AI & Algorithm Governance

China's AI policy remains promotion-first, and the newest documents show that push widening from dedicated compute zones into classrooms, robot showpieces, and provincial industrial parks, while the governance track stays flat. Guizhou, the flagship of the East-Data-West-Computing project, now reports 176 EFLOPS of installed capacity—up from 0.84 EFLOPS at the start of the 14th Five-Year Plan, a >200x jump—with smart computing making up over 90% of the total ¹. This is the single most concrete number yet for the national compute build-out.

Guizhou plans to push capacity to 190 EFLOPS in 2026 and targets a 300bn yuan digital-intelligence industry (including a 100bn yuan computing industry) ¹. The province has 50 large data centers now, up from 7 a decade ago. Outside the western hubs, Fujian's IT innovation park has assembled 21 enterprises across three closed-loop chains—trustworthiness computing, industrial robots, and precision electronics ². The pattern is the same as in Harbin's drone cluster ³: AI is being wired into regional manufacturing, not sealed in special zones.

The human-capital layer is now visible at both ends. Hong Kong's Education Bureau ran "AI Leadership" training for about 1,400 principals and vice-principals, framed by the recently published "Digital Education Blueprint" and an "AI Literacy Learning Framework" that teaches students to use AI responsibly ⁴. Mainland bootcamps continue to channel graduates into entry-level AI trainer roles ⁵. This is the state deliberately building the social and educational base for AI adoption.

Diplomatically, Beijing is using robotics as its global calling card. After Chinese humanoid robots broke records at the second World Humanoid Robot Games, a foreign ministry spokesperson used the podium to stress that China's "underlying logic" is open cooperation—"technology interconnection, standards co-construction, market sharing"—and to invite global cooperation on AI and embodied intelligence ⁶. That reinforces the "new three samples" trade narrative of AI, robots, and innovative drugs ⁷.

The market side still provides a caution. Zhongji Innolight's record H1 results were followed by a 7.72% stock drop on cash-flow and customer-concentration worries ⁸. Hong Kong's July merchandise trade shows exports up 50.7% y/y ⁹, but the surge is broad-based and does not address the specific concentration risks inside the AI hardware chain.

Governance has not moved. There are no new national algorithm or large-model rules in this batch. The clearest control mechanism remains the National Healthcare Security Administration's staged pathway for medical AI—from concept to insurance listing to payment ¹⁰. That permissive-but-staged approach is unchanged.

For outside readers: China is now building AI into its core infrastructure, school curriculum, and export pitch simultaneously, while the governance machinery stays in the background. The balance remains promotion-first; the watch item is whether the staged-review system expands to other verticals as AI deployment scales.