
Editor’s note: This article is republished from Jumpstart, a TechNode Group publication.
Noetix Robotics’ RMB9,998 Bumi makes the humanoid robot market easier to enter, but the price is only the opening move. The company is betting that an affordable machine for education and companionship can attract buyers now, while its teams develop more capable robot intelligence for the future.
That strategy gives Noetix a practical route into the market: sell a useful physical product before general-purpose robots are ready. It also creates a demanding test. The company must manufacture Bumi at scale, support customers, and fund research into embodied intelligence without losing the economics that made the robot accessible.
Founded in September 2023 by Tsinghua University doctoral alumnus Jiang Zheyuan, Noetix has pursued those goals in sequence. It first focused on robot bodies and motion control, then began building an intelligence programme and exploring international sales. Media reports in October said the company had confidentially submitted a Hong Kong listing application, with a possible 2027 flotation in view.
A humanoid with a defined first job
Noetix introduced Bumi in October 2025 as a small robot for programming education and companionship. Its appeal is direct: children can programme its movements, interact with it and watch it dance. That gives families and schools a concrete reason to consider a humanoid without expecting it to perform general household work.
The price was part of the engineering from the start. In a March interview, Jiang said the company used composite materials in Bumi’s body to reduce reliance on metal structures. The design aimed to bring down costs before production, rather than treating affordability as a later pricing decision.
That choice reflects where Noetix saw the immediate technical bottleneck. Jiang said that when the company began, unreliable robot bodies and limited manufacturing readiness were major obstacles. Artificial intelligence methods were changing quickly, so Noetix concentrated first on making a machine that could move reliably.
Physical capability and intelligence are related, but they aren’t the same task. A robot can perform difficult movements while depending on outside models for conversation or higher-level instructions. Interpreting an unfamiliar setting and deciding what to do requires another layer of development.
Bumi’s initial purpose narrows the problem. Programming and interaction can support a product people can buy today, while Noetix works on more demanding capabilities. The trade-off is that customers need lasting value from those first uses. A compelling demonstration can prompt a purchase; continued use depends on the product, its software and the support around it.
Public attention meets factory constraints
Noetix has found ways to put its robots in front of large audiences. Its N2 humanoid placed second in the 2025 Beijing humanoid half-marathon. In 2026, Noetix robots appeared with comedian Cai Ming in a sketch on China’s Spring Festival Gala.
Those appearances bring the technology out of specialist demonstrations and into popular culture. They can also help connect awareness to retail: viewers who recognize a robot may be more willing to investigate a product they can purchase. But attention doesn’t make units, and the jump from a visible performance to repeatable delivery is where the company’s operating model gets tested.
At the AWE exhibition in March, Jiang said orders surged after the gala and set a target of more than 10,000 Bumi sales in 2026. Reaching that figure would require a production system that can keep pace with demand without sacrificing quality or timely service.
Jiang pointed to component availability and production-line management as key challenges. A shortage of one part can hold up a complete robot. Increasing monthly output also means hiring and training more workers, coordinating assembly, and keeping finished units consistent.
For a product priced to broaden access, production problems carry a direct commercial cost. Rework, delayed orders and service visits all consume resources. If the company can’t control those costs, volume alone won’t show that the affordable design works as a business. The 10,000-unit target is therefore a test of manufacturing discipline as much as demand.
Hardware sales and the cost of intelligence
Noetix is now building a second technical pillar. In September, it launched Scalabot, its embodied-intelligence sub-brand, and HERON, a world-model architecture. The company had started assembling a dedicated intelligence team in July 2025, according to Jiang’s September interview with Caijing.
The intended connection between hardware and research is straightforward. More affordable robots could reach more users and settings, creating opportunities to learn from how machines operate in the physical world. Better models could, in turn, help those machines handle more complex tasks.
That loop takes more than deploying robots. The company has to capture relevant experience, organise it into useful data, and train models against tasks that matter. A large installed base may give Noetix access to more real-world use; it doesn’t automatically produce better intelligence. Data collection and model development are their own engineering work.
Jiang told 21st Century Business Herald that intelligence research had become a major area of expenditure. He described hardware as a potential funding source for that effort. The company is asking its product business to support a research programme whose costs arrive before its longer-term commercial results.
Noetix’s backers include Chendao Capital, an investment platform associated with CATL. The reported Hong Kong listing plan could provide another route to capital if it proceeds. For the company, the central business question remains how much its hardware operation can contribute while it scales production and invests in research.
Overseas growth depends on local support
Noetix is also looking beyond China. At CES in January, co-founder Zhang Shipu named five target regions: North America, Europe, the Middle East, Southeast Asia, and Japan and South Korea. He described an ambition to reach thousand-unit scale in overseas expansion during the second quarter, backed by local deployments, customer service and repeatable reference projects.
Those details point to the work behind international sales. Schools need teaching materials and help using the robots. Distributors need parts and repair options. Families need a reason to keep a robot in use after the first burst of curiosity. Shipping a unit is only one step in establishing a market.
Noetix has developed some domestic relationships that connect its product to customers. Its investor, Beijing Industrial Development Investment Management, reported partnerships with programming-education provider Codemao and retailer Kidswant, along with school deployments and donations. These channels give the company ways to place robots within existing education and retail systems.
Replicating that approach abroad means adapting to different languages, curricula and buying habits. Bumi’s price can make it worth a closer look, but local partners and ongoing support will shape whether that interest leads to sustained use.
The next test is execution
Noetix has put a humanoid within reach of more households and classrooms by giving it a defined purpose and designing around a target price. That first product provides a foothold while the company works on intelligence systems that could expand what its robots can do.
Now the company has to make three efforts reinforce one another: higher production, more capable robots and overseas expansion. Affordable hardware gives Noetix a way to meet customers early. Whether that foothold can support the next stage will depend on reliable deliveries, useful customer experiences and a research programme that turns real-world robot use into practical gains.