Who is building AGI?
The labs racing to AGI, and what their leaders say
| Lab | Leader & stated position |
|---|---|
| OpenAI | Sam Altman — confident on the path; superintelligence "a few thousand days" |
| Google DeepMind | Demis Hassabis — ~50% by 2030; cautious on hype |
| Anthropic | Dario Amodei — "powerful AI" possibly 2026–27, with caveats |
| xAI | Elon Musk — most aggressive, by end of 2026 |
| DeepSeek / Qwen (open-weight) | No AGI claim — but ~3–6 months behind the frontier at far lower cost |
The striking thing isn't any one date — it's that the people actually building the systems cluster around "this decade," with the lab leaders (Hassabis, Amodei) more measured than the loudest voices. Full side-by-sides: Aschenbrenner vs Hassabis, Altman vs Musk.
Is any lab actually close to AGI?
Close on capability, not on the finish line. Frontier models reach ~83% on knowledge-work benchmarks (GDPval) and ~80% on agentic coding (SWE-Bench Pro), but no lab has demonstrated a system that autonomously conducts AI research end-to-end — the bar that defines serious AGI. Until one does, "who's building AGI" is really "who's building toward it."
How to track the race objectively
Instead of trusting any lab's framing, watch the falsifiable claims resolve. The AGI-2027 Thesis Tracker distills the whole bet into one auditable score that moves only on evidence — the cleanest way to see whether any of these labs is actually pulling it off. The headline claim resolves by January 1, 2028.
Frequently asked questions
The serious race is run by OpenAI, Google DeepMind, Anthropic, and xAI, with strong open-weight pressure from Chinese labs like DeepSeek and Qwen. Their leaders' public AGI timelines span 2026 to 2030+, but none has demonstrated the autonomous AI research that would settle it.
On capability, the frontier labs (OpenAI, DeepMind, Anthropic, xAI) are neck-and-neck near the top, with open-weight models ~3–6 months behind. But none has crossed the decisive bar — reliable, autonomous, end-to-end work — so 'closest' is a matter of months on capability, not a finish line anyone has reached.
No — mostly that it's close. Lab leaders' public positions range from Musk's aggressive end-of-2026 to Hassabis's ~50% by 2030, with the builders (Hassabis, Amodei) more measured than the loudest voices. None claims AGI has arrived.
Watch the falsifiable, dated claims resolve rather than trusting any lab's framing. The AGI Scorecard's Thesis Tracker distills the shared 2027-ish bet into one auditable score that moves only on evidence.
The live scorecard updates as models ship and verdicts change.
View the live scorecard →Get the weekly AGI progress briefing
Which lab is really closest — by evidence, not press release. Subscribe for the weekly read. Free, no hype.
Subscribe free →