Pateo Turns Cockpit Plants Into Yunji Robot Workshops

Pateo and Yunji Technology signed a pact on August 14 that puts cockpit electronics plants to work on commercial robots. The Shanghai cockpit supplier, listed as 02889.HK, will build control modules for Yunji and swap speakers for skin-made sound.

Pateo CONNECT booked RMB 3,509.7 million in 2025. Yunji, listed as 02670.HK, booked RMB 301.2 million, and hotels still supplied 81 percent of that total.

The Factory Line Is Still a Sideline

Beijing Yunji Technology sells indoor service machines and the software that schedules them. Chairwoman Zhi Tao, who also serves as president, still frames the firm around hospitality agents. The 2025 annual report keeps hotels at the centre of the cash register.

Factory work is the fast-growing slice, and it is still small. Yunji’s factory revenue rose 393 percent last year, to RMB 15.1 million, or 5.0 percent of sales. That is the book the auto-plant trials are meant to thicken.

YUNJI 2025 SALES BY SETTING

The same filing says Yunji already ran a factory AI-agent project and calls itself the sole core supplier to one Fortune Global 500 manufacturer, with a first export of material-moving kit. So car plants are not a first industrial experiment. They are a way to put auto-grade electronics behind a line that is still a rounding error next to room service.

Hardware still dominates the mix. Robots and functional kits brought in RMB 213.3 million, or 70.8 percent of 2025 sales. The AI digital system, including HDOS subscriptions, contributed RMB 87.9 million. HDOS clients jumped from 237 to 2,430. Gross profit was RMB 113.4 million. The company still lost RMB 295.2 million before tax.

Pateo Will Make the Control Modules

Pateo CONNECT Technology (Shanghai) Corporation is a smart-cockpit house founded in 2009 by Ying Zhenkai, the chairman. It listed H shares in Hong Kong on September 30, 2025. Its 2025 results show a business that already stamps electronics at car volumes, and still does not make a statutory profit.

PATEO’S 2025 COCKPIT MACHINE

• Sales: Revenue reached RMB 3,509.7 million, up 37.26 percent, with gross profit of RMB 434.2 million and a 12.4 percent margin.
• Units: Smart cockpit domain controllers hit 1.3 million, up more than 40 percent from 2024.
• Loss: The statutory loss for the year was RMB 1,093.7 million; adjusted net loss narrowed 32 percent to RMB 239.8 million.
• Research: R&D spend was RMB 379.3 million, with 709 specialists, 1,909 granted patents, and 1,039 invention patents.

High-end work sits on Qualcomm Snapdragon 8295 and 8397 boards and Huawei’s Kirin 9610A. The results also describe automated production lines, shared order-to-logistics data, and a push to raise plant quality. That plant story is what Yunji is buying. Under the August pact, Pateo will use precision electronics production to offer OEM and EMS work on parts such as robot control modules, with joint purchasing of chips and other key materials.

On paper the two firms are peers on the Hong Kong board. In the plant they are not. Pateo’s 2025 sales were more than eleven times Yunji’s. The robot maker gets auto-grade boards, shared chip buys, and a factory customer next door. Pateo gets a robotics SKU without having to invent a hotel fleet.

Car Plants Get the Next Robot Shift

The public text of the deal is careful. The companies will discuss robots inside automotive factories, then try embodied machines on material handling and assembly help along auto-parts lines. If those jobs hold, the same kit goes on sale to other car plants and to general manufacturing.

WHERE THE WORK IS MEANT TO LAND

• Plant robots: Joint trials of service machines inside automotive factories, using Yunji’s existing indoor navigation and Pateo’s electronics.
• Line help: Staged tests of embodied machines for material handling and assembly support on component lines.
• Resale: After those jobs work, the pair will take the same package to other auto makers and to wider industry clients.

Yunji already has the physical forms. Its first machines were hall-and-room runners. The current flagship is the UP series, a mobile chassis that takes swappable upper cabins for delivery, cleaning, or monitoring, including on factory floors. Older Run series delivery robots still define the hotel picture most guests see. Average daily robots online in 2025 ran at about 32,000, and the annual report puts coverage above 40,000 hotels.

A car plant is a harsher corridor than a hotel floor. Cycle times are tighter, dust and metal scrap are worse, and a missed tote costs more than a late toothbrush. That is why a cockpit EMS shop matters more here than another lobby demo. Pateo already lives inside OEM quality systems. Yunji’s factory sales, even after a 393 percent jump, still need that kind of board and that kind of buyer.

Sound From the Skin, Not a Speaker

One product clause is easy to miss beside the plant talk. Yunji said it will put Pateo’s intelligent sound system first in new robot work, replacing ordinary speakers so the shell itself makes the voice. Pateo’s own English note calls this sound generation from the skin.

Hotel robots talk in corridors that already hum with HVAC, trolleys, and check-in chatter. A grille speaker on a short cabin is a cheap way to sound like a gadget. Driving the outer skin turns the whole body into the driver, which is the same trick some cars use when they want audio without a visible cone. Pateo’s day job is cockpit acoustics and voice. Wiring that stack into a service robot is a parts sale, and it is also a way to make the machine feel less like a rolling kiosk.

No volume, price, or launch date sits in the public note. The clause still tells you how Pateo thinks about robots: as another enclosure that needs a domain controller, a voice path, and a plant that can hold tolerance. Chip joint-buying sits in the same paragraph, aimed at supply security rather than at a named foundry.

What OpenHarmony Adds to the Stack

The frontier clause is an operating-system bet. The two firms will research a “brain-cerebellum” coordinated architecture on OpenHarmony, the open-source project incubated by the OpenAtom Foundation, and try to put that split into embodied products. Pateo had already written the same split into an HKEX filing in April, four months before Yunji signed.

On April 10, 2026, Pateo entered a framework pact with the Graduate School of AI Operating System at Shanghai Jiao Tong University. The filing says the pair will build OpenHarmony’s original big brain-small brain OpenHarmony plan for robots, intelligent cars, and wearables, with the large brain doing cloud recognition and the small brain doing on-device control.

Jointly and collaboratively construct the original solution of OpenHarmony “big brain (cloud recognition)-small brain (end-side in-time control)” on scenarios such as robots, intelligent automobiles and wearable devices.

PATEO CONNECT, HKEX voluntary announcement, April 10, 2026

That is the same job the industry now describes as a robot brain for planning and a cerebellum for motion. Yunji’s own annual letter already talks in those terms. HDOS is the “core brain” for speech, intent, fleet scheduling, and task split, and the composite robots are the “limbs.” Pateo is offering a car-world OS path under that split, on a stack Beijing wants to keep in domestic hands.

OpenHarmony is no longer a lab toy. The foundation’s OpenHarmony 6.0 Release version, issued on September 27, 2025, put community code above 1.3 billion lines, with more than 9,200 contributors and more than 1,300 products through compatibility tests by the end of August 2025. Cars, metros, grids, and industrial controllers already ship on it. A hotel robot that boots the same family as a cockpit domain controller is the point of the August add-on.

PATEO’S EMBODIED CALENDAR IN 2026

1. April 10, 2026: Signs the SJTU framework on OpenHarmony embodied intelligence, including a big-brain and small-brain protocol platform.
2. July 15, 2026: Says it joined the application working group of MIIT’s humanoid robot and embodied intelligence standards committee.
3. August 14, 2026: Signs the Yunji pact covering plant robots, skin sound, chip buys, EMS modules, and the OpenHarmony architecture.
4. August 17, 2026: Posts the English cooperation note describing OEM and EMS work on robot control modules.
5. August 19, 2026: Tells the market a leading NEV maker picked its 8397 cockpit platform as a standard fit, on top of existing 8295 supply.

The April filing is a framework with no binding sales. The August pact is the same kind of paper: a map, not a purchase order. Yunji is the first named robot company asked to live on that map.

Both Companies Listed in One Autumn

Both names are young public files. Pateo priced 10,436,900 H shares at HK$102.23 on September 30, 2025. Yunji listed on October 16, 2025 under Chapter 18C as a specialist technology name, selling 6.9 million shares after backing from Alibaba, Tencent, and Lenovo. Each spent 2026 proving it is more than the niche that got it listed.

For Yunji that niche is hotels, still 81 percent of sales, with a 23.1 percent rise in group revenue that did not stop a wider pre-tax loss. Factory and technology sites are the proof that the chassis can leave the lobby. For Pateo the niche is the cockpit, and the results say AI software and cloud-style services were only about RMB 95.1 million of the 2025 total. Embodied work, including this pact, is how a cockpit supplier tells investors it can stamp boards for machines that walk.

The noisy public fight in robotics is still about biped demos and who owns the training data that teaches a hand to pick up a cup. This deal lives one layer down. It is about who builds the control module, who buys the chips with it, which voice stack sits in the shell, and which OS splits planning in the cloud from control at the edge. That layer rarely trends. It is where a hotel-robot firm with a RMB 15.1 million factory book meets a cockpit plant that already shipped 1.3 million domain controllers.

Cockpit Orders Still Pay the Bills

Five days after the Yunji signing, Pateo’s cash register spoke in the old language. A voluntary filing on August 19 said a leading new-energy car maker had named the 8397 cockpit platform a standard fit, covering the rest of that OEM’s lineup beside existing 8295 boards. The 2025 results already pointed at 8397 design wins heading into 2026 mass production, plus extra models at an international OEM certified in more than 100 countries.

Pateo has also talked up technical work with NVIDIA, Horizon Robotics, and SenseTime, and a separate robotics “brain” project it has not named. Those notes sit beside the Yunji pact, not above it. The plant that will EMS robot modules is the same plant filling cockpit orders. Until a robot control board shows up as its own line in a results table, the embodied file is a customer and a story, and the cockpit is the bill-payer.

Yunji’s next test is whether auto-parts handling can move factory sales off a 5 percent share without breaking hotel service levels that still fund the group. Pateo’s test is whether EMS margins on robot modules can look anything like a 1.3 million-unit cockpit run. The August 14 paper commits both to try. It does not yet say when the first module ships, or what it will cost.

Disclaimer: This article is news reporting and analysis of a corporate cooperation and of figures published by the two listed companies, and it is for information only. It is not investment advice, a solicitation to buy or sell any share, or a recommendation on Pateo CONNECT (02889.HK), Yunji Technology (02670.HK), or any related security. Readers should consult a licensed financial adviser or other qualified professional before making any investment decision. Revenues, losses, unit counts, and the status of the pact reflect the company filings and notices available as of September 2, 2026, and those items can change with later results, contracts, or disclosures.