Building robots that show up every day
His parents ran a restaurant. Now his robots fold napkins in one.
Lindon Gao arrived in the United States at age 11, the son of a father who had worked long enough to earn citizenship and bring his family over from China. The family settled into the unglamorous grind of American immigrant life - long hours, physical work, meals cooked and served rather than enjoyed. It is not a coincidence that Dyna Robotics, the company Gao co-founded in 2024, deploys robots in exactly these environments: hotels, restaurants, laundromats, gyms. Places that chew through human labor and spit out exhaustion.
Gao has the unusual clarity of someone who grew up watching the jobs he now automates. He is not theorizing about "the future of work." He grew up inside it.
"I want to prove to myself that I'm not a one hit wonder."Lindon Gao, on founding Dyna Robotics after the Instacart acquisition
The first hit, for context, was substantial. In 2021, Instacart acquired Caper AI - the company Gao co-founded to put computer vision inside grocery carts - for $350 million. It was the kind of exit that turns founders into passengers. Gao became VP, General Manager, In-Store at Instacart, leading the team that digitized physical grocery stores for chains like Kroger, ShopRite, and Aldi. Then he left.
The departure was not impulsive. It came after what Gao has described as a period of genuine reckoning - about what comes after a defining win, about the psychological weight of starting over with "some sort of expectations and a track record." He did not want to coast. He wanted to find out if the first exit was luck or judgment.
Fourteen years of not stopping
Gao founded his first business at fourteen, trading gaming collectibles. He sold it two years later. By nineteen, he had built LPG Crafts - a jewelry supply chain management company - into a multi-million dollar enterprise. These are not "early passion projects" that look good in a bio. They were operating businesses, with customers and cash flow, run by someone still in school.
At NYU Stern, he studied Finance and International Business. After graduation, he took the predictable path - Goldman Sachs, investment banking, 2.5 years - and then stepped off it. The discipline was useful. The ceiling was not. He moved to Y Combinator's W16 cohort with QueueHop, an app-enabled anti-theft apparel tag that raised $150,000 in seed capital. QueueHop stalled, but the problem it pointed at - the friction of physical retail - did not go away. Gao and his co-founder York Yang kept pulling at the thread. Caper AI was what came out.
"Save every penny because the company almost died twice along the way."Lindon Gao, on the Caper AI years
Caper's model was elegant: replace the checkout line with a cart that already knows what's inside it. Cameras, weight sensors, machine learning. No scanning. No cashier. Gao sustained himself through two years without salary - including a period he credits to re-reading Marcus Aurelius's Meditations. The stoic's line that "you can't change the world, you can only change your perception of the world" became a private compass during the months Caper nearly did not survive.
There's a specific Caper story worth noting. A 79-year-old grandmother became one of the company's most loyal customers. She was not in any of Gao's target demographic assumptions. He revised his understanding of the product immediately. That revision - speaking directly with customers, watching actual behavior rather than trusting the slide deck - would become central to how he built Dyna.
The robots are already working
Dyna Robotics launched publicly in March 2025 with a $23.5 million seed round led by CRV and First Round Capital, emerging from stealth at a roughly $100 million valuation. The company's three co-founders cover an unusual spread: Gao brought the hardware business experience; York Yang, his co-founder from Caper, brought the engineering; Jason Ma, a former Google DeepMind research scientist and lead author of the Eureka robotics research paper, brought the foundational AI research.
The pitch was not humanoid robots. It was not general-purpose AI. It was a specific, honest claim: Dyna's robotic arms - powered by the DYNA-1 foundation model - could do one class of jobs reliably, repeatedly, and cheaply enough to make commercial sense today.
"Getting to general-purpose robots is not going to happen as quickly as the industry hoped."Lindon Gao
This is the rare robotics pitch that leads with limitation. Gao openly states that foundation models currently run at 10-30% of human-level efficiency. He is not selling the future. He is selling the present - a 99.4% success rate folding napkins in a restaurant for 24 straight hours, with no human intervention, "straight out of the box" in any environment. DYNA-1 requires zero setup time when moved to a new location. It generalizes across tasks and physical settings by design, learning from every deployment.
After speaking with hundreds of potential customers before launch, Gao revised the company's core thesis. "Initially, we thought that people wanted robot employees," he said. "But in speaking to hundreds of customers, we realized that a majority of the people and workforces don't really need a robot employee." What they need is reliable task automation - a robot that handles one category of repetitive, dirty, or dangerous work without calling in sick.
By September 2025, Dyna had raised a $120 million Series A led by Robostrategy alongside CRV and First Round Capital. Additional backers included NVIDIA Ventures, the Amazon Industrial Innovation Fund, Salesforce Ventures, Samsung Next, and LG Technology Ventures. The company's valuation exceeded $600 million. It had grown from 30 employees at launch to 120 within a year.
What the numbers say
Foundation model robots operate at a fraction of human speed - but Dyna's goal is not to impress a benchmark. It's to run a shift.
How Gao thinks
Good engineering is still going to ultimately win.
A strong foundation model is key to scalable distribution. Our models continuously improve with each customer deployment, generating high-quality data.
We are the only company today that is in production environment with dexterous manipulation.
The first time you have no baggage. But now I have some sort of expectations and track record.
Right now the speed of foundation models is around 10-30% of human-level efficiency.
You can't change the world, you can only change your perception of the world.
The thread from there to here
What he has actually done
Sold Caper AI to Instacart for $350 million in 2021 - one of the largest AI-hardware retail acquisitions of its era
Forbes 30 Under 30 - Retail & Ecommerce (2020), recognized alongside Caper co-founders
Raised $143.5M in total funding for Dyna Robotics within its first year of operation
Built Dyna to a $600M+ valuation in under 12 months - backed by NVIDIA, Amazon, and Salesforce
Shipped DYNA-1: the world's first commercially deployed dexterous manipulation foundation model
Y Combinator W16 cohort - launched QueueHop that pivoted into the Caper AI thesis
Specific and true
A 79-year-old grandmother became one of Caper AI's most loyal customers. She was not in Gao's target demographic. He revised his assumptions about technology accessibility immediately - and carried that lesson into Dyna's customer discovery process, where he personally interviewed hundreds of operators before settling on a product thesis.
Caper AI nearly died twice. Gao kept it alive by treating every dollar as irreplaceable - "save every penny" is not a cliche from him, it's what actually happened. He went two years without salary and credits Marcus Aurelius's Meditations as the thing that kept him from quitting.
His parents worked in a restaurant when the family first arrived from China. His robots now fold napkins in restaurants for 16 hours a day. He does not appear to have made this observation publicly. It does not need to be made out loud to carry weight.
Before founding Dyna, Gao described the psychological difficulty of starting over after a major exit: "The first time you have no baggage." This is the rarer kind of founder honesty - not manufactured humility, but the actual texture of what it feels like to start again with something to lose.
Lindon Gao on Building Robots
Lindon Gao (Dyna) Is Building Robots To Do Your Dishes - YouTube
The smaller truths
Wakes at 5 AM, sleeps by 10 PM. Separates his calendar into "focus days" and "manager days."
Sustained by Marcus Aurelius's Meditations during two years without a salary at Caper AI.
Founded his first company at age 14 trading gaming collectibles - and sold it profitably at 16.
Stuyvesant High School alumnus - one of New York City's most academically selective public schools.
DYNA-1 folded over 700 napkins in 24 hours without stopping. That's a napkin every two minutes, continuously, all day.
Once an investment banker at Goldman Sachs; left after 2.5 years to pursue Y Combinator with an anti-theft clothing tag.