Chef Robotics says it has produced 100 million servings. The number sounds like a milestone. It is not what it sounds like.
A company spokesperson defines a serving as one component of a meal deposited into a tray. Not a meal. One ingredient, placed once, into one compartment. A complete tray might have six or eight servings in it. The company declined to say how many complete meals 100 million servings represents. TechCrunch
This is the robot cooking graveyard, and Chef Robotics is one of the few companies still standing in it. Zume burned through $400 million building pizza delivery robots and collapsed in 2023. Chowbotics built a salad-making machine, got acquired by DoorDash, and was shut down. Karakuri, another food robotics startup, also closed. TechCrunch The graveyard has room for more. Chef Robotics, founded in 2018 by Rajat Bhageria, almost joined it.
The survival story is real. The milestone needs context.
Bhageria spent years trying to build a robot that could work a fast casual restaurant line — the kind of place where one employee assembles one complete order with dozens of possible ingredients. He could not make it work. The robotic grasping problem defeated him: training a robot arm to pick up a blueberry without crushing it, or to handle cheese without clumping it, requires data that did not exist. He asked restaurant chains to give him access to install robots for one or two ingredients to gather that data. They said no. TechCrunch
So he turned away the signed contracts and the multimillion-dollar revenue. It honestly sucked, Bhageria said in a 2025 interview. Because I spent the last year and a half of my life trying to convince these people. TechCrunch
What Bhageria found instead was food manufacturing, specifically what the industry calls high-mix production. These are facilities that make thousands of different meals or meal trays — airlines, hospitals, frozen food companies, school lunch programs — with workers standing in cold rooms assembling one ingredient onto one tray, hundreds of times a day, eight hours at a stretch. The work is brutal. The labor shortage is chronic. And unlike a restaurant line, the assembly-line structure means a robot does not need to handle the full range of human dietary chaos in one shift.
Chef is deployed at Amys Kitchen, the organic frozen food company, and Chef Bombay, a prepared meal producer. The company also works with one of the largest school lunch providers in the country. TechCrunch Robots handle individual ingredients across hundreds of recipes, the same motion repeated with different food types. The data from each motion trains the next one.
That is the data flywheel Bhageria talks about: every serving generates real-world training data, which improves the AI models, which improves the robots, which handles more ingredients, which generates more data. TechCrunch Chef own blog says its latest robot, the Chef+, was built on insights from over 80 million servings and more than 100,000 production hours. Chef Robotics The company raised a $43 million Series A in late 2025, per AgFunderNews, bringing total funding to roughly $55 million including earlier rounds and debt financing.
The growth is real by one measure. In April 2025, TechCrunch reported Chef had produced 45 million meals with 40 employees and a $20.6 million raise. TechCrunch By April 2026, the company was claiming 100 million servings. If those numbers are accurate, production roughly doubled in twelve months. Whether that represents doubling of revenue, deployed robots, or customer contracts is not public.
What is public: Chef is now moving into airline catering and ghost kitchens, the faceless delivery operations that supply DoorDash and Uber Eats. The airline catering customer Bhageria called a smaller kitchen is one of the largest such companies in the world. Eventually, he wants back into fast casual restaurants, stadiums, and prisons. TechCrunch
The graveyard lesson is clear: robot chefs failed in public, where diners could see them fail. They survived in private, where the people eating do not have opinions about the automation. The question is whether that survival translates into a business, or whether 100 million component deposits is a milestone the size it sounds like.
