NEW YORK — Physical AI simulation startup Antioch has raised a $32 million Series A round led by Greylock to expand a platform for testing robots and autonomous systems in virtual environments.

Investors including A*, Category Ventures, BoxGroup and Icehouse Ventures also took part in the funding round, according to Antioch’s announcement on Sept. 8. The company's total funding now stands at $40.5 million, which includes the $8.5 million seed round that was announced back in April.

Antioch, based in New York, plans to use the new capital to develop its product, expand its engineering team and improve its simulation technology. The company did not disclose its valuation, revenue or terms of the financing.

The platform is designed to give robotics and autonomy developers a faster way to evaluate changes to perception models, motion planners, controllers, sensors and mechanical designs. Physical testing remains necessary, but it can consume equipment, engineering time and access to test sites. Some hazardous or rare failures are also difficult to reproduce.

Antioch says its software builds simulations around a customer’s hardware, sensors, software and operating conditions. Developers can create scenarios, run evaluations in parallel and compare simulated results with data collected from real systems. The company also offers synthetic-data generation for circumstances that would be costly, dangerous or impractical to capture physically.

The central technical challenge is the “sim-to-real gap,” in which behavior observed in simulation does not accurately predict how a machine will perform outside it. Antioch says its customer-specific calibration can reduce that mismatch, but it has not published independent benchmarking that would allow its accuracy to be compared with competing simulation systems.

The company identified Amazon and AI manufacturing startup Launchpad Build AI as customers. Jason Mitura, Amazon’s vice president of software development and Ring’s chief product officer, said Antioch’s simulations closely matched physical test results, including scenarios excluded from calibration. That assessment was included in Antioch’s announcement and was not independently evaluated.

Antioch is also integrating Nvidia’s Omniverse libraries, Isaac Sim and Isaac Lab into its platform, while working with Nebius on cloud infrastructure for large simulation workloads.

The startup was founded in 2025 by a team with experience at Tesla, Google DeepMind and Meta Reality Labs, according to SiliconANGLE. Now, Saam Motamedi, a general partner at Greylock, will become a part of Antioch's board after they invested in the company.