- Dotika
- Posts
- Meta AI: Hiring and firing paradox
Meta AI: Hiring and firing paradox
ALSO : Alibaba launches AI glasses Quark
Alexandre HOTTON
October 23, 2025
News of the day
1. Meta is cutting ~600 AI roles despite recent hiring, aiming to streamline bureaucracy and boost decision-making speed → Read more
2. Alibaba enters the consumer smart glasses market with Quark AI Glasses, starting at $660, powered by its Qwen AI models → Read more
3. Palantir partners with Lumen Technologies in a multi-year, multi-million dollar deal to integrate AI software for enterprise services, leveraging Lumen's infrastructure→ Read more
4. A Chinese creator is gaining viral fame by hand-crafting videos that perfectly mimic the awkwardness and errors of AI-generated content → Read more
Our take
Hi Dotikers!
Imagine spending billions to attract the brightest minds in AI, offering signing bonuses of $100 million, then... laying off 600 people just months later. That's exactly what Meta just did.
This decision reveals an uncomfortable truth: in the AI race, accumulating talent doesn't guarantee success. Meta realized its structure had become too bureaucratic, slowing down decision-making. So the tech giant made a radical choice: keep the new star recruits in its "TBD Labs" unit and part ways with some of its legacy teams.
What's fascinating is the timing. On the same day Meta announced these layoffs, the company signed a $27 billion deal to build a new AI-dedicated data center. The message is clear: Meta isn't backing down on AI, it's simply betting on a smaller, more agile team.
The trigger? The lukewarm reaction to the launch of Llama 4 last April. Zuckerberg understood that having more people around the table doesn't guarantee better ideas. Meanwhile, Meta's PyTorch team is doubling down on tools: they just announced Monarch, which enables training models on 128 GPUs from a simple notebook. And that's not all: the PyTorch Foundation is welcoming Ray, a distributed computing framework that dramatically simplifies scaling from one machine to thousands of GPUs. Fewer people, but more powerful tools: that's the new strategy.
A.
Meme of the day
Reply