What is Physical Intelligence?

Physical Intelligence: foundation-model-for-robotics tier

Physical Intelligence (PI) is a foundation-model-for-robotics company building cross-embodiment AI brains for general-purpose robotic control. The company operates at the brain-provider tier per DEPLOY's platform-builder vs brain-provider distinction: PI builds the AI substrate that humanoid OEMs + adjacent platform builders integrate as core embodied control.

---

Valuation: $5.6B confirmed (NOT $11B aggregator drift)

Per Agent A source-depth hygiene pass, PI's confirmed valuation is $5.6B. The aggregator-drifted $11B figure is incorrect; the $10B/$38B figures belong to Project Prometheus (a separate Bezos lab), not to Physical Intelligence. Per DEPLOY's framework, the aggregator-drift correction matters at the cohort-positioning layer: trade-press coverage that frames PI at $11B operates outside the verified-source valuation. The framework-in-action correction is documented at canonical worked-example depth in how DEPLOY corrected the Physical Intelligence valuation.

Series B raised $400M in March 2024. Per Agent A note, the current state may have evolved (~$600M Series B context worth re-verifying against current PI public communications + audited funding-round documentation). Per cap-flag-as-trust-signal, specific current funding state cap-flagged pending primary-source verification.

---

Model lineage: pi0 + pi05_base open + π0.6 + π0.7 closed

PI operates a model lineage with mixed open/closed weights:

• pi0: open weights under Apache-2.0 license. The foundation model for PI's cross-embodiment robotic control work; foundational release establishing the cohort's open-weight position.
• pi05_base: NOW open under Apache-2.0 verified via openpi README per Agent A hygiene pass. The verified isOpen status is editorially substantive: trade-press coverage that frames pi05_base as closed operates outside the verified source.
• π0.6: closed weights; not publicly released.
• π0.7: closed weights; not publicly released.

Per DEPLOY's verified-vs-claimed framework, the model-by-model open/closed designation matters at the verification-posture layer: specific isOpen field assignments require first-party verification (openpi README; PI public communications; Apache-2.0 license artifact in repository) rather than aggregator inference.

---

Cohort positioning + deployment maturity

Per the brain-providers cluster framework, PI anchors:

• Foundation-model-for-robotics tier: cross-embodiment robotic control; humanoid + adjacent platform integration as licensee/partner.
• Open-weights position with mixed releases: pi0 + pi05_base open under Apache-2.0; π0.6 + π0.7 closed (cohort's mixed-license archetype).
• Lab-deployment in limited customer pilots: per Agent A entity record, PI's commercial deployment depth is at lab + limited-customer-pilot tier, not scaled-commercial integration partnerships.
• $5.6B confirmed valuation tier: substantial venture-funded scaling at brain-provider tier; structurally distinct from foundation-model-for-driving (Wayve $8.6B) + foundation-model-for-physical-AI-general (NVIDIA broader strategy) positions.

Per DEPLOY's framework, PI sits at foundation-model-for-robotics tier alongside Skild AI + adjacent brain-provider startups + frontier-AI-lab equivalents (OpenAI Robotics + Google DeepMind Gemini Robotics). The verification posture differential: PI's lab-deployment-in-limited-customer-pilots state is editorial signal at the brain-provider verification layer; integration-partner depth at commercial scale is the forward verification question.

---

Cap-flag discipline: aggregator drift + valuation confusion

Per cap-flag-as-trust-signal, aggregator coverage has produced specific drift patterns on PI that the framework cap-flags explicitly:

• $11B valuation: aggregator-quoted figure NOT confirmed; $5.6B is the verified valuation per Agent A source-depth hygiene.
• $10B / $38B figures: belong to Project Prometheus (separate Bezos lab), not Physical Intelligence; aggregator coverage occasionally conflates the two.
• Series B current state: $400M March 2024 confirmed; potential ~$600M Series B context per Agent A note worth re-verifying against current primary-source disclosures.
• Model-by-model isOpen status: requires first-party verification (openpi README + license artifacts); aggregator inference cap-flagged.

Per DEPLOY's framework, the cap-flag pattern is editorial signal at the verification-discipline layer: trade-press coverage that propagates aggregator-drifted figures should be evaluated against Agent A primary-source-anchored verification.

---

Cohort wiring + contrasts

Per the brain-providers cluster framework, Physical Intelligence anchors the foundation-model-for-robotics tier alongside:

• NVIDIA GR00T: foundation-model-for-physical-AI-general tier; N1 (noncommercial NVIDIA OneWay) + N1.7 (commercial-open) license evolution; no verified production deployment.
• Wayve: foundation-model-for-driving tier; AI Driver + GAIA-3 models; commercial via Stellantis STLA AutoDrive partnership.
• Covariant RFM-1: foundation-model-for-robotic-manipulation tier; corporate-state vs model-state distinct per Agent A discipline.
• Dyna Robotics DYNA-1: foundation-model-for-dexterous-manipulation tier; pilot-not-scale deployment posture.
• Skild AI + frontier AI labs: adjacent brain-provider tier entrants.

For the canonical brain-providers cluster context, see the brain-providers cluster. For the platform-builder vs brain-provider tier distinction framework, see brain-provider vs OEM platform tier distinction. For the canonical category umbrella, see what is physical AI. For methodology canonical references applicable to Physical Intelligence: captive vs third-party brain providers (third-party foundation-model-for-robotics tier) + the 9-tier source-quality rubric (PI IR + reputable-press + entity-distinction with Project Prometheus source classification).