The policy wind is blowing toward humanoid robots and AI infrastructure this time.

Seven ministries including MIIT jointly launched 24 key technology projects via open competition, covering humanoid robots, atomic-scale manufacturing, brain-inspired intelligence, AI base infrastructure, advanced industrial equipment, and high-end consumer technology, with clear technical milestones set for 2028.

This isn't an ordinary policy document. Layered on top of the "2026 Humanoid Robots and Embodied AI Real-World Training Special Action" jointly issued by MIIT and SASAC in June, it essentially writes a check for industry + funding + scenarios + orders — full-pipeline support.

What Does 'Open Competition' Actually Mean?

Many people see "open competition" (揭榜挂帅) as bureaucratic jargon, but it's actually one of the most high-impact mechanisms in China's R&D system.

In plain terms: the government lists critical technologies as "bounties" with clear technical specs, deadlines, and supporting resources; whether you're a state-owned enterprise, private company, research institute, or startup — whoever has the capability can "tear the bounty poster off the wall"; after winning, the state provides funding, scenarios, and order support; final deliverables are assessed against the metrics — meet them and get rewarded, miss them and exit.

The biggest difference from traditional research grants: credentials don't matter — capability does; papers don't matter — results do. Many major national programs used to get captured by large institutions through bureaucratic maneuvering, but under open competition, small teams with hard-core tech have a real shot at big projects.

The weight of these 24 bounties lies in the fact that it's not just R&D money — it's scenarios and orders. The biggest pain point for humanoid robots isn't building prototypes; it's the lack of real-world scenarios after building them — factories won't let you test, logistics won't adopt, consumers won't buy. Through the special action and open competition, the state directly opens up scenarios in SOEs, central enterprises, and government projects — something more valuable than cash.

Why 2028?

Why is the key milestone set for 2028? Because that's the industry's shared window globally.

Tesla Optimus targets 2027-2028 for mass production; Boston Dynamics, acquired by Hyundai, is pushing commercialization; Figure AI raised massive funding and is sprinting; Chinese companies like Unitree, UBTECH, and Xiaomi are accelerating. The industry consensus: around 2028 is the critical window for humanoid robots to move from labs to large-scale commercial use.

China is using its national system to push at this inflection point with a clear goal: avoid repeating the chip industry's "strangled" scenario in humanoid robots — a potentially larger industry than EVs. Core components for humanoids — servos, reducers, controllers, sensors, foundation models — China's supply chain has advantages in motors, batteries, and manufacturing, but still needs to catch up in high-end reducers and base algorithms.

The policy logic is clear: China's EV industry took about a decade to go from following to global leadership; the humanoid robot track aims to compress that timeline. Government sets direction, provides resources, opens markets; companies charge ahead.

Of course, national systems aren't omnipotent. Market competition, product definition, user experience — the government can't do those for companies. But in the zero-to-one phase, state willpower is an enormous thrust.

See you tomorrow.