Dyna Robotics
Last reviewed
Jun 4, 2026
Sources
17 citations
Review status
Source-backed
Revision
v1 · 1,606 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Jun 4, 2026
Sources
17 citations
Review status
Source-backed
Revision
v1 · 1,606 words
Add missing citations, update stale details, or suggest a clearer explanation.
Dyna Robotics is an American embodied-AI and robotics startup founded in 2024 and based in Redwood City, California. The company builds robot foundation models for commercial robot manipulation, pairing them with low-cost, stationary dual-arm hardware that it deploys into real businesses such as laundromats, restaurants, hotels, and gyms. Its first model, DYNA-1, was unveiled in April 2025 and is marketed as one of the first robot foundation models to run autonomously in paying commercial settings. Dyna was founded by Lindon Gao, York Yang, and Yecheng "Jason" Ma; Gao and Yang previously built the retail smart-cart company Caper AI, which Instacart acquired for $350 million in 2021. By September 2025 the company had raised about $143.5 million across a seed round and a Series A, the latter valuing it at more than $600 million.
Dyna Robotics was founded in 2024 and emerged from stealth in March 2025. Its premise is that the main barrier to robot adoption is cost rather than capability, so it focuses on inexpensive pairs of stationary robot arms controlled by embodied AI models that each master one task or a narrow set of tasks at a time, beginning with chores like folding and food preparation. The founders describe the long-term goal as "physical AGI," meaning general-purpose physical intelligence for robots.
The three co-founders combine consumer-hardware operating experience with robot learning research:
| Founder | Role | Background |
|---|---|---|
| Lindon Gao | Co-founder and CEO | Co-founded Caper AI, an AI smart-cart company, sold to Instacart for $350 million (2021) |
| York Yang | Co-founder | Co-founder of Caper AI; engineering leadership at Dyna |
| Yecheng "Jason" Ma | Co-founder and Chief Scientist | PhD from the University of Pennsylvania GRASP Lab; robotics foundation-model researcher; lead author of NVIDIA's Eureka reward-design work |
Gao and Yang are repeat founders: their previous company, Caper AI (also styled Caper Inc.), built computer-vision-enabled self-checkout shopping carts and was acquired by Instacart in 2021 for $350 million. Ma completed his PhD at the University of Pennsylvania's GRASP Laboratory, advised by Dinesh Jayaraman and Osbert Bastani, and is associated with reinforcement-learning and robot-learning research, including the Eureka paper on using large language models to design reward functions. Press releases and the company's investors describe him as a former researcher at Google DeepMind and NVIDIA who has worked on foundation models for robotics.
Dyna has raised roughly $143.5 million in two rounds. The seed round was co-led by CRV and First Round Capital, and the Series A was led by RoboStrategy together with CRV and First Round Capital, with participation from several strategic corporate investors including the venture arms of NVIDIA, Amazon, Samsung, and LG.
| Round | Date | Amount | Lead investors | Notable participants |
|---|---|---|---|---|
| Seed | March 2025 | $23.5 million | CRV, First Round Capital | (co-led) |
| Series A | September 2025 | $120 million | RoboStrategy, CRV, First Round Capital | Salesforce Ventures, NVentures (NVIDIA), Amazon Industrial Innovation Fund, Samsung Next, LG Technology Ventures |
The Series A, announced on September 15-16, 2025, brought total funding to about $143.5 million and, according to reporting by Bloomberg and Investing.com, set a post-money valuation above $600 million. Dyna said it would use the capital to expand its research and engineering teams and to accelerate delivery of production-ready, general-purpose robots powered by its proprietary foundation models. The mix of strategic backers (NVIDIA, Amazon, Samsung, and LG corporate funds, plus enterprise-software investor Salesforce Ventures) reflects the broad industrial interest in deployable ai robotics.
DYNA-1, announced on April 29, 2025, is Dyna's robot foundation model. The company positions it as a single-weight, general-purpose model intended to perform a variety of everyday manipulation tasks across different commercial environments rather than a model hand-tuned for a single workcell. It is delivered as a full-stack system: two industrial robotic arms (with quick-swap grippers) mounted so they can sit at an existing workstation, driven by the learned policy.
Dyna's launch centered on a continuous napkin-folding run. A pair of arms running DYNA-1 folded more than 700 napkins over a 24-hour period with no human intervention, which the company reported as a success rate above 99.4 percent. The task is harder than it looks because it requires reliably separating a single napkin from a stack and recovering when several are pulled at once. The company's own materials and several profiles cite higher counts (figures of 800-plus and around 60 percent of human speed appear in the Series A framing), and Dyna says the underlying skills transferred with limited retraining to related tasks such as laundry folding and cup-filling, an example of partial zero-shot generalization to new environments.
DYNA-1 is trained with reinforcement learning and a custom reward model. The reward model is designed to judge whether the robot has completed a sample task correctly more reliably than earlier approaches, which lets Dyna grade large volumes of real-world attempts and feed that signal back into training. Because the model is deployed in production, every customer deployment becomes additional training data: robots stream sensor data, the reward model labels successful and failed manipulations, and the foundation model improves for all customers. This data flywheel, where each deployment compounds the model's advantage, is central to the company's strategy and to its investors' thesis.
Dyna has also referenced task-specific variants in 2026 evaluations, including a shirt-folding configuration (sometimes labeled DYNA-1i) reported at roughly 40 shirts per hour.
Dyna operates a business-to-business Robots-as-a-Service (RaaS) model. Rather than selling hardware outright, it charges a monthly fee per robot that bundles the hardware, software, maintenance, and continuous model updates into a single payment. The approach turns a large up-front capital expense into an operating expense, which lowers the barrier for smaller businesses, and it gives Dyna recurring revenue plus a continuous stream of training data. The founders have stressed that the hardware is deliberately cheap: where many AI-powered robots cost hundreds of thousands of dollars, Dyna aims for an order of magnitude less, on the order of tens of thousands of dollars per unit when sold.
Dyna says its robots moved into paying customer sites within months of launch. By around six months after coming out of stealth, the company reported DYNA-1 systems running roughly sixteen hours a day across hotels, restaurants, laundromats, and gyms, with live sites in San Francisco, Los Angeles, and Sacramento.
The most visible public deployment is at Monster Laundry, a midtown Sacramento laundromat that local outlet KCRA described as the first laundromat in North America to use Dyna's robotic folding system, nicknamed "Sophy Swiftfold." The robot folds towels, linens, and clothing using arms, sensors, and vision; the operator framed it as a labor aid rather than a replacement. Dyna has cited commercial laundry work as a flagship use case, reporting that its robots folded large volumes of towels across multiple commercial clients with a high quality-acceptance rate.
In a 2026 review of real-world robot autonomy, the research group Epoch AI singled out Dyna as having some of the strongest evidence of sustained real-world operation, citing the 24-hour run of 850-plus napkins at about 60 percent of human speed with a 99.4 percent success rate and zero interventions. Epoch added an important caveat: Dyna's commercial deployments are limited to simple, uniform items such as towels and napkins, not the diverse clothing a general household robot would have to handle, so the results demonstrate narrow, reliable specialization rather than broad household generality. That distinction places Dyna in the same competitive landscape as other robot foundation model efforts such as Physical Intelligence, Covariant and its RFM-1 model, Skild AI, and the robotics arms of larger players.