Ask anyone
Try it yourself. Next time you’re talking with someone, ask: “What is the hardest part to build in a humanoid robot?”
The answer will come in under three seconds. The brain. Intelligence. Making it think like us. The bolder ones will add consciousness, emotions, maybe walking. Someone will mention the battery, because there is always someone who mentions the battery.
Nobody, absolutely nobody, will say: the hand.
And yet it is the hand. Not the brain, not the soul, not the backflip that promotional videos love so much. The hand. The one that picks up a grape without turning it into jam.
What is easy for us is hell for the machine
Researchers have a name for it: Moravec’s paradox. What seems hard to us, playing chess or calculating, is easy for a machine. What seems trivial to us, picking up an object, is almost impossible.
So we have built machines that beat world chess champions. None of them can yet fold a towel as well as a four-year-old. Humanity can be proud, but only moderately.
Why the hand? Because it concentrates everything machines handle badly. Thousands of sensors under the skin that feel pressure, texture, slipping. A sense of force that goes from egg to hammer without a thought. A reaction within milliseconds when a glass starts to slide. And an endlessly varied real world, where no object is ever quite in the same place.
The brain calculates. The hand has to feel, adjust and correct, constantly, in the middle of chaos. The real wall was never at the top. It was at the bottom, at the fingertips.
Businesses suffer from exactly the same blindness
Replace the robot with a company. The diagnosis stays the same.
The brain is the executive committee. The vision, the strategy, the five-year plan, the forty-slide deck with arrows pointing upward. That is where everyone thinks the game is played. That is where the budgets go, along with the consultants and the lakeside seminars.
The hand is the shop floor. The salesperson facing the customer, the first five seconds of the welcome, the right gesture at the right moment. It is the part people consider simple, almost secondary. And that is precisely where everything is won…
A customer is a grape. Too much pressure and they back off: you forced them, they won’t come back. Too little and they slip away: you let them leave without a solution. The right touch isn’t learned in a spreadsheet. It is learned by feel, like the hand.
That is the whole difference between a salesperson and a merchant. The salesperson performs a gesture. The merchant feels what they are holding, adjusts their grip and keeps hold of it over time. Brilliant strategies rarely fail for lack of intelligence. They fail at the fingertips.
Best on paper versus the one that delivers
In the race for the hand, two philosophies clash. Together they make a strategy lesson in their own right.
Tesla chose nature’s way. On its Optimus robot, the motors leave the fingers and move up into the forearm. They pull on cables, just as our muscles pull on our tendons. Twenty-two degrees of freedom per hand. Anatomically, it is close to perfect. Less heat in the hand, fewer fragile parts exposed to impacts, simpler repairs. The bet on the future.
Except that Tesla was aiming for 5,000 robots in 2025 and built a few hundred. Among the culprits: getting the hand right. The ideal architecture has an unfortunate habit of staying ideal.
Xpeng, for its part, chose pragmatism. Miniature joints housed directly in the hand, twenty-one degrees of freedom, strictly human size. Less elegant anatomically, but more direct, more responsive, simpler to manufacture. And a production line already running in Guangzhou.
Every business knows this duel. The perfect project still being polished, and the imperfect project already bringing in money. In the field, the elegant solution that can’t be rolled out almost always loses to the decent solution that works on Monday morning.
Winning today or winning tomorrow
So who is right? Both. Just not on the same clock.
Xpeng wins the present: the look, the fluidity, the responsiveness, the ability to deliver. Tesla is playing for the future: reliability, durability, maintenance costs once we are talking about millions of units. One sells what works; the other builds what will last.
The real strategic mistake isn’t choosing one or the other. It is ignoring which clock you are playing on. A company that needs cash within six months can’t afford the anatomically perfect bet. A company that wants to last twenty years can’t live on brilliant demos alone.
A good leader doesn’t only ask “what is the best solution?” They ask “the best for when?”
What if the question lies elsewhere?
One question remains, and I am deliberately leaving it open.
Do we really need a human hand? For a large share of useful tasks, a simple two- or three-finger gripper may well be enough. We are generally born with five fingers on each hand. Are we copying the hand out of necessity, or out of fascination?
The question applies to your projects too. How often do we chase the perfect solution, modelled on the ideal, when a good-enough solution would already have done the job?
One thing is certain. The day someone explains that the hardest part is the brain, look at their hands instead. That is often where the real problem hides. And therefore the real value.
Sources for figures: Xpeng announcements on the IRON robot, Technology.org (“Humanoid Robots in 2026”, July 2026), specialist press coverage of Tesla Optimus. Data available in 2026.