Mark Zuckerberg does not reorganize Meta by internal notes or massive dismissals but via a particularly offensive strategy, supported with billions. After attracting Alexandr Wang, founder of Scale AI, to pilot his efforts to superintelligence, the CEO of Meta finalizes a partial acquisition of the NFDG fund, one of the most influential investors in the IA ecosystem. Mark Zuckerberg wants to build a Dream Team that has the best thinking heads for the next generation.
Governance by Deals
Its objective is not to simply “catch up” Openai or Google, Zuckerberg wants to rebuild its governance AA on a small circle of separate individuals, with an unprecedented financial lever.
Thus Meta has just launched the acquisition of 49 % Scale AI for $ 15 billiona sum that values the company at almost 30 billion. Officially, the agreement aims to “deepen collaboration in data production for AI models”. But for many observers, it is above all a giant acquihireintended to bring Alexandr Wang into the Meta organization chart.
Alexandr Wang, the man at 15 billion
At 28, Alexandr Wang is not a star researcher like Ilya Sutskever or Noam Shazeer and it is not a profile produced in the classic sense either. By recruiting Wang, Zuckerberg offers a captain with a profile capable of federating around IA projects with high capital intensity, with an ambition of civilizational scope. It is an integral recomposition of the Meta IA organization chart which is redrawn.
Friedman and Gross to consolidate the top of the table
Other maneuvers feed the dynamics of Zuckerberg. According to The Information, Meta would also be In advanced negotiations to recruit Nat Friedman and Daniel Grosstwo key figures of the independent IA ecosystem, co -founders of the NFDG fund. Thus Meta would consider partial buyout of the fundvalued several billion, making it possible to acquire a minority participation in startups like Perplexity, The Bot Company, and potentially SAFE Superintelligence, the startup founded by Daniel Gross and Ilya Sutskever. Gross could also leave Safe Superintelligence barely a year after its creation, despite a record lifting of $ 2 billion. The project, supposed to remain “out of the short-term trade pressures”, would thus be short-circuited by a Meta package probably higher than all that the startup could offer, even to its founders.
A reset after reverse
This restructuring comes after a series of setbacks for Meta in AI:
- The chaotic release of Llama 4, with disputed benchmarks and experimental versions better rated than the public version.
- The unexpected rise in Chinese open source models, like those of Deepseek, deemed more efficient and more economical.
- Fatal recruitments despite nine figures, as Sam Altman (OpenAi) has publicly deplored, by evoking bonuses of $ 100 million refused by its best elements.
