They Filmed the Wrong One

Source: LinkedIn video repost — Jean-Claude Tshipama, “La technologie au cœur de la vie quotidienne 🤔,” June 2026. Corroborated against CES 2026 floor reporting and the Qualia fake-robot admission, June 9, 2026.

Watch where the phones point.

A man stands inside a roped-off square at some technology expo, holding his baby. Across from him, a white humanoid robot leans in, articulated and gleaming, and the baby does the only sensible thing a baby can do when a machine reaches for its face: it screams. Behind them, two and three deep, the crowd lifts its phones. Every lens is aimed at the robot. Not one is aimed at the child.

I have watched the clip more times than I want to admit, and that is the detail I cannot get past. Forty-four likes. Five reposts. A caption in French about technology at the heart of everyday life, punctuated with a thinking-face emoji, as if the man who posted it suspected he was looking at the future and wanted credit for noticing. He was looking at the future. He just had the camera turned the wrong way.

Because there are two kinds of capital in that square, and the room has its back to the more valuable one.

The robot is crystallized labor. Every engineer-hour, every actuator someone designed, every line of training data — all of it spent already, hardened into a finished artifact that can stand in a roped square and pose. The baby is the opposite pole. Labor in formation. Capability with not one of its formations begun. And the money, the attention, the awe in that room flow entirely toward the crystal and not one cent toward the formation. Hold that. We will come back to it more than once, because it is the whole thing.

Let me give you the crystal’s price, because the numbers are not small and you should feel their weight.

Figure AI, the American humanoid maker, carries a private valuation of thirty-nine billion dollars — the highest in the field — on more than two and a half billion raised. Its robot spent eleven months on a BMW line in Spartanburg and helped build north of thirty thousand cars. In February, Apptronik raised five hundred and twenty million dollars at a five-billion valuation, with Google, Mercedes-Benz, and the sovereign wealth fund of Qatar all writing checks; its chief executive calls humanoid robotics the space race of our time. Tesla runs something like a thousand Optimus units in its own factories, and Elon Musk has said eighty percent of Tesla’s eventual worth will come from the robot — though on the last earnings call he quietly conceded it is still early-stage research. A unit that cost a quarter of a million dollars to prototype in 2020 can be built for under thirty thousand at scale today. The crystal gets cheaper by the year.

Every figure I just gave you is the price of the finished good. None of it — not one dollar — is the price of the formation.

And here is the part that should give the reposting banker pause. The crystal is not even as finished as the footage suggests. At this year’s CES, the show that exists to make machines look inevitable, the verdict from the floor was merciless: the humanoids were excruciatingly slow, one laundry robot needing roughly two minutes to carry a single garment to the washer, and someone cut the failures into a viral reel. Six days before I sat down to write this, a robotics company partnered with Google’s own DeepMind admitted its viral humanoid video was fake — no real robot at all. The founder’s reply when caught: “Got your attention tho.” A French caption, a thinking emoji, five reposts. The distribution layer is doing the work the hardware still cannot.

So much for the crystal. Now the seed.

In the system I have spent two years building, a human being becomes capable by passing through four formations. Call them four gates. The first is the home — the years of attachment and language and steadiness before a child has done anything the world would pay for. The second is school. The third is the trade: the apprenticeship, the Ausbildung, the slow accumulation of hours at a real bench under a real master. The fourth is early career, the first five years of paid work where everything formed finally compounds. Roughly twenty-five thousand hours from the first gate to a fully formed adult who can do work no machine in that roped square can touch.

The baby in the clip stands at the very threshold. Formation Zero. Not one gate passed. And of the four, the first is the one no balance sheet in this essay is funding — so let me tell you what it actually costs, because I paid it twice.

One of my sons I walked all night. Colic — the kind with no cause you can fix and no end you can schedule. The only thing that touched it was motion: the same slow circuit of the same dark rooms, the same turn at the same wall, until the screaming wore down to hiccups and the hiccups to breathing. I did not soothe him so much as outlast him. Most nights he won. I walked anyway.

The other I held sitting up. His Eustachian tubes stayed infected through his first years, and lying flat put pressure where it hurt, so the arrangement was simple: upright, against my chest, all night, because the angle was the medicine and I was the angle. I did not sleep sitting up because it worked. I slept sitting up because he could not otherwise.

A robot could do both. The circuit, the wall, the turn — teachable in an afternoon. And once one unit learns the precise posture that keeps the pressure off an inflamed ear, five hundred units know it by morning; that is not my nightmare, that is their brochure. But the motion was never the formation. The formation was the cost. I lost the night, I could have done otherwise, and I did not — and that, not the holding, was what his nervous system was actually reading. A machine holds and loses nothing, so it teaches nothing, because nothing was spent. And a baby does not attach to a caregiver.