Chelsea Finn: This is the State of the Art in Robotics
Chelsea Finn: This is the State of the Art in Robotics
Podcast58 min 17 sec
Listen to Episode
Note: AI-generated summary based on third-party content. Not financial advice. Read more.
Quick Insights

Investors seeking direct exposure to the emerging physical AI and robotics sector should monitor upcoming private market developments for Physical Intelligence, as it is not currently publicly traded. For immediate public market exposure to autonomous physical AI, accumulate shares of Alphabet (GOOGL / GOOG) to capitalize on the proven commercial scaling of its Waymo subsidiary. Watch for public software-hardware integration firms and enterprise deployment partners utilizing open-source robotics foundations like the Pi series. Investors should track these partners over the medium term to identify publicly traded value chains benefiting from advanced robotics. Maintain a diversified approach to robotics and autonomous systems as these foundational technologies mature.

Detailed Analysis

Physical Intelligence (Private Company)

  • Physical Intelligence is a robotics AI company founded by Chelsea Finn two years ago, focusing on developing general-purpose robots capable of performing various real-world tasks using advanced foundation models and reinforcement learning.
  • The company recently developed the Pi07 model, which matches or exceeds the performance of specialized, fine-tuned robotics models across multiple downstream tasks without requiring task-specific fine-tuning.
  • Physical Intelligence models have been successfully applied to real-world workflows such as making lattes (achieving over 90% reliability over extended runs), constructing and labeling cardboard boxes at a chocolate factory, folding laundry, cleaning kitchens, and using a knife to slice vegetables.
  • The company integrates reinforcement learning with human interventions to avoid dead-end trajectories, and utilizes a general-purpose value function to improve data efficiency and model reliability.
  • Key technology includes a multi-timescale memory system allowing robots to execute long-horizon, non-repetitive tasks (such as a 10-to-15 minute kitchen cleaning sequence) autonomously.
  • The company's Pi05 and Pi models are open-sourced to build an ecosystem and assist developers and startups in fine-tuning applications.

Takeaways

  • Physical Intelligence represents a pioneering player in the physical AI and robotics sector, though it is currently a private company and not directly available for public stock market investing.
  • Investors looking to gain exposure to the broader physical AI and robotics trend can watch for open-source robotics foundations, software-hardware integration firms, and enterprise deployment partners utilizing these technologies.

Waymo (Alphabet - GOOGL / GOOG)

  • Waymo reached a milestone of a quarter of a million weekly autonomous rides a year prior to the discussion, demonstrating that machine learning-based systems can operate in a trustworthy and autonomous way directly in the physical world.
  • Waymo's success serves as an industry proof-of-concept that long-term autonomy and reliability in physical AI systems are achievable.

Takeaways

  • Waymo's progress highlights the commercial viability of autonomous physical AI systems.
  • As a subsidiary of Alphabet (GOOGL / GOOG), Waymo offers public market investors indirect exposure to leading-edge autonomous driving and physical AI deployment.
Ask about this postAnswers are grounded in this post's content.
Episode Description
Robots can already fold laundry, make espresso, clean kitchens, and assemble things. The harder problem is getting them to do those tasks reliably, for long periods of time, without a human babysitting them. At Startup School 2026, Physical Intelligence cofounder Chelsea Finn explains what it takes to build general-purpose robots that work in the real world. She shares how reinforcement learning pushed robot throughput up 2x, how their systems can run autonomously for hours, and why she believes robotics is entering its GPT era: moving from specialized models toward general-purpose systems that can work across tasks, robots, and environments. Transcript: https://www.ycrootaccess.com/p/chelsea-finn-on-the-next-decade-in
About Y Combinator Startup Podcast
Y Combinator Startup Podcast

Y Combinator Startup Podcast

By Y Combinator

We help founders make something people want.