B rett Adcock has spent the past decade turning tomorrow into money.

He made his first $30 million the old-fashioned way: building and then selling job website Vettery for $110 million in 2018. Everything since has come from putting a present-day price on a future-day product. In 2021, he took the air taxi outfit Archer Aviation public via SPAC, raising $858 million at a $1.7 billion valuation before its aircraft had flown a single commercial flight. He persuaded investors to value Figure at $39 billion while its humanoid robots were still working small-scale factory pilot projects and tightly controlled demonstrations. This May, he raised $700 million at a $6 billion valuation for Hark, an AI assistant company with nothing on the market at the time.

The work is real. Vettery was a legitimate win. Archer built a full-scale electric vertical takeoff-and-landing aircraft. Figure’s humanoids have worked on a BMW production line and have helped build 30,000 X3 SUVs, by the automaker’s count. Adcock, barely 40, has a reputation for raising enormous sums, recruiting talented engineers and getting difficult machines to work. He’s currently CEO of three companies. And when it comes to humanoids, he is a flywheel of enthusiasm.

“Over a long period of time we’ll have maybe ten or 20 billion humanoids on the planet,” he says. “They’ll run the whole supply chain. They’ll build homes for you.”

Adcock also has a reputation for exaggeration. A pilot project becomes a fleet of robots. A falling-out with a marquee partner becomes a technical rout. It hasn’t slowed his rise: Early progress becomes a $39 billion valuation. His work, and his public account of it, have made him worth more than $20 billion and left BMW, OpenAI and Archer correcting the record in public.

Adcock’s claim that while at Archer he “architected, engineered and flight-tested five generations of aircraft”? An Archer spokesperson calls that “wildly inaccurate,” adding that “since his separation in 2022, he has consistently overstated his contributions.” (An Adcock spokesperson dispu­tes this, saying Archer was “attempting to take credit for and diminish his accomplishments.”)

That 2025 BMW pilot? In Adcock’s telling it involved a “fleet of robots performing end-to-end operations.” BMW tells Forbes that just one Figure robot worked at a time, with a second onsite to rotate in while the other one recharged.

Then there’s the OpenAI deal. The AI lab invested in Figure’s Series B in 2024, and the two companies set out to build the models that would run Figure’s robots. It all fell apart in less than a year.

OpenAI attributes the breakup to Figure’s “consistent failure to meet the deliverables required, gaps in software and hardware support” and unspecified “challenges in the working relationship.” Adcock says he fired OpenAI because his team “ran circles” around it.

None of this has tripped him up. Forbes now estimates his fortune at $23.4 billion, mostly thanks to the spike in Figure’s valuation. Because if a humanoid can theoretically do anything a human can, what price can you put on it?

Right now, the future of the humanoid market is essentially illegible. Morgan Stanley says it will be worth $5 trillion by 2050. Citi says $7 trillion. UBS says up to $1.7 trillion. RBC says $9 trillion to $12 trillion. Within this heady fog, Figure’s valuation went from $2.6 billion to $39 billion in 19 months.

That ambiguity is the zone in which Adcock thrives. It’s why he can confidently talk about humanoids taking over a big chunk of the $42 trillion annual global human labor market despite daunting constraints like a paucity of training data or labor unions, or the fact that teaching a humanoid to use scissors remains impossible. Remember, there is no easily scrapable internet of human behavior on which to train humanoids.

Friends such as Marc Benioff are impressed with the maximalist pitch. “He is extremely positive, extremely energetic, and he has a big vision. . . . [He makes] very aggressive statements about the future that are really things that we’ve heard about only in science fiction,” says the cofoun­der of Salesforce, a Figure investor.

“I think every human will have one, like a phone or car."

Adcock also has the well-polished endorsement of AI infra­structure powerhouse Jensen Huang. The Nvidia CEO says Adcock has “an extraordinary ability to imagine the future” and wrangle “the hard work of building it.” Nvidia, too, is a Figure investor.

Humanoids are our oldest fantasy. The Greeks imagined Hephaestus with golden handmaidens who could think and speak. Leonardo sketched a mechanical knight that could sit up and wave its arms. Westinghouse sent Elektro shambling onto the floor of the 1939 World’s Fair: seven feet of aluminum and a 700-word vocabulary that came off a record player. Cable-and-pulley, relays, now neural networks. Every age builds it out of what it has and oversells what it built.

For Adcock, the future is clear. A humanoid in every home, leased for perhaps $500 or $600 a month, doing laundry, unloading dishwashers, cleaning kitchens and picking up kids’ toys. A world relieved of drudgery. “I think every human will have one, like a phone or car,” he says.

F igure’s cavernous 300,000-square-foot headquarters in San Jose, California, is designed to impress, like something out of a James Cameron movie. Robots walk the corridors like rank-and-file employees. Some practice flipping packages in a mock warehouse. Guests are given stickers to cover their phone cameras; staffers double-check to make sure they are properly applied. Adcock grins like a kid showing off an epic papier-mâché volcano. “Pretty sci-fi, right?”

It is. And Adcock prefers to be in the middle of it all. He likes to work in the weeds. His desk is in the center of the office near a mini Figure museum. He works relentlessly. He says he attends a half-dozen product and engineering meetings a day, debating the minutiae of a robot’s foot, prioritizing software bug fixes and wooing new talent.

During a design standup that afternoon, Adcock delights in a rendering of Figure’s future charging station: a circular glass sarcophagus that a humanoid steps backward into, like some glowing sci-fi portal to the future.

“That charger is dope,” he says. Placed in a home, he enthuses, “it’s art.”

In Silicon Valley, Adcock’s companies sit strangely outside the industry’s conventional venture hierarchy. Forbes contacted a dozen top investors from firms on our Midas List who have backed robotics companies; all of them questioned Figure’s valuation, and some of them voiced reservations about Adcock’s claims.

He doesn’t care. He raises anyway, through a separate network of investors who syndicate access to his companies and strategic investors like Salesforce and Nvidia.

“To the extent we need to raise, we have some really big firepower and a lot of interest and demand,” Adcock says. “I think we’ll be fine.”

Last year, when Figure sought to raise $1.5 billion at a $38 billion pre-money valuation, the term sheet reserved as much as $1.2 billion for two relatively unknown VC firms, Parkway Venture Capital and Align Ventures. Neither had the assets for such a substantial amount. Align recently announced a $125 million fund to write small checks for consumer goods companies (it has backed sparkling water brand Olipop and pimple patch brand Starface). The firms filled out their allocations with special-purpose vehicles (SPVs), single-company pools that let investors fill out a round. Align took checks as small as $100,000. When the round closed, it put Figure’s valuation at $39 billion, up from $2.6 billion the year before, vaulting it into the top 20 most valuable private companies in the world, nearly on par with that of its marquee partner, BMW. Adcock argues investors got a deal.

“If we’re able to be the first to solve true intelligence for a humanoid robot in the biggest total addressable market in the world,” he says, “I think $40 billion is probably really cheap.”

Central to that incredible valuation step-up were yet-to-be-deployed fleets of robots at two big customers: BMW and an undisclosed company—which people familiar with the matter identify as UPS.

Adcock publicly said he saw “a path to 100,000 robots over the next four years” between the two companies. In a memo to prospective SPV investors, Align said 100,000 robots over four years would represent “$9b in ARR [annual recurring revenue] under current pricing assumptions.”

“I think every human will have one, like a phone or car."

The BMW pilot, which ended late last year, was a meaningful accomplishment. The automaker confirms it assis­ted in the production of tens of thousands of vehicles, moving some 90,000 parts. But the work was far narrower than Adcock’s public account of it. On social media, he claimed Figure had a “fleet of robots performing end-to-end operations” for the carmaker. BMW tells Forbes that only one Figure robot worked at any given time. When Fortune earlier reported the same thing, Adcock accused the magazine of “mischaracterizations and downright lies” and threatened legal action. He never publicly said what he believed to be false, but a Figure spokesperson tells Forbes that Fortune “mischaracterized the scope of our early collaboration with BMW.”

In June, Figure returned to the automaker’s Spartanburg, South Carolina, plant, and Adcock’s social media lit up with another burst of publicity: videos set to thumping electronic music showing updated Figure robots tugging heavy carts around a BMW facility.

Describing the project, Adcock tells Forbes he thinks “six or eight” Figure robots are onsite and says he’s “pretty sure” BMW is paying for them, though he wants to check both points with his communications team.

“I think we’re getting paid for all the robots now,” he says. “All the customers we have have a certain amount of robot allocation and contracts, and we’re trying to ship a lot more into all of them right now.”

That’s an oddly muddy assessment of a big, important partnership. BMW’s account is clearer. The carmaker tells Forbes that the project was a non-production test involving two or three robots. Now that it has ended, no Figure robots currently operate at the plant.

Of the four robots in the promotional video, BMW says one did the job; the other three did not participate. It also says Figure sent five to ten people to ensure the video shoot went smoothly. Regarding the companies’ relationship, BMW says only that they “continue to collaborate in an ongoing evaluation and exchange of learnings related to humanoid robotics.” It declines to say whether any money changed hands.

Asked about the disparity later, Adcock hedges. “I can’t give an exact number because as we explore new use cases together, we move robots back and forth between our facili­ties for ongoing development work.”

It doesn’t seem like there are any Figure humanoids in active use at UPS, either. A spokesperson for the International Brotherhood of Teamsters, the union that represents UPS workers, says the shipping company has not notified it of any Figure robot testing, which the union says it is required to do for any tests affecting members. UPS declined to comment on any relationship with Figure, as did Adcock.

A third customer, JCPenney parent company Catalyst Brands, is operational, but Figure would not say how many humanoids are working. The two companies share an investor, Brookfield.

Figure is a real company wearing a much larger company’s numbers. Of the 1,000 or so humanoids it has built, it’s not clear how many have been deployed, or where. Who is paying remains an open question.

Figure’s former relationship with AI juggernaut OpenAI follows a similar pattern of conflicting accounts. The frontier AI lab tells Forbes Figure repeatedly missed agreed deliverables and could not supply the hardware and software support its researchers needed. By the end, OpenAI says, it had concluded that Figure was not a viable robotics partner and informed Adcock that it planned to invest more heavily in its own robotics work. Figure rejects that account. Its team moved faster than OpenAI when it came to training models on robots, and they didn’t want to outsource that, so they terminated the relationship, a Figure spokesperson says.

These days Adcock is focusing Figure’s story on the home. Showing off videos of robots training inside Airbnbs, he says that by next year, Figure’s robots could be visiting our houses to help with chores.

But humanoids are nowhere near safe enough for that, says Rodney Brooks, co-creator of the Roomba vacuum, the first home robot anyone actually bought. “I certainly wouldn’t have [a humanoid] around my grandkids,” he says.

A November 2025 wrongful dismissal lawsuit filed against Figure by its former head of product safety, Rob Gruendel, describes an internal fight over exactly those risks. Early last year, the lawsuit alleges, Adcock asked Gruen­del what it would take to put robots in homes. Gruendel developed a road map centered on detecting nearby objects and not running into living things. He alleges that Adcock declined to attend a meeting or respond to messa­ges about this road map. When he pressed for risk reduction measures, the suit alleges that another Figure executive told him that “[Adcock] would shoot us if we did it.”

Gruendel claims that a Figure 02 humanoid, programmed during a July 2025 test to strike at maximum force, generated impacts more than twice the force needed to fracture an adult skull. He also described a malfunction in which a robot punched a quarter-inch gash into a steel refrigerator door, narrowly missing an employee. He says he escalated his warnings, objected when work to certify an emergency stop was canceled and was fired days after his most detailed written complaints.

Figure disputes portions of Gruendel’s account and says it is committed to the “highest safety standards.” It says he was fired for poor performance and claims many of his concerns were trivial or illegitimate. The suit is ongoing.

B rett Adcock grew up on 2,000 acres of corn and soybean fields in Moweaqua, Illinois, around machinery and entrepreneurial relatives who had built companies—a tractor dealership and grain-bin manufacturer—off the farm economy. From the University of Florida he cold-called his way into a boutique investment banking internship. Then came Vettery. Then Archer. Then Figure. Then Cover (which he says hasn’t raised outside capital or hired full-time employees yet), building weapons-detection tech to prevent school shootings. Then Hark.

Adcock builds companies that cut against the current Sili­con Valley ethos. In an era when a single person using Anthropic’s Claude can code, ship and sell a product in days, Adcock raises vast sums, assembles large teams and markets a grand vision before anyone can buy anything.

His latest venture, Hark, is perhaps the purest example. This spring it raised nearly three-quarters of a billion in a round led again by Parkway Venture Capital. It already employs roughly 100 people. The product is a wearable AI assistant that can remember a user’s life and take actions on their behalf. In an August marketing video that opens with a stylist prepping Adcock for a photo shoot, he declares, “At Hark, we’re creating the world’s most capable personal intelligence.” Then the software goes online to order flowers.

Dana Berlin, a Figure and Hark executive, describes today’s AI products as “like nothing, basically,” next to Adcock’s vision. “What I’m telling you is we’re in the process of the AI revolution and there’s this one guy in Silicon Valley, as far as I know, who’s figured out the recipe to create totally different artificial intelligence products.” Hark has no precedent, she says, because “there’s no model of how to use your computer. It doesn’t exist.”

But AI assistants that take action have existed for some time. Google famously showcased its AI assistant making phone calls for restaurant reservations and haircut appointments at its 2018 I/O developer conference. In August, Instinct, a startup of about ten people, released an AI assistant that is already doing things for customers like searching for in-network podiatrists and negotiating with wedding vendors over WhatsApp. A couple weeks later, Meta released Muse, a personal AI agent capable of tasks far more complex than ordering flowers.

That’s what Hark is up against, and it hasn’t brought its wearable to market yet. “I think Figure and Hark are going to explode this year and next,” Adcock says.

A dcock has proven that Figure can build humanoid robots. He has yet to prove that it can build a humanoid-robot business at scale. The company has made genuine progress; BMW says so itself. It’s just that the valu­ation, scale and market penetration all describe a bigger business that doesn’t exist yet. Figure is a real company wearing a much larger company’s numbers.

Adcock is very much a creature of this moment. The mainstreaming of generative AI has made raw vision fundable again. Elon Musk has used social media to sell sci-fi pipe dreams directly to his followers, while OpenAI’s Sam Altman and Anthropic’s Amodei siblings plan to IPO their companies based on the promise that machines are almost smarter than humans. But their businesses also have massive product wins. You can use ChatGPT and Claude today; at least a billion people do. Figure refuses to disclose revenue, and it’s impossible to ballpark revenue accurately from available information. Of the 1,000 or so humanoids it’s built, it’s not clear how many have been deployed or where.

Here’s the thing: The humanoid form is an astonishingly effective piece of marketing. And at this moment in time, it wildly oversells the machine inside it. A human-shaped robot arrives fully loaded with assumptions and expectations of endless functionality. If it has hands, surely it must be able to fold laundry or fry an egg. If it has legs, surely it will soon be pushing a lawnmower or carrying a baby.

It’s a lot easier to sell a vision of a future with a robot in every home when its design immediately registers in our minds as human. It makes it easy to forget that there are 4 million mechanoreceptors in the human body, 17,000 of them in each of our hands. Most full-body compound movements require the coordination of dozens of major muscles, millions of individual muscle fibers and a cascade of billions, if not trillions, of synapses firing in perfect harmony. The training data to re-create even a fraction of that need to be captured, and the machines need to be good enough to use it.

Despite this, Figure is, unsurprisingly, raising again, says Gregg Hill, a founder at Parkway Venture Capital who sits on the company’s board. He said it is in the process of closing “billions” as an extension to its Series C round (the company declined to confirm an active raise), and says its Series D valuation will be considerably higher.

So does Adcock. As ever, the plans are scaling alongside the capital.

“We have some crazy stuff happening here at the com­pany,” he says. “I made a prediction early this year: My stretch goal would be to take a robot to a home it’s never seen before and do long horizon work—laundry end-to-end or maybe make scrambled eggs. I feel very confident.”