Sharpa plans Dairy Queen shifts for its North humanoid in Shanghai

Why it matters

Sharpa is moving North from exhibition demos into a branded restaurant workflow, where repeatability, intervention rates and cost per order will matter as much as dexterity.

Sharpa, the robotics maker co-founded by Hesai executives David Li, Kai Sun and Shaoqing Xiang, plans to put its North humanoid behind the counter at a Dairy Queen China store in Shanghai in August, where the robot will prepare Blizzards for customers using the restaurant's existing equipment.

The deployment surfaced in a July 18th post on X from Humanoid Scott (@GoingBallistic5), who described the partnership as "Project Edison." Sharpa's own customer deployment page confirms the August timeline and identifies DQ China as the partner. The public launch is scheduled for August, despite the X post's claim that North was rolled out on July 18th.

Sharpa says North is currently training for the assignment. The public materials do not identify the store, its opening date, the hours North will work, the number of robots involved or the commercial terms between Sharpa and DQ China.

The deployment gives Li, Sun and Xiang a customer-facing test of a thesis they carried into Sharpa after building lidar maker Hesai: useful robots will depend on precision hardware, manufacturing discipline and control systems that can survive repeated work outside a laboratory. Hesai's corporate biography describes Li as a robotics Ph.D. who worked at Western Digital, Sun as a Stanford-trained mechanical engineer and Xiang as a former Apple iPhone hardware systems integration engineer.

A 60-step order in a standard kitchen

Sharpa says each ice cream order requires North to execute 50 to 60 consecutive manipulation steps. North must retrieve and hold a paper cup, measure toppings, operate the blending equipment, open cabinets and complete Dairy Queen's signature upside-down Blizzard presentation.

The physical setup is central to Sharpa's pitch. DQ China is not rebuilding the kitchen around North. According to Sharpa, the robot will use the store's existing handles, tools, ingredients and machines. That forces North to work in a space designed for human employees rather than a fenced automation cell with robot-specific fixtures.

Several parts of the order expose the weaknesses of conventional industrial grippers. North must hold a metal blending ring against the cup with its thumb and index finger while regulating the force applied to a deformable paper container. North must also operate a narrow cabinet handle and remember that an Oreo Blizzard requires three scoops even when each scoop leaves the visual scene looking largely unchanged.

Sharpa says North uses tactile feedback to avoid crushing the cup during blending and the final flip. Those assertions come from Sharpa's own demonstrations and deployment materials; Sharpa has not published third-party reliability data, order completion rates or intervention rates for the Dairy Queen task.

North has 67 degrees of freedom and uses Sharpa's Wave hands, which provide 22 active degrees of freedom at roughly human scale. Sharpa says the robot's CraftNet architecture combines vision, language, motion planning and tactile control, including a high-frequency reflex layer that adjusts its grip after contact.

Sharpa introduced North at CES on January 6th. In Sharpa's account of the CES demonstrations, North played table tennis, dealt cards, took instant photographs and assembled paper windmills. The Dairy Queen deployment raises the difficulty in a different direction: repetitive orders, changing lighting, moving customers, deformable packaging and food-service equipment that cannot be treated as fixed laboratory props.

A field test for Sharpa's full-stack strategy

Sharpa first built its identity around Wave, a dexterous robotic hand that entered production in October 2025, according to Sharpa. North and CraftNet extend that component business into a full humanoid system, while the Dairy Queen work places Sharpa closer to selling completed shifts rather than individual pieces of robotics hardware.

That shift carries a different commercial burden. Dexterous hands can be sold to universities, robot manufacturers and research laboratories based on specifications and developer access. A restaurant operator needs predictable throughput, safe operation around customers, cleaning procedures and a cost per completed order that competes with human labor or specialized food-service automation.

Sharpa's published material does not provide North's price, the cost of training it for a Dairy Queen workflow or the staffing required to supervise the August operation. Sharpa also does not say whether North will accept orders, handle payments, clean equipment or replenish ingredients. The confirmed task is the preparation and presentation of ice cream inside an operating store.

Li has described service work as Sharpa's initial route to commercial adoption. In a June 30th interview with Singapore's Economic Development Board, he said field deployments would supply the operating data and reliability work needed before broader general-purpose use. He also said North was expected to become available to enterprise customers by the end of 2026.

Sharpa has already announced planned food, retail and logistics pilots in Singapore with JTC and Grab. The Shanghai store provides a sharper public test because customers will be watching the robot carry out a recognizable branded workflow and perform the upside-down flip associated with the Blizzard product.

Hesai supplies the hardware base

Sharpa's relationship with Hesai extends beyond having the same three founders. A March 25th exchange filing published through the SEC says Li, Sun and Xiang together hold a majority of Sharpa's voting rights. The filing also shows that Hesai agreed to supply Sharpa with lidar products, robotic actuators and related manufacturing services through December 31st.

The agreement carries an annual transaction cap of RMB 100 million, which is a ceiling rather than a purchase commitment. Hesai said the arrangement covers actuator assembly, testing and integration into Sharpa's dexterous hands and other robotics products. All three founders abstained from Hesai's board vote because of their interests in Sharpa.

That supply relationship gives Sharpa access to hardware development and manufacturing capabilities built during the founders' previous decade at Hesai. It also makes execution at Dairy Queen a test of a tightly connected founder-controlled stack: Hesai-supplied sensing and actuation, Sharpa's Wave hands, the North body and Sharpa's CraftNet control architecture.

The August opening will determine whether that stack can repeat a 50-step food-service task under public operating conditions. A clean promotional run will establish the demonstration. Sustained order volume, intervention rates and economics will determine whether North has completed a shift that restaurant operators would pay to automate.