While Tesla’s Optimus walks slowly around a lab and Figure AI sends single-digit units to BMW for testing, China just put over 1,000 humanoid robots to work in actual logistics centres — and they’re hitting 85 percent of human efficiency.
RobotEra, a Beijing-based company backed by Geely, Alibaba, and Lenovo, has deployed its L7 humanoid robots across more than 10 sorting centres operated by China Post and SF Express. These aren’t demos. They’re not pilots. They’re sorting parcels, scanning packages, and running 24/7 on conveyor lines alongside human workers — and doing it at a rate that makes them commercially viable right now.
The Numbers That Matter
Let’s be specific, because the specificity is the whole point:
- 1,000+ units deployed — transitioning from pilots to batch delivery at thousand-unit scale this quarter (Q2 2026)
- 85% human efficiency — in centres where they’re fully operational. Initial multi-city deployments (Shenzhen, Huzhou, Hangzhou, Hefei, Beijing) started around 70% and climbed
- 24/7 operation — no breaks, no shifts, no sick days
- RMB 50 million single order from SF Express alone
- 200M USD raised, valuation over RMB 10B (~1.4B USD)
- Cumulative orders exceed RMB 500M, half international
The L7 itself is a serious machine: 171cm tall, 65kg, top speed 14.4 km/h, 20kg dual-arm payload, 55 degrees of freedom, 360-degree perception, powered by RobotEra’s ERA-42 embodied AI visual-language-action model.
Why This Is Different From Every Other Humanoid Announcement
Here’s the thing about most humanoid robot news: it’s always “coming soon.” Tesla says 2027 for volume. Figure has a handful of units at BMW. Agility’s Digit is in early Amazon pilots. Apptronik just signed with Mercedes — for pilot testing.
RobotEra isn’t announcing a pilot. They’re announcing deployment at scale with a concrete efficiency metric in live commercial operations. That’s a different category of news entirely.
The 85% figure is particularly meaningful. It’s not a benchmark score. It’s not a controlled demo. It’s the actual throughput rate in a real sorting centre doing real work. At 85% of human efficiency, the economics start to work — especially when you factor in 24/7 operation. A human works 8 hours with breaks. An L7 works 24. Do the maths: even at 85% per hour, you get roughly 2.5x the daily output per robot versus a human on a single shift.
China’s Logistics Beachhead
Logistics sorting is the obvious first use case for humanoids. The tasks are repetitive, the environment is semi-structured, and the labour shortage is acute — China’s ageing demographics make this an existential supply chain problem, not a nice-to-have.
But here’s what makes it a beachhead rather than a destination: once you’ve proven the robot can handle parcel sorting at commercial efficiency, the path to warehouse picking, delivery loading, and eventually last-mile delivery is incremental, not revolutionary. RobotEra is already working with Chinese customs for cross-border inspections.
The L7 isn’t a specialised sorting machine. It’s a general-purpose humanoid that happens to be sorting parcels because that’s where the money and demand are right now. The same platform, with software updates, goes into manufacturing, retail, healthcare.
The NZ Angle
For New Zealand, this is a shot across the bow for NZ Post, Mainfreight, and every freight handler watching automation timelines. The standard assumption has been “humanoids are 3-5 years away from commercial viability.” RobotEra just cut that to zero — in China at least.
NZ logistics companies aren’t deploying humanoids this year. But they should be watching this closely because:
- The timeline just collapsed. If 1,000 units are commercially deployed now, the follow-on scale (10,000 units, 100,000 units) is years, not decades, away
- China’s supply chain automation advantage compounds. Cheaper, faster, 24/7 sorting means Chinese logistics operators gain a structural cost advantage — and NZ imports and exports flow through those same networks
- The talent pipeline matters. NZ’s logistics workforce is already tight. The question isn’t “will robots replace these jobs?” but “will we have the skills to work with the robots when they arrive?”
The Takeaway
The humanoid robot industry has been stuck in an endless cycle of impressive demos and vague deployment promises. RobotEra’s 1,000-unit commercial deployment at 85% efficiency is the first time anyone has crossed the “actually doing the job at scale” threshold — and it’s not a Western company.
Whether RobotEra’s L7 specifically becomes the dominant platform matters less than the signal: the transition from “humanoid robots are coming” to “humanoid robots are here” just happened. In China. In logistics. At commercial scale.
The rest of the world is still running pilots.
🔍 THE BOTTOM LINE: China didn’t just demo humanoid robots — it deployed 1,000 of them commercially at 85% human efficiency in real logistics operations. The “3-5 years away” narrative just expired. The question for everyone else is no longer when humanoids arrive, but how fast they scale once the economics work.
Sources
- RobotEra L7 deployment announcement via X
- SF Express partnership details
- RobotEra ERA-42 embodied AI model specifications
