#TooBusyToAttend: International Humanoid Forum
Between the Prototype and the Production
Janusz Nowakowski · February 2026 · Biel/Bienne, Switzerland · Switzerland Innovation Park
I asked a robotics researcher where we are with humanoid robots. He paused. "Hard to say." That was the most honest answer at the International Humanoid Forum in Biel last week. And it told me more than any demo.
A Siemens engineer shared a story from a factory visit. A warehouse manager wanted a humanoid robot to take a paper delivery note from the loading dock to the office desk.
The engineer's answer: "An email would be cheaper."
This is where humanoid robotics actually is in 2026. Not the viral clips. The real gap between what impresses on stage and what survives a shift.
Four bridges to production
Dr. Anders Schulz from Siemens laid out four bridges that must be crossed before humanoids enter production:
Safety. You cannot power off a 70 kg humanoid. It falls. Onto people. No safety standard exists for self-balancing machines.
Uptime. Prototypes run 2 to 4 hours. A factory shift runs 8 to 12.
Dexterity. The robot's own hand covers the camera while grasping. It goes blind. Tactile sensing is early.
Cost. Target for 12-month ROI: 20 to 50k EUR. We are not there.
The dance test
A robot can dance techno. Pre-trained, fixed movements. But it cannot pair-dance. Because pair-dancing means sensing another body in real time. Adjusting to pressure. Leading and following through touch. That is not a software update. That is a sensory gap no foundation model will close.
The night shift equation
But here is the turn.
A professor from Bern University of Applied Sciences told me what his Swiss industrial partners say. Typical small-batch, high-mix production: "We do not need speed. We would let the robot work through the night. Slow and steady is enough."
That changes the equation. You do not need a perfect humanoid. You need a good-enough one that shows up for the night shift.
The room
The room in Biel was full of engineers solving exactly this. Siemens, NVIDIA, Disney Research, EPFL, ETH, and dozens of startups. Zero hype talks. Zero politics. Real problems, real progress, real engineering.
These people are building the iPhone before the world knows what a smartphone is.
The seatbelt question
And the seatbelt question hangs over everything. The first cars had no licenses, no road signs. Volvo invented the seatbelt decades later and gave the patent away free. That might never have happened if regulators had demanded it first. Switzerland is trying the other approach. Prove it is safe, then deploy.
Biel is not trying to be Silicon Valley. It is trying to be the place where humanoids become safe enough to work next to people. That is harder. Slower. And probably more important.
Tomorrow: the builders I met on the floor. And why the biggest opportunity in tech right now is not in software.
Physical AI needs infrastructure decisions too.
CloudIndustry advises engineering teams on cloud platform strategy, vendor selection, and building systems that stay independent and maintainable. We have worked with regulated industries in Switzerland and internationally.