Physical Intelligence
NativeApplied AIPrivateUpdated November 15, 2026The robot foundation model company building AI that understands the physical world.
- Website
- www.pi.website
- Founded
- 2023
- CEO
- Karol Hausman
- HQ
- San Francisco, CA, United States
- Stage
- Private
- Revenue
- Not disclosed
- Total Funding
- $2.1B
- Employees
- 201-500
Executive Summary
Updated Nov 15, 2026
What they do
- Builds general-purpose vision-language-action foundation models that control many types of robots
- Ships the π model family, from π0 and π0.5 to the steerable π0.7
- Offers the Pi API beta for uploading robot data, managed fine-tuning and hosted inference
Why it matters
- Often treated as the research leader in generalist robot models, with top academic founders
- Heavily funded: about $2B raised, with reported talks near a $20B valuation
- Open-weight openpi releases let teams test π0 and π0.5 before committing commercially
Best fit
- Robot makers who want a ready-made robot model instead of building their own
- Automation integrators and fleet operators testing new tasks in manipulation, warehouse or manufacturing settings
- Corporate R&D robotics teams comparing VLA models with open-weight baselines
Watch-outs
- Pi API is a waitlisted beta with no published pricing or SLAs
- Success rates on unseen tasks are self-reported and imperfect, so oversight is needed
- Publicly named production deployments are still few and come mostly from startup partners
Products & Offerings
π0.7
Steerable generalist vision-language-action model, about 5B parameters, that takes language, strategy metadata and visual subgoals for dexterous, multi-step tasks across different robots.
Pi API
Hosted service where customers upload robot demonstration data, fine-tune π models on compute managed by Physical Intelligence, and run the resulting policy on their own robots.
openpi (π0 / π0.5 open weights)
Open-source repository with π0, π0-FAST and π0.5 model weights plus JAX and PyTorch fine-tuning code for robotics researchers and developers.
π0.5
Upgraded generalist robot policy focused on open-world generalization, shown cleaning kitchens and bedrooms in homes it had not seen before.
π0
The company's first generalist robot policy, which trained on data from many tasks and many robots and introduced its flow-matching VLA architecture.
Latest News
AI Magazine · 13 days ago
NeutralExchange4Media · 17 days ago
PositiveHoodline · 17 days ago
NeutralBigGo Finance · 19 days ago
NeutralUpdated Nov 15, 2026
Physical Intelligence is the most credible independent vendor of general-purpose robot models, with strong research, big funding and a new hosted fine-tuning API. Commercial terms, reliability guarantees and enterprise-scale references are still immature. Buyers with robotics programs should run a structured pilot, starting with the open openpi models or the Pi API beta, and compare it against Google DeepMind Gemini Robotics, NVIDIA GR00T and Skild AI before committing to a production rollout.
Signal Provenance
Latest observation date and source for each signal category
| Signal Category | Last Updated | Source |
|---|---|---|
Funding Signals Last funding round, Funding stage | Nov 11, 2026 | Cited web research |
Hiring Velocity Headcount, Headcount growth | Nov 15, 2026 | Research agents (estimate) |
Technical Output Research papers, GitHub activity | Nov 15, 2026 | Hugging Face Hub + news feed + research agents |
News Sentiment News mentions, News sentiment | Nov 15, 2026 | Verified news feed |
Partnership Signals Enterprise partnerships, Distribution partners | Nov 15, 2026 | Research agents (estimate) |
Revenue Indicators ARR estimate, Revenue growth | Nov 15, 2026 | Research agents (estimate) |
Leadership Signals CEO tenure, Founder-led | Nov 15, 2026 | Research agents (estimate) |
Competitive Position Category rank, Market share | Nov 15, 2026 | Research agents (estimate) |
Why this Bluescore?
6 dimensions, each scored 0–10 against all 250 companies. Each adds its weight × score to the Bluescore — see methodology
Physical Intelligence's Bluescore of 60.7 rests most on Talent & Research (8.8, top 5%) and Product Depth (6.7, top 15%). Its weakest area is Customer Adoption (4.0, 9th percentile).
In plain English
Physical Intelligence earns its 60.7 Bluescore mainly from Talent & Research, where it scores 8.8 of 10 at the 95th percentile, and Product Depth, at 6.7 of 10 and the 85th percentile. Customer Adoption, at 4.0 of 10 and the 9th percentile, and Business Model & Revenue, at 4.4 of 10 and the 45th percentile, hold the score back. The biggest lever is Business Model & Revenue, which carries 21% weight and still has 11.8 left, so a stronger score there would move the Bluescore more than any other weak area.
Written by AI from the scores below and checked against them; updated nightly.
Risks
Updated Nov 15, 2026