Figure raises more than $1 billion in Parkway-led Series C at a $39 billion valuation, 15 times its 2024 price

The round resets the price of a humanoid developer. At $39 billion post-money, Figure is valued at 26 times Skild AI's July 2024 post-money and 15 times its own February 2024 figure, and every later humanoid round will be read against it. Investors in rivals now have a public marker for what a vertically integrated humanoid company with its own model and factory can command before it reports revenue.

It also shows where the money in physical AI is flowing. Figure's three stated uses are manufacturing, compute and data, and they depend on one another. Spending on GPU infrastructure and human video data treats the robot as a vehicle for a learning system, while BotQ is the route to cheaper hardware. Rivals that have only one of those three will find it harder to raise at similar terms.

Apptronik is the closest US comparison on model and capital: a general-purpose humanoid aimed at logistics and manufacturing, financed with a $350 million Series A in February 2025 and a $53 million addition in March, without a disclosed valuation. Agility Robotics has the longest commercial record with its Digit robot in warehouses but last disclosed new equity in 2022. 1X Technologies, backed by the OpenAI Startup Fund in 2023, is aiming at homes, the same segment Figure now names first in its use of funds.

Tesla is the rival that cannot be priced. Optimus is funded from a carmaker's balance sheet, has access to Tesla's own AI training clusters and can be deployed in Tesla plants. In China, Unitree, UBTech and AgiBot compete on cost and volume with deep domestic component supply chains. Figure's response is to own the model and the factory, a strategy that requires capital on the scale of this round.

Figure's disclosed rounds rose from $70 million in 2023 to $675 million at a $2.6 billion valuation in February 2024 and to more than $1 billion at $39 billion now. Taking the $1 billion floor, new investors bought roughly 2.6% of the company ($1bn / $39bn), a small dilution for the amount raised. That ratio points to a market where a few humanoid leaders can raise large sums without giving up much ownership.

The comparison set is thin. Skild AI's $1.5 billion post-money from July 2024 is the only peer post-money in our verified data, and Apptronik, Agility and 1X have not disclosed valuations for their latest rounds. The 15-fold step-up should also be read with care, because Figure's 2024 release did not say whether $2.6 billion was pre-money or post-money, and committed capital may not be fully paid in.

BotQ concentrates Figure's supply chain around parts it builds itself: actuators, hands, batteries and final assembly. Figure said in March 2025 that its supply chain could support 100,000 robots or 3 million actuators within four years, which implies roughly 30 actuators per robot by its own arithmetic. Moving from CNC machining to injection moulding, die casting, metal injection moulding and stamping shifts the bottleneck to tooling and to suppliers of gears, bearings, magnets and motor components.

The chip side of the cap table matters as much. Helix runs on dual embedded GPUs onboard, and training depends on large GPU clusters, so NVIDIA's presence is strategic. Intel Capital and Qualcomm Ventures give Figure relationships with alternative chip suppliers for onboard compute. LG Technology Ventures connects Figure to a Korean group with batteries, displays and motor know-how.

The first signal is the paid-in amount. Figure described more than $1 billion as committed capital, so any later statement of a final close, or the timing of tranches, will show how firm the round is. The second is BotQ output; the company gave capacity, not production, and the first reported delivery numbers will test the factory claims.

The third is the data program. If Figure publishes results showing Helix policies trained mainly on human video, that would support the thesis that data, not teleoperators, limits progress. The fourth is customers: named buyers beyond BMW, and any home pilot with pricing, would give the first demand-side evidence for the valuation.

Thesis: Figure has bought time and compute to prove that a single learned model on in-house hardware can generalise across homes and workplaces, and the $39 billion price assumes it succeeds. Confidence: medium. The confidence is not high because Figure disclosed no revenue, fleet size or customer count, the round size is a floor of committed capital, and the earlier valuation basis is unspecified.

The confidence is not low because the facts that were disclosed are specific and consistent. Helix has a documented architecture with stated rates and parameter counts, BotQ has a stated capacity and in-house component list, and the syndicate includes the chip supplier the company depends on. Those details are what investors are paying for, and they can be checked against output over the next 12 months.

How much of the more than $1 billion in committed capital has been paid in, and are there milestones attached to later tranches? What share of the round came from Parkway and from new investors such as Brookfield, Salesforce and T-Mobile Ventures?

How many robots has BotQ built since March 2025, and when will Figure report deliveries? Will the home segment, listed first in Figure's use of funds, be sold, leased or offered as a service, and under which safety standards? How many hours of human video does Figure plan to collect, and how will it measure transfer from video to robot behaviour?