China seen leading AI race as talent shifts to Anthropic and OpenAI

This piece captures two converging trends in the race to AGI: a geopolitical perception shift and an intensifying talent war among frontier labs. Public opinion in most surveyed countries now sees China as the leading AI power, even though US-based firms like OpenAI, Anthropic, Google and Meta still dominate frontier model development. That perception gap matters because it shapes where governments direct subsidies, data center build‑outs and regulatory favor – all of which feed directly into who can afford to train the most capable systems. At the same time, the article underscores how concentrated technical power is in a small cadre of researchers. John Jumper’s move from Google DeepMind to Anthropic, and Noam Shazeer’s from Google to OpenAI, signal that elite scientists are increasingly willing to leave tech giants for labs that promise more focus on frontier models and fewer constraints. For Anthropic and OpenAI, these hires strengthen their claims to be the best place to do cutting‑edge AI work. For Google, repeated high‑profile departures raise questions about whether it can keep the talent density needed to stay at the very front of the capability curve.