Antioch Raises $8.5M to Build “Cursor for Physical AI” as Robotics Simulation Race Heats Up

Antioch, a New York-based startup positioning itself as a “Cursor for physical AI,” is betting that the next big leap in robotics will come not from more hardware, but from better simulation software. The company has raised $8.5 million in seed funding at a $60 million valuation to build tools that help engineers train and test robots in highly realistic virtual environments.

The core challenge Antioch is tackling is the long-standing “sim-to-real” gap in robotics—where machines trained in simulations often fail when deployed in the real world. Because real-world robotic data is scarce, expensive, and difficult to collect, companies often resort to building physical testing environments or gathering large-scale operational data, both of which significantly slow down development.

Antioch’s platform aims to change that by allowing developers to create digital versions of their robots, connect them to simulated sensor systems, and run large-scale testing in virtual environments. These simulations can be used for reinforcement learning, edge-case testing, and generating training data that closely mirrors real-world conditions, helping robots behave more reliably once deployed.

The company draws comparisons to AI-powered coding tool Cursor, arguing that just as software development has been transformed by intelligent tools, robotics development is now reaching a similar inflection point. Backed by investors including A* and Category Ventures, Antioch is led by founders with backgrounds spanning Meta Reality Labs, Google DeepMind, and a previous startup acquired by Chainalysis.

As major autonomy players like Waymo continue refining real-world driving systems using advanced simulation models, Antioch believes the demand for accessible simulation infrastructure will explode. While the technology is still evolving, the company sees a future where most autonomous systems are designed, tested, and iterated primarily in software—making simulation a foundational layer of the physical AI economy.

source: techcrunch