Dyna Robotics raises US$120 million Series A from Robostrategy, CRV and First Round to scale its DYNA-1 robot foundation model into paid laundry and restaurant work

Dyna Robotics, a Redwood City, California, start-up that sells stationary robot arms for folding and other repetitive service work, said on 15 September 2025 that it had raised US$120 million in a Series A round. The round was led by Robostrategy, CRV and First Round Capital, according to the company's press release issued at 5:00 p.m. Eastern Time that day.

Salesforce Ventures, NVentures (NVIDIA's venture capital arm), the Amazon Industrial Innovation Fund, Samsung Next and LG Technology Ventures also took part, the release said. Dyna did not disclose a valuation. The money came almost exactly six months after a US$23.5 million seed round announced on 25 March 2025 and co-led by CRV and First Round Capital, which brings Dyna's disclosed equity funding to US$143.5 million. In a post on Dyna's website, chief executive Lindon Gao wrote that the round closed as the company turned one year old.

What Dyna says it has built

Dyna's pitch rests on DYNA-1, which the company describes as a single-weight, general-purpose robot foundation model. In the Series A release Dyna said the model holds a success rate above 99% over 24 hours of non-stop operation and can be deployed in new environments without additional training data. A research page Dyna published in June 2025 gives the underlying test. DYNA-1 folded more than 850 napkins over a 24-hour autonomous run at a 99.4% task success rate with no human intervention, working at roughly 60% of human folding speed. The same page shows the model folding shirts of different fabrics and sizes and learning a cup-filling task with 0.7% of its training data in under two weeks.

The company credits much of that reliability to a reward model, a second network that scores how well each attempt went so the robot can practise, recover from mistakes and generate fresh training data on its own. In his Series A post Gao wrote that Dyna runs hundreds of model variants each week through offline and online evaluations, and that a robot's total cost of ownership has to stay below thresholds such as US$50,000 or human labour remains cheaper.

Customers and deployments

According to the release, Dyna robots work in hotels, restaurants, laundromats and gyms, and after six months of deployment they run up to 16 hours a day. The one customer Dyna has named in detail is Monster Laundry, a full-service laundry in Sacramento. In a case study dated 14 October 2025, Dyna said its towel-folding robot, which the owners nicknamed Sophy Swiftfold, folded more than 200,000 garments for 10 commercial customers over three months with a 99% quality acceptance rate. Monster Laundry raised its commercial order capacity by 25% and saw store traffic rise 15% after the robot arrived, the case study said, and it put the cost of a traditional industrial folding system at more than US$300,000.

Founders and the seed

Gao and co-founder York Yang previously built Caper AI and exited it for US$350 million, according to Dyna's seed announcement. The third co-founder, Jason Ma, is a former DeepMind research scientist whose robotics work the release said had been recognised at ICRA, CoRL and ICML. At the seed stage Dyna framed its strategy as mastering single commercial tasks on low-cost stationary arms, starting with folding and food preparation, and using the real-world data those jobs produce to improve its foundation model. In a May 2026 post, Yang wrote that Dyna's longest-running deployment had reached about 10 months of daily use.

Where it sits among robot-brain start-ups

The Series A was small next to the largest robot foundation model rounds tracked by MEKAWIRE. Figure AI announced a Series C of more than US$1 billion at a US$39 billion post-money valuation one day later, on 16 September 2025. Skild AI raised US$300 million at a US$1.5 billion valuation in its July 2024 Series A and US$1.4 billion at US$14 billion in January 2026. FieldAI announced US$405 million across two rounds in August 2025, Paris-based Genesis AI raised US$105 million in June 2025, and Generalist AI raised US$400 million in June 2026. Physical Intelligence, which raised a Series B in November 2025, is the closest rival in approach because it builds a general model rather than a humanoid body. Dyna differs from most of these companies in betting on commercial revenue from narrow tasks before generality.

Dyna's own disclosed operating metrics are collected below. All figures come from Dyna's releases, research page and customer case study and have not been independently audited.