Antioch, a 16-month-old startup founded by a former Tesla Autopilot engineer, has raised $32M in Series A funding led by Greylock Partners, with A* Capital, Category Ventures, BoxGroup Ventures and Icehouse Ventures joining. Angels include Palantir CTO Shyam Sankar, Foxglove CEO Adrian Macneil and Nvidia executive Ian Andrews.
The company sells simulation software that calibrates a virtual model to a customer’s own hardware and runs it on cloud infrastructure, so a team that could afford one physical experiment can run thousands of simulated evaluations in parallel. Co-founder Harry Mellsop, who worked on Autopilot at Tesla before starting Antioch in May 2025 with four co-founders, says high-fidelity simulation is the way autonomous systems get validated faster, safer and more reliably.
Amazon’s Ring unit is an early customer. Its vice president of software, Jason Mitura, says Antioch’s simulations have closely matched physical test results, including in scenarios deliberately held out of calibration. Nvidia is a partner through its open physical AI stack, including Omniverse libraries and the Isaac Sim and Isaac Lab robotics frameworks, and Nebius supplies cloud infrastructure.
The bet echoes a broader shift in physical AI: robotics companies are discovering that testing and data, not just hardware, decide how quickly autonomous machines reach the real world.
