Robots in the Classroom: How School Robotics Went Mandatory in 2026

School robotics stopped being an after-school club this year. In Beijing, more than 1,400 primary and secondary schools are running compulsory AI lessons for every grade, and the hardware in those rooms is doing the teaching work: block-programming platforms, image-recognition robots, robotic arms and, in at least one school, a full-size humanoid. China’s Ministry of Education wants an introductory AI course in every elementary and middle school in the country from the 2026-27 academic year. The United Arab Emirates already made AI compulsory from kindergarten to grade 12, and India starts at class 3 next year.

That is the boom. The honest version of it is smaller than the headlines suggest — eight to ten class hours a year is roughly one lesson a month, and the best available research says a robot in the room only helps when a human teacher is standing next to it. But the direction is unambiguous, and it has consequences beyond education: it is a state-scale demand signal for cheap, safe, serviceable robots, and it is quietly turning a generation of eleven-year-olds into people who are not impressed by a machine that walks.

What Beijing actually mandated

The anchor fact comes from China Daily, and it is worth quoting precisely: this semester, over 1,400 primary and secondary schools in the city will provide no fewer than eight class hours of AI education covering all grade levels from primary through senior high schools. The programme started with the September 2025 semester and follows two guidelines that between them define the positioning, content and instructional pathways for AI in the city’s classrooms. Beijing is the first provincial-level region in China to do this comprehensively, which in a country where policy travels city by city is the part that matters.

Eight hours is not a lot. What makes it interesting for anyone who follows robots is how those hours are spent. China Daily lists the classroom kit as block-based programming platforms, iFlytek robots with image recognition capabilities, and large language models that help students generate executable code from Chinese instructions. In other words, the abstraction is deliberately physical: the child writes an instruction, and something in the room moves.

At Haidian District Experimental Primary School, grades one to six get progressive, project-based lessons. Shi Yuan, head of the school’s information centre, framed the goal in a way that any hardware reviewer will recognise: “We turn abstract concepts into tangible projects. It’s not about making every child an expert, but about helping each one understand and interact confidently with AI.”

Guangqumen Middle School is where it gets closest to what this site normally covers. China Daily describes students there operating a robotic arm with a smart headband during an AI class, and one student who was put in charge of operating a humanoid robot at the school sports meeting — walking it forward and waving to the students in the stands. Seventh graders take a semester-long information technology course; eighth and ninth graders design “intelligent study companions” using generative AI; the high school runs an AI class developed with partners including the Chinese Academy of Sciences and Huawei that reaches into embodied intelligence, brain-computer interfaces and autonomous driving. Pei Yi, the school’s director of talent cultivation, put the thesis plainly: “In the future, human-machine collaboration will be inevitable. Students may not all become AI specialists, but they must learn how to use AI effectively.”

A twelve-year-old driving a robotic arm with an EEG headband is a better indicator of where consumer robotics is going than most trade-show demos. It is also, bluntly, a procurement pipeline: 1,400 schools that now need hardware that is cheap enough to buy in bulk, safe enough to put in front of children and robust enough to survive them.

It is not just Beijing

Hangzhou — home to DeepSeek and to Unitree — made AI classes compulsory in its primary and secondary schools with a floor of at least ten class hours a year, going beyond the capital’s minimum. The South China Morning Post describes a curriculum that starts with everyday AI applications and basic devices in the first two years of primary school and climbs to data, algorithms and AI project development by high school, delivered flexibly: an intensive week, or folded into information technology and science lessons, or run through after-school programmes.

Above that sits the national layer. China’s Ministry of Education issued guidance on strengthening AI education in primary and secondary schools in December 2024, followed by further guidelines in May 2025. The requirement that matters for the next twelve months: every elementary and middle school is to offer an introductory AI course from the 2026-27 academic year, with the goal of making AI coursework basically universal at K-12 level by 2030. Primary students learn basic concepts through hands-on activities with robots and simple programming; secondary students move to data literacy, model training fundamentals and AI ethics.

The guardrails are stricter than most Western coverage acknowledges. Primary pupils are prohibited from independently using open-ended content generation tools. Teachers are barred from using generative AI as a substitute for their primary teaching role, from relying on AI to answer students’ questions directly, and from using AI-generated content to evaluate students. Feeding sensitive data such as personal information or exam questions into AI tools is strictly forbidden. There is also no standardised national AI testing planned before 2030 — a deliberate choice to keep the subject exploratory rather than exam-driven.

The pipeline above the school gate

The K-12 mandate is the visible half. The other half is a vocational and university build-out aimed squarely at staffing the humanoid industry. More than 40 institutions — technical colleges, universities and specialised training centres — are running humanoid-focused programmes under standardised curriculum guidelines, training engineers in whole-body control, sim-to-real transfer and dexterous manipulation rather than general robotics. Seven universities, among them Shanghai Jiao Tong and Zhejiang, have moved to offer “embodied intelligence” as a major in its own right.

There is even a school for the robots. In Shanghai’s Zhangjiang district, a humanoid training centre of around 5,000 square metres was set to reach full operation in July 2026, hosting more than 100 robot models from over a dozen companies and using the data they generate to train the next generation of machines faster. That is the same logic as the supply chain advantage we have written about before, applied to training data instead of parts.

The American route: incentives, not mandates

The United States is pursuing the same destination through a different mechanism. Executive Order 14277, “Advancing Artificial Intelligence Education for American Youth,” was signed on 23 April 2025. It creates a White House Task Force on AI Education, directs the Department of Education, the National Science Foundation and the Department of Labor to build K-12 resources and teacher training, and launches a nationwide Presidential AI Challenge, with agencies working to a 22 July 2026 deadline for that launch. There is no federal hour count, because there constitutionally cannot be one — US curriculum is set district by district.

Which is why the American version of this story looks like a single small district doing something unusual. The Salamanca City Central School District in western New York became one of the first in the country to put a humanoid in a classroom, using Realbotix’s M-Series robot with the company’s Optio AI platform: multilingual homework help, avatars trained on district-approved curriculum, and support outside school hours. The pilot launched in June 2026, with an expansion to roughly 500 high school students planned for the autumn. The framing from the district is careful — the robot supports the teacher, it does not replace one. Realbotix CEO Andrew Kiguel described it as “moving beyond lab demonstrations and pilots to deliver real, embodied AI directly into classrooms.”

One district with one robot is not a national programme. It is, though, the first time a US public school has had to write a purchase order for a humanoid, and purchase orders are how categories start.

Who has mandated what

Does a robot in the room actually teach anything?

This is where the enthusiasm has to meet the evidence, and the evidence is polite but unimpressed. A 2026 meta-analysis covering 146 studies of social robots teaching cognitive skills found that they could improve learning outcomes — but chiefly when the robot personalised its responses, and when a human teacher was present. Robots deployed as solo instructors could not manage the complexity of classroom content, respond to spontaneous student behaviour, or handle unforeseen occurrences. The authors’ summary of why is the sentence worth keeping: children and young people learn best when they are engaged not just cognitively but also with other people.

The trials behind that number are mostly small robots doing narrow jobs: NAO units used in language and coding lessons across Japan, South Korea, Europe and East Asia; voice assistants trialled in New Zealand primary classrooms; on-screen AI avatars providing individualised support. The blended model — teacher plus robot — is where the gains show up, which is exactly the model both Beijing and Salamanca have chosen, whether from reading the literature or from common sense.

The awkward part: the classroom robot business is not healthy

Here is the thing that should temper anyone’s excitement about a state-scale demand signal. The two robots the world actually associates with classrooms are NAO and Pepper, both from the French firm Aldebaran, with more than 19,000 NAO units sold over the platform’s life. Aldebaran filed for bankruptcy in France in February 2025 and was placed into receivership that June, after its German shareholder stopped funding it in favour of distributing a new range of products from China. In July 2025, Shenzhen-listed Maxvision Technology bought the core assets, including all NAO and Pepper intellectual property, and set up a French subsidiary to keep development and support running with the original engineering teams.

So the most recognisable Western classroom robot is now Chinese-owned, at the exact moment Chinese schools became the largest coordinated buyer of classroom robots on the planet. For US buyers that collides with the other direction of travel: the FCC’s decision to add foreign-made humanoids and quadrupeds to its Covered List in July 2026 blocks new models from being imported, marketed or sold. A school district shopping for a teaching robot in 2027 will find that the cheap, mature options and the politically comfortable options are not the same list.

What this means if you are buying for a child at home

Do not buy a humanoid. Nothing in any of this suggests that a walking robot teaches a child anything a cheaper machine cannot, and full-size humanoids remain developer hardware — even the cheapest credible one, the Unitree R1, starts around 4,900 dollars, and humanoid pricing is slippery once hands, spares and support enter the quote. What the school programmes are actually teaching is the loop: give an instruction, watch a machine execute it, work out why it did the wrong thing.

For that loop at home, three things work at three different ages. For younger children, an expressive companion or robot pet gets the attachment and the routine going — our picks are in the best robot pets for kids guide. Once a child can read code, an open, hackable platform beats a sealed toy every time, which is the argument in our open-source robots roundup. And if you want the version we have actually built and lived with ourselves, the Stack-chan desktop robot is a genuinely good first project: cheap, printable, programmable and small enough to fail safely.

The wider point is worth sitting with. A generation of children in Beijing, Hangzhou, Dubai and, from next year, across India is being taught that a robot is a normal piece of classroom equipment — a thing you instruct, debug and are mildly annoyed by. That is a bigger shift than any product launch. The people who will buy household robots in the 2030s are currently in year six, and they are not going to be impressed by a backflip.

Frequently asked questions