Meta AI Brain Drain Leaves Zuckerberg Empty-Handed—So He Buys a Founder, a Lab, and a Second Chance
Just two years ago, Meta was riding high on its LLaMA models—open-source breakthroughs that positioned the company as a counterforce to the secrecy of OpenAI and Google. The Meta AI brain drain, however, is an issue.
But in 2025, Meta’s AI glow has faded. Not because the tech failed, but because its creators walked out the door.
Of the 14 researchers behind the original LLaMA paper, only three remain at Meta. The others have left to power next-gen models at competitors like Anthropic, DeepMind, and Mistral, contributing to the Meta AI brain drain.
Where They Went—And What It Signals
Paris-based startup Mistral, founded by former LLaMA leads Guillaume Lample and Timothée Lacroix, has become the spiritual successor to Meta’s early open-source mission.
Others have landed at Anthropic, Microsoft, DeepMind, and Cohere.
These weren’t junior staff—they were the architects of Meta’s AI credibility.
| Name | Current Role | Company | Left Meta | Time at Meta |
|---|---|---|---|---|
| Naman Goyal | Technical Staff | Thinking Machines | Feb 2025 | 6 yrs, 7 mos |
| Baptiste Rozière | AI Scientist | Mistral | Aug 2024 | 5 yrs, 1 mo |
| Aurélien Rodriguez | Director, Model Training | Cohere | Jul 2024 | 2 yrs, 7 mos |
| Eric Hambro | Technical Staff | Anthropic | Nov 2023 | 3 yrs, 3 mos |
| Timothée Lacroix | Co-founder & CTO | Mistral | Jun 2023 | 8 yrs, 5 mos |
| Armand Joulin | Distinguished Scientist | Google DeepMind | May 2023 | 8 yrs, 8 mos |
| Guillaume Lample | Co-founder & Chief Scientist | Mistral | Early 2023 | 7 yrs |
Meta’s $14.3B Pivot: Buying Into Superintelligence
In June 2025, Meta announced a $14.3 billion investment for a 49% stake in Scale AI, bringing on CEO Alexandr Wang to lead a new AI lab focused on building superintelligence—AI that exceeds human cognitive abilities.
Wang will report directly to Mark Zuckerberg and remain on Scale’s board. Meta says it will share more about this initiative and “the great people joining this team” in the coming weeks.
By not acquiring Scale outright, Meta avoids antitrust scrutiny while effectively absorbing its technical leadership. Scale, known for labeling training data for OpenAI, Google, and Anthropic—mostly through outsourced labor—has also grown into a defense contractor, striking AI deals with the Pentagon and the Qatari government.
Cold Emails, Hot Offers: Zuckerberg Gets Personal
Zuckerberg isn’t waiting on recruiters. According to insiders, he’s been directly messaging AI talent via WhatsApp and cold emails, offering seven- and eight-figure compensation packages. Some have reportedly defected from DeepMind and Google to join Wang’s new team.
This recruiting blitz comes after a disappointing rollout of LLaMA 4, where Meta faced accusations of gaming benchmarks and delaying the long-hyped “Behemoth” variant.
Zuckerberg has now made it clear: two of Meta’s 2025 goals are (1) making Meta AI the most-used personal assistant and (2) building “full general intelligence.”
Internal Conflicts and a Culture in Flux
But inside Meta, not everything is aligned. Longtime AI chief Yann LeCun has resisted the AGI hype, insisting that new architectures are needed. His reluctance may have delayed Meta’s response, allowing rivals to define the frontier.
LeCun’s FAIR team and Ahmad Al-Dahle’s generative AI unit reportedly clash over strategy. Meanwhile, morale has wavered after high-profile exits, overwork, and project resets, exacerbating the Meta AI brain drain.
A Gamble on Scale—and Redemption
For Alexandr Wang, the move is part moonshot, part milestone. At 28, the Scale founder was once the world’s youngest self-made billionaire. Now he’s steering Meta’s most ambitious turnaround yet—building a team that blends academic firepower, defense tech rigor, and global AI scale.
He told Scale employees the Meta deal is “a massive new investment” that validates their mission and will provide payouts to vested shareholders, while keeping Scale independent for future growth.
For Meta, this hybrid structure—hiring Wang, investing in Scale, and launching an AGI lab—follows a now-familiar Big Tech playbook: control the talent, not the whole company.
Can Meta Win Back the AI Narrative?
The irony is stark. Meta once led open-source AI. Now, its founders are pushing the next frontier from outside, while Zuckerberg tries to reframe Meta as an AGI leader.
What comes next—be it a revolutionary model or another setback—will define whether Meta’s new brain trust can restore the momentum it lost.
The Meta AI brain drain is real. But the fight for AI dominance is far from over.
Follow us on WhatsApp, Telegram, Twitter, and Facebook, or subscribe to our weekly newsletter to ensure you don’t miss out on any future updates. Send tips to editorial@techtrendsmedia.co.ke
