Restaurant Robots Move from Booths to Kitchens: Efficiency and Cost Still Key
Food service robots are now moving from exhibition halls into real restaurant operations, handling greeting, cooking, and food delivery.
Yet once they stand at the stove and counter, beyond the novelty, efficiency, cost, and stability will determine whether they can truly stay.
In late August, a dual-arm robot developed by Sharpa went into service at a DQ store in Shanghai. It independently completes 55 consecutive steps from cup retrieval, to ice cream dispensing, to topping addition and mixing, producing a dessert that must pass the 'upside-down without spilling' test. This is the first time a robot has used the store's existing standard equipment, ingredients, and tools to make a product for real customers in an actual restaurant.
Leaving the lab was not easy, though. The robot takes about 6-7 minutes per product, far longer than a skilled worker's 2-3 minutes. Sharpa co-founder Li Yifan said they are not pursuing speed at the moment, but rather ensuring the process runs reliably; speed will improve as training data accumulates. The equipment currently carries a price tag of several hundred thousand yuan, with mass production expected to bring it down to around 100,000.
CFB Group CEO Xu Weilun, DQ's parent, said cost is not the primary consideration at this stage; what matters is whether the robot can produce consistent, standard products. If even a highly standardized process like ice cream-making cannot be reliably completed, more complex culinary tasks become even harder.
Unlike the ice cream scenario's zero-modification approach, cooking robots have seen wider deployment in Chinese kitchens. Chain brand Xiaocaiyuan has activated robots in nearly 300 stores, and others like LaoXiangJi are also promoting them. Industry reports generally indicate a 30-50% improvement in meal preparation efficiency.
According to the 2026 China Cooking Robot Industry Development Report, the market size reached 3.7 billion yuan in 2025 and is projected to hit 4.59 billion in 2026. Yang Jiancheng, founder of Oak Deer Robot, said its AI cooking robots now cover nearly 350 Chinese cities and over 20 foreign countries, with more than 13,000 restaurant deployments and nearly one million digital recipes accumulated.
Oak Deer's calculations show a commercial cooking robot pays for itself in roughly 12-15 months, essentially freeing chefs from intensive repetitive tossing so they can focus on dish development and flavor innovation. In Hangzhou, some restaurants lease the equipment for 198 yuan per month with a 15,000 yuan deposit, claiming a 60% labor cost reduction.
However, this model has clear limitations. Commercial cooking robots suit standardized batch production but not personalized single-person orders, nor can they flexibly adjust flavors according to customer requests. Some high-end restaurants that value traditional craft refuse to adopt robots, insisting that a chef's understanding of ingredients and on-the-spot adaptation cannot be replaced by machines.
From ice cream stores to Chinese kitchens, robot deployments reveal a core insight: full automation is not the sole goal of industrial development; refined human-robot collaboration is a more realistic direction.
Robots handle standardized, repetitive processes, while humans take on personalized service, quality control, and flavor innovation. This division of labor has proven viable in some chain brands. After smart upgrades, multiple community noodle restaurants in Beijing have freed their staff from repetitive steps like boiling and draining noodles, allowing them to focus on soup base preparation and customer service.
But moving from flagship stores to widespread adoption requires solving several key issues: can execution efficiency approach human levels, can long-term operational stability be guaranteed, and does the investment yield a reasonable return? Hu Yanping, a distinguished professor at Shanghai University of Finance and Economics, analyzed that the marketing ROI of robot-enabled stores currently far exceeds their ROI as labor. Only when robots can consistently complete tasks at acceptable cost and speed for extended periods can they be said to have created full productivity.