Figure AI
448 words · 3 min read · 2 sources
Figure is a Bay Area startup building general-purpose humanoid robots for the workplace. As of 2026, it's one of the leading commercial humanoid programmes outside Tesla and Boston Dynamics — and arguably the most focused.
The company concept: Figure is a Bay Area startup building general-purpose
Difficulty 3/5 · ClassroomFigure is a Bay Area startup, founded in 2022, building general-purpose humanoid robots for the workplace. As of 2026, it's one of the leading commercial humanoid programmes outside Tesla and Boston Dynamics — and arguably the most focused.
💡 Think of it like…
Think of it like a household object that does the same job — the underlying idea is the same, just adapted for robots.
Why it matters
Without figure ai, many company systems in robotics simply couldn't work.
Figure is a Bay Area startup, founded in 2022, building general-purpose humanoid robots for the workplace. As of 2026, it's one of the leading commercial humanoid programmes outside Tesla and Boston Dynamics — and arguably the most focused.
The founder and the bet
Brett Adcock founded Figure after exits from Vettery (HR tech, sold to Adecco) and Archer Aviation (electric aircraft). His pitch to investors in 2022: humanoids are the next platform shift, the same way smartphones were in 2007 — and a focused startup can move faster than Tesla or any big company.
He raised an unusually large early-stage round — $675 million by early 2024 — from Microsoft, NVIDIA, OpenAI, Jeff Bezos, and others. That's about 10× the typical Series B for a hardware startup. The argument from investors: humanoids will be a winner-takes-most market, and capital is the moat.
The three products
Figure 01 (2023) — the first prototype. About 1.7 m tall, 60 kg, 28 actuators. Public demos of it walking and making coffee.
Figure 02 (2024) — the commercial prototype. Stronger actuators, better hands (16 degrees of freedom), onboard AI compute. Started a pilot at BMW's Spartanburg plant doing component placement in 2024-2025.
Figure 03 (2025-2026) — the production version. Better dexterity (the famous laundry-folding demo). VLA-based learning architecture. Targeting tens of thousands of units in deployment by 2027.
The OpenAI partnership (and its end)
In 2024, Figure signed a high-profile partnership with OpenAI — OpenAI would build the brain, Figure would build the body. The partnership ended in early 2025. Figure brought AI in-house.
The story given: Figure wanted faster iteration than the partnership allowed. The story not given: a humanoid's intelligence and embodiment are too tightly coupled to separate cleanly.
This is now the prevailing pattern. Tesla, Figure, Boston Dynamics, 1X — every serious humanoid company is vertically integrated on AI.
How Figure is different from Tesla
- Smaller, faster, more focused. Single product. ~150 employees.
- Enterprise-first. Figure sells to BMW, OpenAI, and others as a B2B robot. Tesla's Optimus is positioned for both consumer and B2B.
- No car business. Figure is purely robots — no division of attention, no revenue cushion.
The bet on Figure is essentially that focus beats vertical integration when both sides have similar AI capabilities.
What's at stake
If Figure's commercial pilots scale — if BMW reports that Figure 03 reliably saves money on the factory floor — Figure becomes the proof point that humanoids can be a business, not just a research demo. That moment is probably within the next 18 months.
See Figure 03 folds laundry for the most recent Figure milestone in detail.
Ask R2 Co-pilot anything you didn't understand about Figure AI. It'll explain it plainly.
Keep going
Atlas (Boston Dynamics)
Atlas is Boston Dynamics' bipedal humanoid robot — the most acrobatic robot ever built, and now an electric ma…
CompanyBoston Dynamics
Boston Dynamics is the American robotics company famous for making the most agile and acrobatic robots on Eart…
ConceptHumanoid robot
A humanoid robot is a robot built in the rough shape of a human body — two legs, two arms, a head with cameras…
Last updated · 2026-05-19
Community discussion
0 questions & insightsLoading discussion…
Spotted something off? Report an error →