Figure AI Enters 2026 as Humanoid Robotics' Best-Capitalized Bet — a $39B Valuation, Its Own AI Brain, and a Factory Race

Figure AI Enters 2026 as Humanoid Robotics' Best-Capitalized Bet — $39B Valuation, Its Own AI Brain, and a Factory Race

SUNNYVALE, Calif. — When Figure AI told investors last autumn it had blown past $1 billion in committed capital at a $39 billion post-money valuation, the number was startling enough on its own: a roughly 15x jump in 18 months for a company that had yet to sell a robot at scale. What made it land harder was the cast list behind it — Nvidia, Microsoft, Intel Capital, Salesforce, Qualcomm, LG, T-Mobile and Jeff Bezos's Bezos Expeditions, all writing checks into a humanoid-robot startup that, three years earlier, founder Brett Adcock had seeded largely out of his own pocket.

That Series C, led by Parkway Venture Capital and closed on September 17, 2025, is the round that defines Figure's standing as it moves through 2026 — not the $48 billion "Series D" that has circulated in some quarters. The Vault could not verify any closed Series D at a $48 billion valuation; the most recent confirmed financing remains the $39 billion Series C. What is verifiable is more interesting than a single headline number: Figure has become the best-funded pure-play in a humanoid-robot field that is suddenly awash in capital, and it is spending that money on two things — a proprietary AI model and a factory — that will decide whether the valuation holds.

The money, and who is behind it

"This milestone is critical to unlocking the next stage of growth for humanoid robots, scaling out our AI platform Helix and BotQ manufacturing," Adcock said when the round was announced, framing the raise less as a vanity valuation than as fuel for two specific bets.

The round pushed Figure's total funding above $1.9 billion and was notable for the strategic logic baked into its investor list. Nvidia is both a backer and a supplier, with the capital earmarked in part to build GPU infrastructure for training Figure's models. Microsoft, Salesforce, Qualcomm, LG and T-Mobile each bring an enterprise or hardware angle, while Brookfield Asset Management — which owns vast portfolios of warehouses and real estate — represents exactly the kind of physical environment Figure wants its robots to work in. Adcock, who founded Figure in 2022 after building recruiting startup Vettery and co-founding eVTOL maker Archer Aviation, self-funded the company's earliest stage before the outside money arrived.

Helix: building its own brain

The most consequential technical story at Figure is Helix, the in-house vision-language-action (VLA) model the company began rolling out in 2025 to replace its earlier reliance on OpenAI for robot reasoning. Helix runs on the robot's onboard compute, allowing a Figure machine to take a plain-language instruction and translate it into real-time, continuous control of its upper body — wrists, torso, head and individual fingers — without a round trip to the cloud.

Figure has touted Helix as the first VLA capable of high-rate full-upper-body control and the first to run two robots simultaneously, letting them coordinate on tasks neither could finish alone. In January 2026 the company introduced Helix 02, extending the model toward full-body autonomy — walking, balancing and manipulating as a single continuous system rather than a stack of separate controllers. The strategic point is independence: by owning the model the way it owns the hardware, Figure is betting that the data flywheel from its own deployed fleet, not a licensed third-party brain, becomes its durable advantage.

From pilot to production

The commercial proof point is BMW. Figure's previous-generation Figure 02 ran an 11-month deployment at BMW Group's Plant Spartanburg in South Carolina, handling sheet-metal loading on an active line and contributing to the production of more than 30,000 X3 vehicles. Its successor, Figure 03, has since been deployed at the same plant — reportedly around 40 units — with expansion planned across additional Spartanburg workstations through 2026–2027 and pilots slated for BMW's German plants.

Underpinning all of it is BotQ, Figure's manufacturing facility, where the company says it has reached a cadence of roughly one robot per hour. That number matters because in humanoid robotics, the binding constraint is increasingly manufacturing throughput, not algorithms. The company has even taken to noting that robots now outnumber humans at its own facility — a milestone meant as much for investors as for engineers.

The capital-intensive race

Figure does not have the field to itself, and the funding gap is closing. Tesla plans to put several thousand Optimus units on its own production lines by the end of 2026, with external sales targeted for 2027 — a vertically integrated rival with effectively unlimited balance sheet. Apptronik raised $520 million at a $5 billion valuation in a round backed by Google. 1X has begun delivering its NEO home robot to early adopters at roughly $20,000, or $499 a month. Agility Robotics' Digit continues warehouse trials with Amazon, and Physical Intelligence is pursuing a model-first approach aimed at being the "brain" for many robot bodies rather than building its own.

China complicates the math further: Unitree shipped more than 5,500 humanoids in 2025 — more than all U.S. competitors combined — and is targeting 20,000 units in 2026, with TrendForce forecasting global shipments could breach 50,000 units this year, a roughly 700% surge.

The throughline is that physical AI is staggeringly capital-intensive. Unlike software, every unit carries real bill-of-materials cost, every deployment demands months of on-site integration, and the data needed to make the robots generalize can only be collected by robots already in the field. That dynamic rewards whoever can fund the longest runway — which is precisely why Figure's $39 billion war chest, rather than any single demo, is its most important asset.

What to watch

The questions that will define Figure's 2026 are concrete. Does the BMW pilot convert into a paid, scaled rollout across plants, or stall as a high-profile trial? Can BotQ push past one-robot-per-hour toward the volumes that justify a $39 billion mark? Does Helix 02's full-body autonomy hold up outside curated demos? And does the long-rumored next mega-round — the one some have already labeled a $48 billion Series D — actually materialize, or does the broader humanoid market cool before the capital arrives? For now, Figure sits at the front of the pack on funding and ambition. The harder test, turning a record valuation into robots that earn their keep on real factory floors, is only beginning.