Four years ago, we decided not to fight over hardware. In July 2026, Amazon — one of the retail sector’s operators with the most capital and proprietary technology — has arrived, in practice, at the same conclusion.
We already covered on this blog how Stellantis handed over part of its European production to a Chinese manufacturer, confirming at industrial scale that hardware had stopped being a point of differentiation for anyone who doesn’t manufacture it. Amazon has just written the retail version of that same story, and this time it isn’t about cars or manufacturing: it’s about the magic checkout that the unattended retail industry has spent years selling as the product.
Amazon pulls back the store, licenses the technology
Over the past several fiscal years, Amazon has been closing and scaling back its own operation of its cashier-less formats — Amazon Go, Amazon Fresh — in several markets. The physical store that Amazon ran with its own staff, its own lease, and its own site risk has shrunk.
What hasn’t shrunk is “Just Walk Out,” the computer-vision and sensor technology that lets shoppers leave the store without stopping at a till. That same technology, licensed to third-party operators, is now deployed in more than 360 stores — twice as many as the year before — a figure reported by trade press and not independently verified by us.
The takeaway is uncomfortable for anyone who’s spent years selling the checkout as the business: one of the operators with the most capital and proprietary technology in the world couldn’t make running its own stores add up. And instead of pushing on, it did what a technology company does once it understands where the real margin sits: sell the technology to those who know how to operate, and step out of operations.
The contrast worth looking at
While Amazon retreats from running its own operations, the market keeps rewarding the opposite promise. VenHub — robotic convenience stores — went public on the Nasdaq in January 2026, promising, according to trade press coverage we haven’t independently verified, “hundreds of locations” on the back of more than 1,000 pre-orders. By July 2026 it has five stores open, in Los Angeles.
This isn’t a judgment on VenHub as a company. It’s a data point about the cycle: the stock market keeps rewarding the promise of the checkout at the very moment the player with the most information — and with capital and reputation already on the line — chooses to let it go. When the party that knows the most walks away from operating the very thing another has just gone public promising, it’s worth asking what the first has learned that the second hasn’t yet.
The honest counter-argument
There’s a counter-argument that deserves acknowledgment before it’s rebutted: Amazon hasn’t let go of the good part, it’s kept it. Licensing “Just Walk Out” is scalable and high-margin, with none of the cost of the premises or the staff. What Amazon let go of is the commodity part — the store, the lease, the day-to-day — and that, someone will say, is precisely the part that’s worth the least.
The answer is that this reading confuses which part is hard with which part is worth little. What Amazon actually let go of — picking the right location, restocking, handling incidents overnight, deciding which store to close because it isn’t working — is exactly what no one has managed to make systematically profitable. It’s the uncomfortable part, and that’s precisely why it’s the part where the business is won or lost. The 360 Just Walk Out licenses aren’t going to compete with each other over who has the best camera. They’re going to end up competing to solve the very thing that resisted Amazon: operating well, every day, with no one watching.
The pain isn’t restocking, it’s the 3 a.m. incident
It’s worth not romanticizing the problem. At three in the morning, in a store with no attendant, the problem is almost never restocking — that runs on daytime routes, it’s logistics already solved. The problem is the incident: the product that jams, the payment that fails, the refrigeration unit that breaks down, the customer who’s already paid and doesn’t get what they bought. None of that gets solved by the computer-vision sensor or the sharpest camera. The better the market works, the more hardware gets bought and the more it evens out between operators. Keeping it running twenty-four hours with no one behind the glass — that is pure operations. And it’s exactly what not even Amazon, with all its capital, managed to solve while running things itself.
What we built four years ago
That’s what we decided to build when we designed uNobo Connect: not a checkout, not a sensor, but the 24-hour automated retail operating system that sits above the hardware — manufacturer-agnostic, built for the real-world incident rather than the trade-show demo. The associated network that operates uNobo Connect in Spain today, through Country Partner uRetail S.L., adds up to more than 200 stores, 800 machines, and 180 operators.
That network doesn’t sell the technology to whoever operates it and then walk away. Every 3 a.m. incident has a protocol, a technical support team, and a person who responds, because that’s where it gets decided whether the operator trusts the system again the following month.
Amazon has just learned, with data and capital to spare, what we decided four years ago without needing the lesson: the checkout isn’t the business. The business is everything that has to be operated well so the checkout works without anyone watching it.
Aggregated uNobo Connect network data as of 16 July 2026. Metrics anonymized. Auditable under NDA.
If you operate or want to operate 24-hour automated retail in a European market without an assigned Country Partner, the conversation with whoever wants to be first in their market is different from the one we’ll have once the network is complete. Write to info@bonoboservices.com or use the form at bonoboservices.com.
Román Suárez — Founder, Bonobo Services OÜ