During 2025 no AI lab will hold a technical moat: US and Chinese models will converge, and some European models will catch up to roughly the same level.
Partly right ●●●●●
“Technical moat will continue to be elusive. Instead, there will be more convergence on broadly similar models, across both US and China; some systems in Europe will catch up to roughly the same place.”
- Said on
- 1 January 2025 · Annual predictions post on his Substack, Marcus on AI ("Medium confidence" list)
- Source
- 25 AI Predictions for 2025, from Marcus on AI · Marcus on AI (Substack)
- Should come true
- 2025
- How sure they sounded
- Medium confidence
- Categories
- AITechnologyGeopolitics
- Importance
- 3 of 5 · The competitive structure of the AI industry matters to the industry and to national tech policy.
Outcome
Partly right
US and Chinese frontier models did converge in 2025 (DeepSeek, Qwen and others close to US models; Stanford's AI Index reported near parity), so no lab held a lasting moat. But Europe did not catch up: Mistral's December 2025 Large 3 performed roughly where OpenAI's models had been about 18 months earlier and still trailed on the hardest tasks.
- https://hai.stanford.edu/ai-index/2025-ai-index-report
- https://fortune.com/2026/08/05/mistral-europe-ai-sovereignty-us-openai-anthropic-google-deepmind/
History
- 2026-10-17 07:40:27 · created by claude
- 2026-10-17 07:40:27 · categorized by claude
- 2026-10-17 07:40:58 · scored by claude