Reconfiguring a factory robot can take weeks or months when a production line changes. Industrial Next says its self-contained robotic stations could cut that changeover to minutes, without traditional programming. The San Francisco-based startup presented at the Startup Battlefield stage at TechCrunch Disrupt 2026 and announced a partnership with BMW Group Technology Office to evaluate its AI-powered dexterous manipulation for manufacturing.
Industrial Next builds stations powered by proprietary foundation models that connect vision, task context, and motion. Its goal is to help manufacturers automate complex, delicate work that has traditionally required people. The company’s pitch is not that factory equipment changes instantly: it is that robots can be taught tasks through simulation and demonstration instead of relying on hard-coded scripts.
That distinction matters because most factory automation is built around a specific task and product. Traditional industrial robots are typically programmed for highly specific, repeatable work. A small change to a product or process can make a robot fail or cause costly downtime, and manufacturers may simplify products to make them easier to produce consistently.
Industrial Next says its stations can be reconfigured, combined, and redeployed across product lines. Its public material describes task learning in hours, while the company’s “minutes” claim refers to retooling a production line. The source material does not independently verify that performance claim.
The founders’ idea grew out of the gap between advanced vehicle technology and aging factory equipment. Industrial Next was founded in 2021 by Lukas Pankau and Allen Pan, both former Tesla employees. Pankau was a lead electrical architect at Tesla before moving to Waymo. Pan led Tesla’s autonomous factory efforts and later became chief architect at Neolix, an autonomous driving company.
Pankau and Pan came up with the idea while working together at Tesla. Their first encounter was less than polished: Pankau tripped over Pan, who was asleep on a factory catwalk after an all-nighter. They did not get along at first, but soon found they shared an obsession with getting things done and had complementary skills.
After years of watching cutting-edge vehicle and self-driving technology run up against decades-old factory equipment, the two concluded that manufacturing needed to be rebuilt. Pankau described the mismatch at Tesla: “We had all these fancy things that we could do on cars. I could run machine learning algorithms to design a car for me. But I couldn’t figure out how to program a robot, and this was pretty common for engineers at Tesla,” he said. “You had this massive divide, right? You had people who are really good at what they’re doing–designing cars and designing self-driving systems–and then you had them effectively stumped at the factory looking at this equipment from the 1980s. And so we started rethinking what this should look like using the technology that we have today and using everything that we know today.”
Their frustration followed them to their next jobs. They stayed in touch and often talked about machines that never worked the way they wanted. Pankau said he was at Waymo, “kind of bored out of my mind because I was missing that pace of Tesla.” Pan was in a similar position at Momenta, a developer of self-driving systems in China. Recalling their decision to start the company, Pan said: “We’re like, we’re not getting any younger, and we’re still obviously passionate about it. We still keep complaining about industrial technology. Like, this is three years later, we should start the company.”
Flexible robotics has a data problem. Internet-trained AI models can draw on vast datasets, but robotics lacks comparable training data for complex physical tasks. Pankau put the gap this way: “We can make robots do all these cool things like dance around, because there’s data for that, but we can’t make them assemble high-precision things.”
Industrial Next says it uses high-fidelity simulation to teach robots precise, dexterous work. Pankau said the company’s simulation engine closely matches real-world dynamics and gives robots a general understanding of a process. That, he said, can let them perform high-precision tasks “almost without seeing the parts at all.” In practice, the robots train in detailed virtual environments intended to mimic real-world physics, then learn new tasks through simulation and demonstration.
Pankau said some Industrial Next workstations are already used to make one of the world’s most popular smartphones. They handle tasks manufacturers had tried and failed to automate for years, including threading a delicate flex cable and snapping a tiny connector cap into place. He did not disclose the customer.
He also described a “film peeling” station that removes protective film from camera modules and buttons. According to Pankau, the robots work at or above human speed with 100% accuracy, even after being moved to a phone model and production line they have never seen.
Pankau said the company was founded to achieve the kind of performance the founders wanted at Tesla, where they sometimes bypassed robot stations because traditional technology did not let them program the equipment effectively. Industrial Next’s longer-term goal goes beyond phone assembly: Pankau said the company ultimately wants to build a full car end-to-end without any human touching it.
