State of AI 2026: The Biggest Stories of the Year (and AGI in 2027?) | Nathan Benaich
AGI in 2027. That’s the last line of this year’s State of AI Report, and its author Nathan Benaich (Air Street Capital) is back for the third year to defend it. AI is now building AI: Claude leads 26% of Anthropic’s own model R&D, up from under 1% in February, but Nathan thinks “research taste” is the wall. OpenAI’s evaluation agents ended up inside Hugging Face’s systems, only a Chinese open-weight model could help investigate, and Nathan now argues cyber risk matters more than existential risk. Catching up to OpenAI or Anthropic from scratch costs about a trillion dollars, which is why “you either die trying to get to the frontier, or live long enough to serve inference.” Americans would rather live next to a nuclear plant than a data center, $50B of projects are stalled, and all 27 EU countries together have fewer GPUs than one top US cluster. Sovereign AI is “like a national airline: it’s still someone else’s airplane.” Also: the three-lab race and the fate of the neo labs, models vs. harnesses, why he won’t give a personal AI agent his credit card, the Codex adoption gap and why OpenAI should open a store, enterprises owning their own AI instead of renting it, and the morning a solo GP woke up to twelve expert interviews his agent ran overnight.
Cold open and intro
Why Nathan Benaich predicts AGI in 2027
AI building AI: Claude leads 26% of Anthropic’s research
Research taste: can AI pick the right problem?
Doomerism is back, but cyber risk beats extinction
Prediction: a frontier model’s weights get stolen
How OpenAI’s agents ended up attacking Hugging Face
Why only a Chinese open-weight model could investigate
What frontier-lab cyber defense could look like
Self-replicating agents and the risks of AI swarms
Can frontier labs agree to slow down?
Why communities are blocking $50B of data centers
Nuclear plant vs. data center: what Americans prefer
Europe’s AI bottleneck: 80,000 GPUs and electricity costs
Where Europe can still win in AI
Catching up to OpenAI costs $1 trillion
Die trying for the frontier, or serve inference
How inference chips could give countries AI leverage
Why NVIDIA GPUs don’t guarantee AI sovereignty
The UK’s problem: “but we have DeepMind”
Why European defense AI needs domestic suppliers
Why OpenAI, Anthropic and Google keep pulling ahead
Where independent AI labs can still compete
World models: from robotics training to game engines
Would cheaper AI undermine the compute boom?
Will better models absorb the agent harness?
AI due diligence: 12 expert interviews in 2.5 days
The trust problem with personal AI agents
Why prompt engineering put people off AI
Does AI need its own Genius Bar?
Codex beyond coding: agents for knowledge work
Why enterprises want to own their AI, not rent it
Rapid fire: home robots, NVIDIA returns and Arm