China’s AI boom showcased ahead of WAIC, says Australian AI scientist

Toby Walsh’s remarks, amplified by Xinhua, underscore how far China has come from follower to agenda‑setter in AI hardware, models and humanoid robotics. His framing is telling: Chinese open models like DeepSeek and rapidly cheaper humanoids are not just a domestic story – they’re pitched as infrastructure that will let many more countries experiment with advanced AI at lower cost. That runs directly against a narrative in Washington and Brussels that frontier capability should remain tightly concentrated in a few allied labs.([en.people.cn](https://en.people.cn/n3/2026/0716/c90000-20478351.html))

If China can sustain a pipeline of competitive open‑weight models and sub‑$20k humanoids, the global diffusion of “good enough” AGI‑adjacent capability could happen faster than Western policymakers expect. That has two implications: first, it blunts the soft power of Western labs, because you no longer need an OpenAI or Anthropic contract to get serious capability. Second, it accelerates the need for international safety norms that assume many actors have access, not just a handful of regulated giants.

For frontier labs in the US and Europe, the message is that the competitive threat isn’t just in benchmarks; it’s in the combination of capable models, manufacturing scale and willingness to ship hardware that puts embodied intelligence into factories, hospitals and disaster zones worldwide.