A one-year forecast for Chinese AI labs to build an OpenAI- or Anthropic-style system
The observer argues that DeepSeek has most of the needed components but needs more data, environments and compute.
TLDR
On October 3, 2026, an observer predicted it would take about a year for Chinese AI efforts to build an OpenAI- or Anthropic-style system and begin narrowing the gap. They argued that DeepSeek has most of the components they associate with “RSI,” but needs more data, environments and compute. They also said DeepSeek had only begun expanding its DSec effort and that onboarding people would take time. In a follow-up, they clarified that writing “Chinese models” rather than labs was a slip, though models would be heavily involved.
Combined views
3.1K
2 Sources, first seen ago
