DigestAI news desk

Cut through the AI noise.

Generative AI & Models35 min read

JS Denain of Epoch AI says Chinese models lag US labs by about six to eight months

In a podcast with Nathan Lambert, Epoch AI senior researcher Jean‑Stanislas “JS” Denain discussed the state of AI acceleration and the US‑China capability gap. He noted OpenAI’s blog showing a 2× per‑month increase in Codex spending by researchers, but said this does not constitute strong evidence of an imminent self‑sustaining intelligence explosion. Denain expressed uncertainty about how…

1 source

Key points

  • Epoch AI’s JS Denain estimates Chinese public model releases lag US labs by about six to eight months
  • He cites a 2×‑per‑month rise in Codex spending by researchers but says it isn’t solid proof of imminent AI self‑sustaining acceleration
  • Both agree robotics may see industrial gains, but likely over a longer horizon than two‑to‑five years

The conversation turned to robotics, where both participants agreed that large‑scale industrial automation may be far off the two‑to‑five‑year horizon, citing reliability and regulatory challenges. Regarding Chinese models, Denain estimated the public release gap between Chinese labs and OpenAI/Anthropic to be roughly six to eight months, while acknowledging measurement noise and possible over‑fitting in benchmark indices. He contrasted this with other views that place the gap at two to four months for specific models such as Kimi K3 and GLM 5.2.

Model page: Kimi K3 →

Full story from Interconnects · by Nathan LambertOpen source ↗

Debating RSI, the US-China Gap, and Jaggedness with JS Denain of Epoch AI

Interconnects · 22 September 2026

Loading the full article…

This text was published by Interconnects and written by Nathan Lambert. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Generative AI & Models

All →

Related stories