, the CEO hopes this new team will deliver improvements to its Llama models and build better AI tools. It remains unclear how the group will work alongside Meta’s existing AI teams, though some employees are expected to move over to the new unit.

Unlike most of its rivals that are constantly raising funds to hire top talent and build more data centers, Meta has enough cash flow to invest tens of billions of dollars into AI.

A point Zuckerberg reportedly raised with potential hires is how Meta could fund a multi-gigawatt data center, which would effective give its researchers access to one of the most powerful AI data centers in the world.

Intensifying competition

This news comes even as OpenAI released o3-pro this week, a more powerful version of its o3 AI reasoning model. OpenAI says that expert evaluations and reviewers consistently prefer o3-pro over o3 in “every tested category” and across key domains. Reviewers rated o3-pro higher for clarity, comprehensiveness, instruction-following, and accuracy, according to the company.

O3-pro is available for ChatGPT Pro and Team users starting Tuesday, replacing the o1-pro model. Enterprise and Edu users will get access the following week. O3-pro is also accessible via OpenAI’s developer API.

Image credit: iStock/Derick Hudson