CEOInterviews.AI
Start App
Try: AI strategy tariffs earnings guidance layoffs interest rates growth

Jason Ma

Co-founder & Chief Scientist, Dyna Robotics

Search every verified Jason Ma interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Jason Ma, co-founder and chief scientist at Dyna Robotics, has recently discussed the company's development of robot foundation models and their deployment in commercial settings. In talks and interviews, he described Dyna's approach to training models on over one million hours of data, including human video, and reported that the company's robots fold thousands of napkins daily for multiple locations of the restaurant chain Din Tai Fung. He stated that Dyna's models can learn new tasks from as little as 45 minutes of demonstrations and that the company aims for a zero-shot success rate above 99% before deploying robots at customer sites. Ma also mentioned customers such as Best Western in the hotel space, where robots fold towels. Ma presented Dyna's latest foundation model, Dyna-2, which he described as the first world-action model to demonstrate a scaling law from training on more than one million hours of data. He emphasized that predicting both future videos and actions was crucial for cross-embodiment transfer, and he argued that scaling data alone is insufficient without the right modeling objectives. Ma contrasted generalist models that achieve 80–90% success rates with Dyna's focus on reliability, and he outlined the company's use of a reward model to detect errors and collect targeted recovery data. He noted that Dyna's long-term plan includes developing a robot model and hardware platform for consumer use, but its current go-to-market strategy focuses on enterprise customers.

Recent appearances

  • The Bitter Lesson of Robotics: Why Scaling Wins — Jason Ma, Dyna Robotics

    October 13, 2026 · 📺 Latent Space

    From training a world-action model on over one million hours of human video to deploying robots that fold thousands of napkins daily at Din Tai Fung, Dyna Robotics is pushing physical AI beyond impressive demos and into real-world production. In this episode, Dyna Robotics co-founder and Chief Scientist Jason Ma joins swyx on-site to unpack how robots can learn new tasks from as little as 45 minutes of demonstrations, why predicting the future of the physical world may be the key to general-purpose robotics, and what it actually takes to build robots that work reliably for real customers. We…

    Watch on YouTube ↗

  • Keynote: Building Embodied AGI Through High-Performance Robots That Scale | Day 2

    September 28, 2026 · 📺 Fellows Forum

    Keynote: Building Embodied AGI Through High-Performance Robots That Scale Speaker: Jason Ma, Co-founder at Dyna Robotics Recorded at Fellows Forum 2026 (September 23 to 24, 2026).

    Watch on YouTube ↗

  • Robot Demos Are Easy. Reliability Is Hard — Jason Ma, Dyna Robotics

    September 24, 2026 · 📺 AI Engineer

    A generalist robot that succeeds 80 to 90% of the time makes a great video but a poor product. Jason Ma, co-founder and CTO of Dyna Robotics, shows how Dyna got its model, Dyna-1, to a 99.4% success rate folding restaurant napkins for 24 hours straight, including recovering after it pulled over the whole stack. The key is a reward model that watches the robot and scores its progress. When the score dips, the robot is making a mistake, so the team can collect targeted recovery data and fine-tune again in a human-in-the-loop active-learning cycle. Ma also covers Dyna's pre-training data pyramid…

    Watch on YouTube ↗

  • Actuate 26 | Dyna-2: A one million hour scaling law for world action models by Jason Ma

    September 14, 2026 · 📺 Foxglove

    Jason Ma is co-founder at Dyna Robotics. He received his PhD from University of Pennsylvania, where he studied scalable reinforcement learning algorithms and foundation models for robot learning. His research has been recognized with honors such as RSS Pioneers, Apple and OpenAI Fellowship, Best Paper Finalist at ICRA 2024, Top 10 NVIDIA Research Projects of 2023, and covered by popular media such as the Economist, Fox, Yahoo, and TechCrunch.

    Watch on YouTube ↗

  • Jason Ma (Dyna): Reinforcement Learning for Real-World Production Robotic Systems

    June 08, 2026 · 📺 Zhanpeng He

    Invited talk at ICRA 2026 Workshop on RL in the Era of IL (https://rl4il-icra.github.io/).

    Watch on YouTube ↗

  • Robust Robot Foundation Models for Real-World Manipulation with Jason Ma, Cofounder at Dyna Robotics

    March 09, 2026 · 📺 Humanoids Summit

    In December 2025, Jason Ma, Co-Founder of Dyna Robotics, presented at the Humanoids Summit Silicon Valley 2025, discussing the development of robust robot foundation models for real-world manipulation. In this talk, Jason explores how robot foundation models and embodied AI are enabling robots to perform complex manipulation tasks in dynamic, real-world environments. The presentation highlights the challenges of robotic perception, control, and learning, and how large-scale robotics data can improve the reliability and adaptability of robotic systems. This session provides insights into how…

    Watch on YouTube ↗

  • Actuate 2025: Foundation Reward Models for Robot Learning with Jason Ma

    November 12, 2025 · 📺 Foxglove

    Actuate 2025: Foxglove's developer conference for people who build robots. Jason Ma is co-founder at Dyna Robotics. He received his PhD from University of Pennsylvania, where he studied scalable reinforcement learning algorithms and foundation models for robot learning. His research has been recognized with honors such as RSS Pioneers, Apple and OpenAI Fellowship, Best Paper Finalist at ICRA 2024, Top 10 NVIDIA Research Projects of 2023, and covered by popular media such as the Economist, Fox, Yahoo, and TechCrunch. Featuring: Jason Ma, Co-founder, Dyna Robotics Learn more about Actuate: ht…

    Watch on YouTube ↗

  • He's Building $600M Robots That Fold Your Laundry | Jason Ma, Dyna Robotics

    September 25, 2025 · 📺 Founders in Motion with Thea Ngo

    Meet the $600M startup building AI robots for the real world. Jason Ma, co-founder of Dyna Robotics, shares how his team trains robots to fold laundry, pack boxes, and run 24 hours nonstop. We talk about why AI is the true bottleneck, how to take robots from lab demos to laundromats, and what it means to power the physical economy. Chapters: 00:00 Intro 02:04 What is Dyna solving? 04:30 Choosing a Starting Use Case 06:30 Training the Model 13:50 Why hardware AND software 18:30 Challenges of Real World Deployment 20:48 What is in the Dyna R&D Roadmap? 21:38 Leaving the large tech labs 24:10…

    Watch on YouTube ↗

  • EI Seminar - Jason Ma - Recent Progress on Foundation Model Supervision for Robot Learning

    November 18, 2024 · 📺 MIT Embodied Intelligence

    Title: Recent Progress on Foundation Model Supervision for Robot Learning Abstract: Achieving general-purpose robotics requires robots to quickly learn diverse tasks without extensive training data or hand-engineered controllers for each scenario. While recent efforts in crowd-sourcing robot datasets have expanded available training data, these remain orders of magnitude smaller than datasets used in vision or language foundation models. Rather than solely focusing on scaling robot data, my research develops algorithms that train new and leverage existing foundation models from non-robot dom…

    Watch on YouTube ↗

  • Jason Ma: Foundation Reward Models for General Robot Skill Acquisition

    October 12, 2023 · 📺 Montreal Robotics

    General-purpose robots need to learn and adapt skills in unstructured environments. However, given the diverse nature of robotics tasks and lack of internet-scale data, the recipe of training foundation models, such as GPT-4, is difficult to replicate in robotics. Instead, I posit that foundation models for embodied intelligence should (1) provide actionable information for robots to acquire and refine skills in new environments, and (2) be capable of learning from offline non-robot data. Toward these goals, I will present my recent progress in learning foundation reward models for general rob…

    Watch on YouTube ↗

The summary and quote tags on this profile are produced with AI assistance from verified, first-person interview transcripts, then checked by our team to confirm the speaker's identity and the accuracy of every quote. See how we verify →