Meta’s big AI division, the one they call the Superintelligence Labs, is getting a major overhaul. It’s being split into four separate parts. The move seems to be an attempt to streamline their work on this, well, incredibly ambitious goal.
According to an internal memo from chief AI officer Alexandr Wang, the idea is to get more focused. He reportedly said that to take superintelligence seriously, they need to organize around the key areas that will actually get them there. It makes a certain kind of sense, I suppose. Big, vague goals often need to be broken down.
What the New Structure Looks Like
The new setup includes a group led by Wang himself, called TBD Lab. Then there’s the existing FAIR team for fundamental research. Another unit will handle products and applied research, led by former GitHub CEO Nat Friedman. And finally, a whole division just for AI infrastructure, which is arguably the backbone of all this.
Meta confirmed the reorganization happened but didn’t add much color to the story. They’re keeping the details pretty close to the chest for now.
A Pattern of Aggressive Moves
This isn’t happening in a vacuum. It follows a pretty intense period where Meta has been snapping up talent from all over. We’re talking about people from OpenAI, Anthropic, Google DeepMind. Back in June, they invested a staggering $14 billion in Scale AI and basically brought its CEO, Wang, over to run the whole show.
The competition for people who can do this work is fierce. OpenAI’s Sam Altman has publicly complained about Meta offering huge packages, we’re talking up to $100 million in some cases, to pull his employees away. It’s a real arms race for brains.
And with a reshuffle like this, there’s always fallout. Rumor has it that some executives might be on their way out. There’s also talk that Meta is considering using outside AI models in its products, which would be a real shift from their usual “build it all ourselves” attitude.
Betting the Company on AI
None of this is cheap. The company’s CFO, Susan Li, recently said capital expenditures could reach $72 billion this year. A huge chunk of that is for AI infrastructure. That’s an almost unimaginable amount of money.
Mark Zuckerberg is clearly all in. He’s been talking non-stop about making superintelligence central to Meta’s future. In a recent post, he wrote about being extremely optimistic that it will help humanity. He also mentioned a new era of personal empowerment, though what that actually means in practice is, well, still pretty vague.
It feels like a huge gamble. They’re spending colossal sums and restructuring their entire AI operation around a concept that still feels like science fiction to a lot of people. Whether this new structure helps them get there faster, or just creates more internal complexity, remains to be seen.
