FieldAI
Foundation model
Irvine robotics-software lab building physics-first 'Field Foundation Models' that let any robot navigate unstructured, GPS-denied environments.
FieldAI develops Field Foundation Models (FFMs), general-purpose embodied-AI 'brains' for robots operating in unstructured outdoor and industrial environments without GPS, prior maps or teleoperation. Unlike LLM-derived models, FFMs are built physics-first and are intrinsically risk-aware, reporting their own confidence so customers can set safety thresholds - letting one brain generalize across robot types (humanoids, quadrupeds, vehicles) and tasks. The company traces to NASA JPL and DARPA autonomy programs (Subterranean Challenge, RACER) and reports deployments across hundreds of complex industrial environments, including defense and infrastructure use cases. (Company-reported facts.)
Research mapping
Physical AI / Sim-to-Real, Digital Twins & Validation (L9) / Foundation model
Evidence boundary
Original analysis and independent corroboration remain under development. Commercial scale and durable technical advantage remain under review.
What makes it defensible
Physics-grounded, risk-aware foundation models built from JPL/DARPA field-robotics programs - generalization with quantified confidence, the piece LLM-derived robot models struggle to provide.
Position in the stack: Embodiment-agnostic autonomy software layer (robot brain) licensed to robot makers and operators across defense, industrial and infrastructure end markets.
Who buys it
Founding team
Ali Agha (Ali-akbar Agha-Mohammadi) — Founder & CEO — Former NASA JPL robotics technologist and group leader; Principal Investigator on DARPA's Subterranean Challenge and RACER programs.
SourceFinancing detail
- Last round
- Series A
Investors: Bezos Expeditions, Khosla Ventures, NVentures (NVIDIA), Temasek, Canaan Partners, Intel Capital, BHP Ventures, Emerson Collective, Prysm, Gates Frontier, Samsung
Mapped funds and investors
Direct relationships supported by published evidence.
Reported holdings as-of
No sourced relationships mapped yet.
Disclosed investment history
No sourced relationships mapped yet.
Showing up to four relationships in each group. ETF weights describe the fund’s portfolio, not ownership of this company. Private investment history does not establish current ownership.
Evidence & media
-
Can physics stop robots hallucinating? — FieldAI's Ali Agha (podcast)
What this source supports: Third-party interview: FieldAI runs a physics-based + data-driven FFM on-robot (no cloud/GPS); software operating across 34 robot embodiments.
What it does not establish: Interview; claims not audited.
Source -
How AI is teaching robots to navigate the real world (YC ECDemoDay talk)
What this source supports: Third-party demo-day talk: shared "brain" across a fleet, co-learning across sites, deployments in difficult dynamic domains.
What it does not establish: Founder pitch; no independent validation.
Source -
FieldAI nears $700 million round at $10 billion valuation (follow-on coverage; round not confirmed closed)
Source
Timeline
-
Financing
Series A — $405,000,000 total disclosed
Source