What it means that robots now run faster than Usain Bolt
Humanoid robots are now fast enough to beat the world's greatest sprinter — and that tells you something important about how quickly the physical side of automation is arriving.
- 9.39s
- Tiangong Ultra's 100m time (vs. Bolt's 9.58s)
- 21.50s
- last year's winning 100m time — same robot
- 2,056
- robots at the 2026 games in Beijing
- 2.88m
- high jump cleared by X-Humanoid (human record: 2.45m)
This week in Beijing, a humanoid robot called Tiangong Ultra ran 100 metres in 9.39 seconds — faster than Usain Bolt's world record of 9.58 seconds. Two thousand and fifty-six robots competed, from 666 teams across 16 countries. One year ago, the same robot ran the same race in 21.50 seconds. That is not a small step forward. It is a step change in roughly twelve months.
The run was not clean. Tiangong Ultra slammed into a safety mat at the finish line. A rival robot, Lightning, clocked 9.47 seconds, then collapsed and had to be carried off the track. These machines now move faster than any human alive — but they still lose control in the final metres. The gap between peak speed and reliable, repeatable performance is real, and honest observers say it matters.
But the one-year improvement is the story worth watching. Cutting the 100-metre time by two-thirds in a single development cycle is not incremental progress — it mirrors what happened with AI software a few years ago, when capabilities that seemed years away arrived in months. China has made humanoid robotics a national priority. The games are designed as a public checkpoint: 30 athletic events plus 20 real-world work scenarios covering warehousing, retail, food service, assembly, and household tasks — showing exactly where companies plan to deploy these machines.
If you run or build a business where physical tasks — picking, packing, moving, assembling — are still done by people, the time to think about this is now, not in three years. The robots competing in Beijing this week are not prototypes. The same companies making them are already taking orders for factory and warehouse deployment. The question is not whether physical AI will reach your industry. It is who decides when.