Bosworth states Meta fell behind on frontier AI, specifically with a "Llama gap" where future bets were pulled forward for Llama 3, killing the pipeline for Llama 4. He predicts users will become model-agnostic, caring only that the AI product works, similar to not checking an app's database. Meta's strategy focuses on "personal super intelligence" delivered through glasses and assistants, emphasizing consumer experience and daily habits over raw model intelligence.
Meta's strategic pivot away from model leadership implies that durable value in AI shifts to consumer product integration and distribution, not just raw model capability.