Dyna Moves Into Scaled Deployment

Date: 2026-08-27
Type: ARTICLE
Desk: Commercial Deployment / Physical AI
Topics: Commercial Deployment, Din Tai Fung, Physical Ai, Robot Foundation Models
Deck: Din Tai Fung is expanding its use of Dyna robots as the company says production performance has crossed the customer ROI threshold and its deployed fleet is set to reach hundreds by H1 2027.
URL: https://korthosrobotics.com/research/dyna-moves-into-scaled-deployment
Referenced profiles: Dyna Robotics (https://korthosrobotics.com/ecosystem/company/dyna-robotics)

Dyna Robotics says its robots have crossed the customer ROI threshold and are now moving to scaled commercial deployment.
Din Tai Fung is rolling Dyna robots out across its restaurant network, with deployments across hotels, logistics and data centres expected to take the fleet into the hundreds by H1 2027.
In production, Dyna says it increased napkin folding from 35 to 95 per hour while lifting the quality pass rate from 75% to 93%.
At Din Tai Fung, each robot needs to produce 1,500 table-ready napkins during an 18-hour shift. Dyna-2 now produces 1,590, compared with around 480 for Dyna-1.
Dyna also says a new deployment can now go from setup to meeting the production ROI bar in as little as three days, after building reusable deployment, monitoring and data infrastructure across its fleet.
Earlier this month, Dyna-2 was released, a world-action model pretrained on more than 1M hours of human video that showed human-to-robot transfer scaling.
It came amid a run of robot foundation-model releases, including Skild AI's S1 and Generalist's GEN-1.5, both focused on allowing robots to learn new tasks from very small numbers of demonstrations. S1 uses a single video demonstration as context, while GEN-1.5 reported one-shot learning from as little as 3 to 12 seconds of demonstration data without fine-tuning.