The startup that proved humanoids can actually work for a living.
Figure AI is a Sunnyvale, California-based humanoid robotics startup backed by OpenAI, Microsoft, NVIDIA, Intel Capital, and Parkway Venture Capital. While most humanoid companies are still in the demo phase, Figure has actual commercial deployments — BMW’s Spartanburg manufacturing plant runs Figure 02 units on production lines.
What They Build
| Robot | Status | Key Feature |
|---|---|---|
| Figure 02 | Commercial deployment | 16-DOF hands, BMW factory integration |
That’s it for now — one robot. But Figure 02 is one of the few humanoids doing real industrial work at scale.
The Investor List Matters
Figure AI has raised significant capital from:
- OpenAI — Strategic partner, also provides AI models
- Microsoft — Cloud infrastructure and enterprise distribution
- NVIDIA — Compute hardware and AI training infrastructure
- Intel Capital — Manufacturing and supply chain expertise
- Parkway Venture Capital — Deep-tech robotics focus
This backing gives Figure access to cutting-edge AI models, compute infrastructure, and enterprise channels that competitors lack. It’s not just money — it’s the full stack needed to build intelligent robots.
Production Scale
As of May 2026, Figure AI scaled production from one robot per day to one per hour.
At one per hour, that’s approximately 8 robots per day, or ~2,000 per year if sustained. That production rate indicates they’ve solved significant manufacturing challenges — supply chain, assembly, calibration, quality control — that other humanoid companies haven’t cracked yet.
Autonomy Approach
Figure 02 operates at a Hybrid autonomy level. It performs routine manufacturing tasks autonomously — material handling and assembly with minimal human intervention. The AI handles object recognition, grasp planning, motion execution, and error recovery.
Human operators remain in the loop for task assignment, exception handling, and safety oversight. Figure calls this “AI-assisted autonomous operation” rather than full autonomy.
The company is actively working toward greater autonomy. The production ramp suggests they’re moving from prototype to scaled deployment, which typically requires more robust autonomous capability.
Combat Relevance
Figure AI has shown no interest in combat entertainment. But the platform’s capabilities are relevant:
- Hand dexterity: 16 DOF hands could manipulate weapons or defensive equipment
- Vision system: Object recognition and tracking could identify opponents
- Autonomous operation: The AI stack could theoretically be retasked for tactical decision-making
- Durability: Industrial deployments prove sustained physical stress handling
The primary barrier to combat application is software, not hardware. Figure’s neural networks are trained for manufacturing, not combat. Retraining would be required.
The broader point: Figure AI is building humanoids for practical applications. The combat leagues (UFB, URKL) are using different, less advanced hardware. The gap between the best humanoid technology and the best combat humanoid is significant.
What They Don’t Do
- No combat demonstrations
- No combat-specific software
- No stated interest in entertainment robotics
- Focused exclusively on manufacturing and logistics
Timeline
| Year | Milestone |
|---|---|
| 2022 | Figure AI founded |
| 2023 | Figure 01 prototype unveiled |
| 2024 | Figure 02 announced |
| 2025 | BMW Spartanburg deployment begins |
| 2026 | Production scaled to 1 per hour |
Related
- Figure 02 — Their deployed humanoid robot
- Boston Dynamics Atlas — More dynamic, industrial-focused
- 1X NEO — Domestic-focused competitor
- Robot Database Hub — Compare all platforms
Last updated: May 2026 | Status: Active commercial deployment | Autonomy focus: Hybrid manufacturing