Antioch, a physical-AI simulation startup founded by former Tesla Autopilot, DeepMind and Meta Reality Labs engineers, has closed a $32 million Series A led by Greylock Partners. The round takes total funding to $44.75 million after a $4.25 million pre-seed and an $8.5 million seed in April.
Who's In The Round
Greylock led, with participation from A* Capital, Category Ventures, BoxGroup Ventures and Icehouse Ventures. The company has been building the platform in tight coordination with NVIDIA and Nebius Group – both are named strategic infrastructure partners, giving Antioch a straight path onto Blackwell-class GPU capacity for simulation workloads.
Software Speed For Robot Testing
The pitch is straightforward: robotics, autonomy and perception teams cannot iterate at software speed when validation happens on physical hardware. Antioch's simulator runs perception-grade sensor stacks, physics and scenario generation so a policy update can be tested against millions of virtual miles overnight instead of waiting on a fleet week. Amazon Ring VP Jason Mitura, cited in the announcement, said "Antioch's simulations have closely matched our physical test results" in Ring's own validation runs.
Where The Money Goes
Founder and CEO Harry Mellsop – ex-Tesla Autopilot – said the round funds product work, engineering hires and a push to deepen sim-to-real fidelity. Co-founders Alex Langshur and Michael Calvey previously built Transpose (acquired by Chainalysis in 2023); two additional co-founders come from Google DeepMind and Meta Reality Labs.
Physical AI's Testing Bottleneck
Antioch lands in a rapidly consolidating tools layer around physical AI. It sits alongside NVIDIA's own Isaac and Omniverse stack and the wave of teleoperation-data platforms like XDOF. The bet: as humanoid and mobile robot fleets multiply, the software that decides whether a policy is safe to ship becomes as important as the models themselves.
Reporting based on coverage from SiliconANGLE, The Robot Report and Seedtable.