Zuckerberg''s ''AI Headhunting War'' Accelerates
The axis of the AI talent war is again shifting to Meta. Andrew Tulloch — former OpenAI researcher and co-founder of AI startup Thinking Machines Lab — recently joined Meta, causing the Silicon Valley AI talent landscape to shake again. According to WSJ, Meta had attempted to acquire Thinking Machines in August but the deal fell through, with reports of packages up to $1.5 billion offered to Tulloch individually (denied by Meta as "inaccurate and absurd," but Tulloch''s joining materialized nonetheless).
Since 2025, AI personnel from Apple, Google DeepMind, and OpenAI have successively moved to Meta: 3 former Apple AI researchers to Meta Reality Labs; 2 former OpenAI researchers joining Meta''s generative model team; a former DeepMind researcher as lead researcher on the LLaMA project. This reflects Meta''s "AI-First Transition" strategy — shifting from metaverse-centered Reality Labs to "fusion of AI and XR" as its core axis. LLaMA 3 established Meta''s vision of "making AI models the central axis of an open ecosystem."
Meta needs AI for metaverse because "presence" expands through AI-based embodiment: LLaMA models as core engines for conversational agents in the metaverse; Ray-Ban Meta smart glasses'' real-time voice recognition/translation/recommendation based on AI core models; reality space duplication technology (Hyperscape Capture) using AI-based spatial reconstruction. Meta spent $18B+ on AI infrastructure in 2025 Q2 alone — roughly equal to combined OpenAI and Anthropic total investments.
The talent shift represents a philosophical battle: Meta (open AI ecosystem through LLaMA) vs. OpenAI/Apple (AI should be controlled). CB Insights predicts "40%+ of AI talent will be incorporated into large company proprietary ecosystems by 2026." Meta''s investment: $60B+ AI R&D budget planned over next 3 years; LLaMA-based AI Companion, smart glasses, and metaverse integrated services scheduled. AI talent movement signals the direction of the next internet hegemony — Meta aims to be the "Android of robotics," providing software platforms that other manufacturers adopt, just as Google''s Android dominated smartphones.


