Figure AI: The Startup Building the Robotic Workforce of the 21st Century

Back to: Top 10 US Startups 2026

The history of industrial robotics is a history of specialization. The robotic arms that weld car frames, the automated guided vehicles that shuttle shelves at Amazon fulfillment centers, the CNC machines that cut precision parts for aerospace manufacturers — each is extraordinarily capable within its designed domain and extraordinarily limited outside of it. The dream of a general-purpose robot capable of performing the full range of physical tasks that humans perform has been the holy grail of robotics for fifty years.

Read more: UK Regulator Probes TikTok's Content Moderation Practices

Figure AI, founded in 2022 by Brett Adcock — who previously founded the aviation startup Archer Aviation — is pursuing that holy grail with a focus and set of technology partnerships that no previous robotics company has assembled. With $2.6 billion in funding from Microsoft, OpenAI, Nvidia, Jeff Bezos, and the venture arms of Intel and LG, plus a manufacturing partnership with BMW and an AI partnership with OpenAI, Figure is building the foundational platform for a new kind of labor market.

Company Overview

Figure AI is headquartered in Sunnyvale, California, with its primary manufacturing facility in San Jose. The company employs approximately 300 people with expertise in mechanical engineering, computer vision, motion planning, and large language model integration. Adcock's leadership style emphasizes rapid prototyping and willingness to deploy robots in real production environments before they are fully polished — real-world feedback is irreplaceable, and waiting for perfection is a luxury the market will not allow. The Figure 02 robot stands approximately five feet six inches tall, weighs around 70 kilograms, and can lift objects up to 55 pounds. It navigates via onboard cameras and LIDAR, manipulates objects with multi-fingered hands, and communicates with human co-workers in natural language via a voice interface powered by OpenAI models.

Business Model

Figure's commercial model is built around robot-as-a-service: rather than selling robots outright, Figure leases them on multi-year contracts including ongoing software updates, maintenance, and performance guarantees. This lowers the adoption barrier for customers reluctant to make large capital commitments on unproven technology, creates a recurring revenue stream for Figure, and maintains the company's ongoing relationship with its deployed robots — generating real-world training data necessary to continuously improve capabilities. The BMW partnership, Figure's flagship commercial deployment, involves Figure 02 robots working in BMW's Spartanburg, South Carolina assembly plant, performing material handling tasks: moving components from storage to the assembly line, loading and unloading parts, and performing quality checks on finished components.

Read more: xAI and Grok: How Elon Musk Is Betting $50 Billion on an AI Moonshot of His Own

Innovation Factor

What distinguishes Figure from earlier humanoid robot projects is not primarily mechanical sophistication but the quality of AI giving it general-purpose environmental understanding. The partnership with OpenAI has given Figure access to the most capable multimodal language models in existence, enabling Figure 02 to look at a work environment, understand verbal instructions from a human supervisor, reason about the sequence of steps required to complete a task, and execute that task physically. This embodied AI capability — making a robot that has learned one task in one factory able to perform a different task in a different factory with minimal additional training — is the key to the market scale that Figure's investors are betting on. The robotics research community calls this physical intelligence, and it is what makes Figure's approach fundamentally different from traditional industrial robotics, which requires precise programming for every specific task in every specific environment.

Market Position

Figure competes in an emerging market that did not exist five years ago. Its primary competitors are Tesla's Optimus program, Boston Dynamics' Atlas platform, and well-funded Chinese humanoid startups including Unitree and UBTECH. Tesla's entry is particularly significant given its manufacturing scale and AI experience through its automotive business. But Figure has a head start in actual commercial deployments that Tesla has not yet matched. The broader market opportunity is staggering: global manufacturing employs approximately 300 million people; logistics and warehousing employs hundreds of millions more. If humanoid robots can perform even a fraction of these tasks at costs competitive with human labor — a threshold Figure's economics suggest is achievable this decade — the market would dwarf any previous category of industrial equipment. See also: Waymo and Zipline on how autonomous systems reshape physical operations across industries.

What's Next

Figure's roadmap centers on scaling its BMW deployment, expanding to additional manufacturing customers in automotive and electronics, and deploying Figure 03 — with improved dexterity, longer battery life, and more sophisticated manipulation capabilities. The company is also developing a proprietary AI foundation model specifically optimized for physical manipulation tasks, what Adcock calls a physical intelligence model that will eventually run entirely on the robot's onboard compute without requiring constant cloud connectivity. The longer-term vision — deployment of millions of Figure robots across the global economy — depends on solving the remaining hard problems in robot dexterity and environmental generalization, and doing so before Tesla, Boston Dynamics, or a well-funded Chinese competitor beats them to market at scale.