Physical Intelligence's Pi-Zero Revolutionizes Robotics with $1.1B Funding and Open-Source AI Model

August 17, 2026
Physical Intelligence's Pi-Zero Revolutionizes Robotics with $1.1B Funding and Open-Source AI Model
  • Physical Intelligence aims to build a universal AI brain for robots with the Pi-Zero model, positioned as a general-purpose foundation for robotics.

  • PI has raised about $1.1 billion in two years, including a $400 million Series A in late 2024 at a $2.4 billion valuation and a $600 million Series B in late 2025 at a $5.6 billion valuation, with investors such as OpenAI, CapitalG, NVIDIA, and Sequoia.

  • The physical AI market for robotics was $8.6 billion in 2025, with projections reaching $117.4 billion by 2034; deployment costs have fallen about 35% since 2021 and ROI timelines have shortened to roughly 18–24 months.

  • Pi-Zero’s code was open-sourced in early 2025 as a reference architecture; by mid-2025 Pi-0.5 demonstrated open-world generalization, Pi-0.6 introduced RECAP for demonstration, correction, and self-improvement, and Pi-0.7 in 2026 enabled zero-shot learning for unseen tasks.

  • The competition includes Figure AI and Google DeepMind’s robotics efforts, underscoring a race to own the intelligence layer of robotics beyond hardware.

  • Pi-Zero is a 3-billion-parameter transformer built on Google’s PaliGemma vision-language model, trained on more than 10,000 hours of real-world robot data across seven platforms and 68 tasks.

  • PI was founded in 2024 by prominent AI/robotics researchers, including CEO Karol Hausman, Chief Scientist Sergey Levine, Chelsea Finn, and VP of Engineering Brian Ichter, all with strong academic and industry pedigrees.

  • FAQ material outlines PI’s goals, founders’ credentials, knowledge of Pi-Zero’s architecture, and how PI differentiates itself from traditional robotics through zero-shot learning and rapid R&D execution.

  • Pi-Zero accepts plain-language instructions and converts them into precise motor commands across robot systems, enabling real-time task execution without retraining for each task.

  • The “one brain for every machine” vision advocates a single scalable AI system capable of controlling multiple robot types and tasks, with potential to transform humanoid and industrial robotics markets.

Summary based on 1 source


Get a daily email with more AI stories

More Stories