The short answer is uncomfortable: no general-purpose humanoid is for sale to a commercial customer today. This page exists to tell you exactly where the demo ends and where what you can actually deploy now begins.
All the confusion in this category comes from mixing what is sold with what is shown in a video. Here they are, separately.
Small and full-size humanoids bought from an open catalog, with SDK and documentation. Used for R&D, universities, corporate labs and demonstrations.
Open catalog priceThey are not designed for productive work. They are for learning, not for operating.Indoor delivery, floor service, reception and guidance. Partial humanoid form on a wheeled base. It is the only human-shaped category that works commercially and at scale today.
Through a distributorThrough a distributor. It is what can actually be deployed in Mexico today.The ones in the videos: bipedal manipulation in plants and warehouses. They are in pilots with selected anchor customers, under direct agreements with the manufacturer.
Not buyableNo price list, no distribution channel, no confirmed route to Mexico.The most promoted use case on social media and the furthest away of all. It requires manipulation in cluttered environments, which is robotics' open problem.
Still far offManufacturers themselves speak of many years before the residential market.If someone offers to quote you one of the industrial humanoids from the videos, they have no way to deliver. There is no open distribution channel for that equipment, and we would rather lose that sale than pretend it exists.
None of these facts helps us commercially. We publish them because a customer who buys with inflated expectations cancels within the year, and that customer costs the category more than it costs us.
Teleoperation —a human guiding or correcting the robot remotely— is the silent norm of today's humanoid deployments. It is not fraud: it is how these systems are trained. The problem is that it is almost never disclosed, and the buyer believes they are watching autonomy.
One company claims zero teleoperation in its demos, but nobody independent has verified it. Another openly acknowledges that its main point of failure was the robot's forearm —a credible fact precisely because it goes against its own interest.
Our stance is simple: if we can't open the source, we don't publish the figure. It is the same rule we used to correct our own home page when we found data that didn't hold up to verification.
Open surfaces, repetitive routes, point-to-point delivery: those already have a robot. Fine manipulation in cluttered environments: not yet. The distinction is made task by task, not area by area.
Autonomous cleaning is the mature category, with equipment in operation and verifiable results. It is the pilot that teaches your organization to work with robots before the hard part arrives.
See industrial cleaning robots →Telecommunications certification, binding tariff classification and a privacy notice if there is a camera. Handled calmly it is paperwork; discovered at customs it costs the project.
It is the real gap in the Mexican market: several leading brands have no spare parts or service in the country. A robot without local support is not an asset, it is a stalled liability waiting for a shipment.
It depends on what for. Research and development humanoids can be bought from a catalog, and anthropomorphic service robots —the ones for floor service and indoor delivery— already operate commercially. What doesn't exist yet is a general-purpose humanoid you can buy and put to work on a productive task. The ones you see in videos from the best-known manufacturers in the United States and China are not for sale to a commercial customer today.
It depends on the category. Research models are sold through an open catalog, with a published price. Anthropomorphic service robots that do operate commercially are quoted through a distributor, with support included. Any figure you see for the headline industrial models is a reference price: the manufacturers themselves say to contact them, which in practice means there is no price list because there is no open sale.
Almost never, and this is the most hidden fact. The head of one of the most transparent manufacturers in the sector acknowledges that opening doors works almost always, but picking up objects still fails often. Another manufacturer told its investors that its units on the factory floor are mainly there for learning, not productive work. And Stanford's AI Index measures a drastic drop between success in simulation and success in real-world tasks. When you see an uncut video, ask how many takes came before.
The head of a US manufacturer with real logistics deployments said this year that the residential market is still many years away, and that humanoids' physical intelligence doesn't exist yet. In bounded, repetitive industrial environments there are pilots working today; in open-ended tasks, there aren't. Our reading is that the useful window for a Mexican company in the short term is specialized service robotics, not general-purpose humanoids.
Because it doesn't suit them. No company in the sector publishes operating fleet size, mean time between failures or downtime. It is the biggest information gap in the category. The huge unit figures that circulate are manufacturing targets or order backlogs, not robots at work. When someone quotes you a fleet number, ask them to separate what was built, what was sold and what is switched on today.
Certification before the telecommunications regulator under the Mexican official standard for wireless equipment, mandatory for any device with WiFi or Bluetooth before it is sold. There is also no specific tariff classification for humanoids, so the right move is to request a binding classification ruling before quoting: it is the difference between a calculated customs cost and a surprise at clearance. And if the robot has a camera, the data protection law applies, with a notice at the point of capture.
Because the decision you do make today isn't which humanoid to buy, it's who you prepare with. Whoever understands today which tasks in their operation can be automated, what can already be solved with service robotics and what is worth waiting for, will reach the humanoid window with the groundwork done. Those who arrive to buy without that prior work will pay for the whole learning curve.
We give you the reading without embellishment: what you are looking for that exists today, what is solved with another technology and what is worth waiting for. If the answer is to wait, we tell you.
No cost · no calls · CDMX · GDL · MTY · remote