Language models are celebrated for their spikes. Robots are judged by their weakest link, because there are so many layers between intelligence and real-world impact.
Models, hardware, reliability, operations. Everything has to work. Every single day.
Dyna is here to show how it’s done.
BREAKING: Inside the first company to achieve 99%+ success rates with a robot foundation model and now the first company with working physical agents.
@DynaRobotics co-founder and CEO Lindon Gao (@Lindon_Gao) sat down with @HerbertOng, @GoingBallistic5 and @Rewkang to talk about US-made robots already working real shifts at hotels, restaurants and data centers.
Lindon says customers send Dyna a video of the task, and a new site running a task Dyna already knows can be up in as little as three days. Its robot Taku can now run an entire laundry room workflow end to end, so staff can walk away.
Lindon on where to see them:
"If you go to some of the Best Westerns and Hiltons in California, you'll be able to see our robots at work. So yes, robot is ready for production in prime time."
Lindon on customer demand:
"Your customers are consistently texting you and asking you, when can I get my next batch of robots?"
Timestamps
(00:00) US-made robots already working today
(03:40) Ready for production prime time
(07:18) Acquired by Instacart, scaled to hundreds of cities
(10:55) Robotics is defined by your weakest link
(13:44) The learning loop must spin fast
(15:11) We needed to build our own hardware
(20:22) 99% success is just the starting line
(21:28) Taku completes the entire job end to end
(25:46) Observability to root-cause issues remotely
(27:04) First physical agent shipped in the field
(31:54) Teach the robot in 10 minutes
(33:23) Taku rang the NASDAQ bell
(34:30) 2027 is going to be about deployments
(38:16) Robot as a service, a monthly fee
(39:02) Deployed zero-shot, no extra post-training
(41:11) Scalable across 100,000 US locations
(43:40) If we nail laundry, we nail anything
(49:18) The benchmark is real-world value
(52:47) US ahead in models, China in hardware
(54:28) Ship hundreds of thousands of robots
(55:39) Manufacturing sites in California and Texas
(1:00:59) I thought high-quality data was everything
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me.
I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA.
I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation.
Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO.
Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours.
Jensen
Last month, we hosted @JasonMa2020 (cofounder & chief scientist @DynaRobotics) at @savant_vc to chat about deployment. Learned a ton about distribution, early versions of Dyna, hard lessons, and more.
00:26 From a PHD to NVIDA to co-founding a company
01:51 Why start with a task as simple as folding napkins?
04:40 Do you see Dyna as a model or deployment company?
07:30 Partnership with Din Tai Fung
09:10 How did you hit those numbers?
10:59 What are the hardest lessons you didn't expect in deployment?
13:11 Besides deployment, what other moats would you face?
14:15 What is the core and why should customers care?
16:20 Dyna 2.1
19:04 The Multitasking Vision
19:54 How important is it to own your own embodiment?
20:58 Incontext learning ability with Dyna 2.0
22:32 Does it take a lot of custom engineering to skill it?
24:05 Thoughts on Astra?
25:30 Do you see Astra and similar models as a threat?
26:56 What's after Dyna 2.1?
27:49 Any deployment opportunities you guys are trying to find more value in?
28:58 What are the main challenges with moving into this new phase?
31:19 Do you foresee the need to expand?
31:54 In 5 years, will there be robots that maintain other robots?
33:19 Everyone is building an agentic layer. Is it still possible to build only one part?
35:38 Where are Vams more relevant?
37:44 What do you look for in researchers?
39:59 What does it take to get to 99% success rate?
Managed to make it out of Austin after CoRL on my second attempt Saturday night! Lots of roboticists got stuck 😅 Silver lining is I got to hang out a bit more with one of my favorites @ChenTessler, who also got stuck
Anyways, here is my ranking of the best demos I saw at the conference
🥇Taku from @DynaRobotics
This was easily the conference favorite. It was such a clean demo, it just worked, and it worked so well it was easy to imagine it being deployed in the real world! No gimmicks, no tricks, they just let it run. Props to the team, they’re still killing it.
Zero-shot long-horizon, whole body manipulation is here. Heard from many people that this is the most impressive demo they saw at CORL this year :) Come check it out!
Yesterday, we unboxed Taku at @corl_conf put it to work.
Powered by Dyna-2.1, it performed long-horizon, whole-body mobile manipulation zero-shot, with no on-site data collection or post-training.
Come see it live and meet the team at Booth 500 in Griffin Hall!