In this episode, Sunny Ray sits down with Anto Patrick, co-founder of Cosmic Brain, to trace his path from a childhood spent building robots and reading his mother's physics textbooks to leading a startup focused on training humanoid robots through refined video data. Anto shares stories of his early inventions, including a smart water bottle that was eventually used by the military, and his time working on one of Elon Musk's large scale AI infrastructure projects. He explains how Cosmic Brain acts like a refinery, turning raw human action data into skills robots can learn, aiming to build an app store style marketplace for robotic capabilities. The conversation moves into big picture questions about whether robots can help eliminate poverty, what safeguards are needed as AI takes on humanoid form, and why Anto believes job loss fears are overstated. Throughout, he reflects on lessons learned from working near Elon Musk and what responsible development of physical AI should look like as the technology accelerates.
Cosmic Brain co-founder Anto Patrick on turning human skills into robot data and why humanoids could end poverty, not jobs.
What's your earliest memory of building things, and when did you start tinkering?
I got my first motor and battery at age seven, and that's when I started building. My first project was a small amusement park with roller coasters and a giant wheel. I also built a robotic head that could move and talk, kind of like a Siri. My mom being a physics teacher meant I could always borrow, or steal, her textbooks and just start experimenting from there.
How did the robot fight club come about?
In early 2025 there was a lot of bias in the Bay Area, people didn't believe humanoid robots were real outside research labs. A few friends and I wanted to break that taboo, so we took robots for walks in San Francisco streets and eventually built a caged fight with two humans controlling the humanoids. It drew huge attention and let us actually explain what we were building at Cosmic Brain.
What was your aha moment connecting AI and robotics?
I don't think there's one aha moment, it's an evolutionary domino effect. Neural networks led to transformers and LLMs, which led to diffusion models, then world models, then physics engines and robotics breakthroughs, mostly happening in 2025. When I saw China manufacturing robots at scale while the US built the intelligence layer, I knew it was time to jump in and become an early believer.
What did you learn from working alongside Elon Musk?
Elon is one of the few people passionate enough to build things most people don't care about, rather than just chasing money. He had both the scientist's mind and the resources to make things real, even when nobody believed in him for over a decade. He pushes teams to treat physics as the only real constraint and to build things that have never been done before.
What is Cosmic Brain and how did it start?
It started after conversations with Sam Altman, Elon, and others at the DC inauguration in early 2025 about America needing robotics to solve its labor crisis. I quit my job, got angel checks from parties in Washington DC, and teamed up with my friend Chi, who spent over 20 years at Google, and Ed Miner, an early YouTube engineer who understood large scale video data.
Can you explain what Cosmic Brain actually does?
Think of it like an oil refinery, but for data. We collect huge amounts of human action video from around the world and pass it through our AI translation layer, which cleans and converts it into a format robots can understand, essentially a video to robot transfer protocol. The end goal is an app store style skills marketplace where you download abilities like cooking or mowing the lawn into your home robot.
Do you believe robots can really help eliminate poverty, as Elon has claimed?
Yes, it's definitely possible if we invest enough and move fast enough through the technology and regulatory bottlenecks. Robots could build houses, clean streets, and do labor for free once deployed, driving down the cost of goods and housing. A family could even have a robot nanny, freeing up money and time so people can live more creatively and comfortably.
How do you think about the ethical risks and safeguards as robots take human form?
I think about it constantly, because for the first time superhuman intelligence has a human form that can act and behave like us, which is dangerous in the wrong hands. That's why we're building something we call the ten commandments for robots, so they can't harm humans or disobey commands. Whoever controls and refines the data ultimately controls the ethics and guardrails of robotics.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview