Physical Intelligence is a Foundation Models for Robots company founded in 2023 and based in San Francisco, United States. It has raised $1.1B in total funding, most recently a Series B in 2025 at a $5.6B valuation.
| Date | Stage | Amount | Valuation | Lead investors |
|---|---|---|---|---|
| Nov 20, 2025 | Series B | $600M | $5.6B | CapitalG |
No articles ingested yet for Physical Intelligence. Once the hourly news pipeline is live, every article the classifier tags as mentioning this company appears here with its one-line AI summary and sentiment.
Vision-Language-Action (VLA) model designed for generalist robot control, built on large-scale pretraining and flow matching-based action generation to enable dexterous manipulation across different embodiments. Trained on data from 7 robotic platforms and 68 unique tasks, demonstrating strong zero-shot and fine-tuned performance on complex tasks like laundry folding, table bussing, and box assembly. Released as open-source code and weights via the openpi repository. Fine-tuning requires between 1–20 hours of data for most tasks. Available in both JAX and PyTorch implementations.
Vision-language-action foundation model based on π0 that uses co-training on heterogeneous tasks to enable broad generalization. Uses data from multiple robots, high-level semantic prediction, web data, and other sources to enable broadly generalizable real-world robotic manipulation. Optimized for generalizing to new settings rather than accomplishing new skills or exhibiting high dexterity—can perform a variety of tasks in entirely new homes, though it does not always succeed on the first try. Pre-trained on 10k+ hours of robot data.
Vision-language-action foundation model that can perform dexterous tasks across robots, scenes, and skills. π0.7 exhibits emergent compositional generalization, the ability to follow diverse instructions and visual subgoals, and strong out-of-the-box performance even on tasks that previously required fine-tuned specialist models. Steered through multimodal prompting that includes not just language commands but additional context describing manner or strategy, it can follow diverse language instructions in unseen environments, provide zero-shot cross-embodiment generalization such as enabling a robot to fold laundry without seeing the task before, and perform challenging tasks like operating an espresso machine at specialist-level performance.
We don't have a live feed for this company's ATS. Their careers page has every open role.
View all careers ↗