Warehouse jobs were never supposed to come with homework. Genesis AI just raised half a billion dollars on the strength of 100-odd people who put on a sensor glove at work and let it record their hands, and the recording is what the robot learns from. The glove is not a perk. It is a resignation letter the worker writes in advance, in force data, without being told that is what it is.
The company is Paris-based, it closed a $500 million Series A at a $3.5 billion valuation on Tuesday, and it intends to spend the money putting its robot into factories and warehouses. The Next Web logged the round at $500 million against a $3.5 billion valuation. the investor list reads like a semiconductor summit — Nvidia’s NVentures, AMD Ventures, Bpifrance, Samsung Next, Ericsson Ventures, with Eric Schmidt and Xavier Niel among the angels. That is not venture capital chasing a demo. That is the component supply chain buying a seat at the table before the table is built.
The robot is called Eno. eWeek notes the machine has no legs and no head, and this is the detail the pitch deck leans on: two arms on a rotating wheeled base, built for manufacturing, logistics and laboratory floors, where the work is grasping, sorting, packing and labelling rather than climbing stairs. Genesis argues that the hard part of commercial robotics was never locomotion. It was hands.
Which is where the glove comes in. 24/7 Wall St. priced the glove at roughly $300. Workers wear it, with headset cameras, while doing the jobs they already do, and the force data it captures maps one-to-one onto Eno’s hands. The company calls that the unlock. The worker, if anyone asked, might call it something shorter.
Zhou Xian — The Robot Has No Head, and It Does Not Need Your Whole Job Either
Genesis was founded in 2024 by Zhou Xian, who is chief executive, and Théophile Gervet, a former Mistral AI research scientist who serves as president. They met as graduate students in a Carnegie Mellon lab. Genesis shipped a $105 million seed round in July 2025 and has now stacked a Series A on top of it, which is a strange thing to be able to do in a category where almost nobody has shipped a durable product — and the funding is not the interesting number. The interesting number is 30 to 40. That is how many Enos exist, and they are running pilots with customers Genesis will not name.
Xian’s case rests on dexterity, and he makes it in the least romantic way available: the hands are the bottleneck, not the brain. Xian told TechCrunch the hands wear out that a set of Eno hands survives only a few weeks of regular work before it needs replacing. Read that twice. The most expensive part of this robot is a wear item, and the company raising $500 million to deploy thousands of them has not yet solved how long a hand lasts. That is not a reason the plan fails. It is a reason the plan needs volume, and volume is exactly what the round buys.
He is also refreshingly unsentimental about the shape of the thing. Asked why Eno has no head, he told Reuters he has never understood the requirement, since there is nothing alive in there that would need one. He is right, and the same logic is what should worry the people whose hands trained it. If the human form was never the point, then the human was never the point either — the grasp was.
Théophile Gervet — 100 Workers Wore the Gloves, and the Gloves Are the Product
TechCrunch put the glove count above 100. They live in different countries, they work ordinary shifts, and their hands are the reason GENE — the model behind Eno — can fold a box, tape it and peel a label off printer tape without being told the sequence. Genesis also runs a simulator to stress-test what it learns, but simulation did not teach it how much force a cardboard flap needs. A person did, at work, for a wage.
the same demo dropped a box on the second run: the arms packed and labelled a carton cleanly, then fumbled a box and fought the label on the retry. Genesis’s own people called the failure useful — hiccup data. It probably is. It is also the most honest sentence in the whole story, because it tells you what the current state of the art actually is: a system that succeeds on the second attempt when a human is standing nearby to reset the table.
Here is the asymmetry nobody puts in the press release. When a worker trains a colleague, the knowledge stays inside the building and the colleague gets promoted. When a worker trains this, the knowledge leaves the building as weights, gets copied to a contract manufacturer’s line in the United States, and comes back as a unit that does not need a lunch break. One hundred people are not a workforce. They are a dataset with a commute.
Warehouse Jobs — The $300 Glove Is a Severance Package You Pay For
the same piece runs the arithmetic for that worker and it is worth sitting with, because it is the arithmetic every person on a picking floor is about to run privately. Take a 60-year-old who has spent decades picking, packing and loading. Say the job pays $55,000. Say it disappears this winter. Social Security cannot start before 62, which means two years of pay — $110,000 — has to come from somewhere, and the somewhere is severance, savings, unemployment and whatever a 60-year-old body can still get hired to do.
The trap is on the other side of that gap. Full retirement age for anyone born in 1960 or later is 67, and claiming at 62 cuts the monthly check by about 30% permanently. On a $2,800 benefit at 67, that is $1,960 instead — a $10,080 annual shortfall that runs for the rest of a life and widens every year the cost-of-living adjustment lands on a smaller base. The layoff is a two-year problem. The claim is a twenty-year one.
This is what the glove actually costs. Not $300 to Genesis. $300 is the hardware. The invoice goes to the person who wore it, and it is denominated in the years between the last paycheck and the first benefit check.
Eno — Thirty Robots Now, a Contract Manufacturer Next Year
The rollout schedule is the part to watch, because it is unusually concrete for this sector. 30 to 40 of them exist, running pilots with customers the company declines to identify. Genesis is already testing Eno with LG CNS in manufacturing and logistics, a partnership that puts a Korean conglomerate’s name on the risk. A contract manufacturer starts mass production in the United States next year, with a wider rollout expected in 2027. Genesis told the Journal it expects a robotics model that simply works out of the box sometime next year — a claim with a date attached, which makes it falsifiable, which makes it rare.
And the money is not only Genesis. Path Robotics raised $300 million this month in the same category. The capital is arriving faster than the hands are lasting, and the sector’s own honest answer about reliability is that the second attempt is better than the first.
The containment here was never technical. It is contractual, and there is no contract. Nobody signs a form when they put the glove on that says: this recording will be used to make your role unnecessary, and your name will not appear anywhere in the training set.
The 60-Year-Old Picker — Two Years and $110,000 Is the Invoice
What Genesis has built is real and it is impressive on its best day. A pair of arms that can pack and label a carton is a decade of robotics research compressed into a startup’s second year, and the people who made the data possible are not villains or victims in the abstract — they are workers doing a job that now has a second output nobody described to them.
That is the whole story. AI-mekna has spent two years being sold as a tool that helps the person holding it. This round is a $3.5 billion bet that the person holding it is temporary, and the proof of concept is already wearing the glove.
The robot does not need a head. It has a hundred of them, and they all clock in.
Sources: TechCrunch, The Next Web, 24/7 Wall St., eWeek.