Physicl said on November 6, 2026, that it raised $10 million in a seed round led by Inovia Capital, with participation from U.S. Venture Partners (USVP). The startup also launched Studio, a tool for generating training data for robotics and computer vision models.
The pitch rests on what the company calls an "edge case gap." Robotics and vision teams need examples of rare, hard-to-capture situations, and those are the very scenarios real-world collection tends to miss. Founder and CEO Alex de Vigan put it bluntly in a November 6 funding post: "physical worlds cannot be scraped. They have to be built."
What Studio does
Studio produces targeted data from Physicl's own library, which the company says holds 10,000 environments and 100,000 underlying assets. Its press release phrases the asset figure as "more than 100,000." Both numbers are the company's inventory claims, not independent measurements.
The tool renders images. According to the product's FAQ, it does not run physics forward through time, a distinction the company draws itself. Physicl's materials describe the output as "COCO-shaped," a phrase that should not be read as full compatibility with COCO, the widely used image-dataset format.
Studio is free during beta, per Physicl's November 6 launch explanation. In the materials reviewed, the company has not established paid pricing or recurring revenue.
The investors
Steve Woods, an Inovia partner, supplied the investor comment in the release. "Physicl is tackling a critical piece of that puzzle: scalable, high-quality physical data grounded in real-world physics," he said.
Woods wrote in a LinkedIn post that he first met de Vigan roughly five years ago, while conducting technical diligence on de Vigan's previous company. Inovia's portfolio directory lists Physicl under Robotics.
De Vigan's post thanked Dafina Toncheva of USVP for backing the team from the beginning, though it gave no individual check sizes. Physicl disclosed no valuation in the materials reviewed, and RoundOS records the November seed as the company's only disclosed round.
Customers and origins
Matthias Niessner, CEO of SpAItial and a professor at TUM, is quoted in the founder's post: "Physicl gives any researcher access to virtually any physical training data they can imagine."
The company's origins are harder to pin down. A March 17 launch announcement, issued at Nvidia's GTC conference, said Physicl was launched by members of the Nfinite team. That release claimed a library of millions of simulation-ready assets and environments, a figure the November materials do not match, and the research does not explain the gap. Nor is the legal relationship between the two companies established.
Basic facts about the company's size and base also diverge by source. SignalBase lists 44 employees, while Physicl's own company page says "Thirty of us." The Next Web described Physicl as Paris-based; the company identifies itself as a U.S. corporation with three office locations.
Competition
Antioch, which builds simulation and evaluation tools along with synthetic-data infrastructure for physical AI, announced $32 million in new financing on September 8, 2026. That brings its total to $40.5 million, including an earlier $8.5 million seed. Its scope is broader than that of Studio's documented renderer.
Nvidia is both an ecosystem foundation and an overlapping tooling provider. On October 21, 2026, it announced expanded Isaac tooling, including Isaac Lab 3.0 and workflows for generating data and for training and evaluating models.
What's next
Studio remains in free beta. Physicl has not said when it will begin charging, and it has published no timeline for paid plans.
