Latest / Minds, Bodies, and Terawatts / Figure AI Valued at $39B After Series C

Transcript

Figure AI Valued at $39B After Series C · Minds, Bodies, and Terawatts

• 0:00
  [Host] This is Minds, Bodies, and Terawatts, where your next coworker might come with a monthly subscription. Axios reported on April 10th that Figure AI just closed its Series C at a thirty-nine billion dollar valuation, a fifteen-x jump in eighteen months, and they're offering robots at roughly a thousand dollars a month.
• 0:16
  [Guest] A thousand bucks a month — that's less than the average temp worker costs per week. Figure is packaging the whole deal as Robot-as-a-Service: hardware, software updates, maintenance, all bundled into one subscription. You don't buy the robot. You lease the labor.
• 0:29
  [Host] So a warehouse manager doesn't need a capital expenditure budget or a robotics engineer on staff. They just sign a lease?
• 0:35
  [Guest] That's the play. It removes the two biggest adoption barriers in one stroke. No six-figure upfront cost, no in-house maintenance team. Figure handles the fleet. You handle the work orders.
• 0:44
  [Host] Who's funding this kind of valuation? Thirty-nine billion for a company founded in 2022 is aggressive.
• 0:49
  [Guest] NVIDIA, Intel Capital, Qualcomm Ventures, Salesforce. The investor list reads like a roster of companies betting that physical AI is the next platform shift. Figure ran an eleven-month pilot at BMW's Spartanburg plant building cars with Figure 02. And Figure 03 showed up at the White House in March alongside Melania Trump.
• 1:06
  [Host] A humanoid robot at the White House. How does Figure stack up against the competition on price?
• 1:11
  [Guest] Unitree has publicly priced its hardware starting around fifty-six hundred dollars. Tesla has said it's targeting twenty to thirty thousand per Optimus unit. Figure's thousand-dollar subscription sits in a sweet spot for commercial buyers who want enterprise-grade capability without ownership risk.
• 1:24
  [Host] And what makes Figure 03 different from earlier models? Because if you're asking a warehouse operator to swap out their temp workforce for a subscription robot, the technology needs to justify that leap.
• 1:34
  [Guest] Their proprietary AI, Helix, is a vision-language-action model. You don't program it. You talk to it.
• 1:39
  [Host] So the interface is natural language, not code?
• 1:41
  [Guest] Right. Double the frame rate of the previous generation, quarter the latency, palm cameras detecting forces as light as three grams. That's the weight of a paperclip. The barrier to deployment dropped from "hire a robotics engineer" to "hire anyone who speaks English."
• 1:55
  [Host] Okay, but a thousand a month per robot still adds up. Fifty robots is six hundred thousand a year.
• 2:01
  [Guest] Run the comparison. Fifty temp workers at fifteen dollars an hour across two shifts costs about three million a year, plus benefits, turnover, and training. Fifty robots at twelve thousand each annually is six hundred K, running around the clock with no overtime, no workers' comp claims. The math is stark.
• 2:16
  [Host] I want to push back on that, because the math assumes these robots actually perform at spec in messy real-world environments. BMW's Spartanburg is a controlled factory floor. What happens when you drop Figure 03 into a chaotic third-party logistics warehouse where nothing is standardized?
• 2:30
  [Guest] Fair challenge. We don't have independent data on uptime rates or mean-time-between-failures for Figure's fleet yet. The BMW pilot was curated.
• 2:37
  [Host] So you're extrapolating from a controlled factory floor to the real world.
• 2:41
  [Guest] Exactly. If these robots hit 70% uptime instead of 95%, your cost advantage evaporates and you still lost the experienced workers you let go. The subscription model actually makes that worse, because you're paying monthly whether the robot is productive or not.
• 2:54
  [Host] So the risk transfers to the customer on the performance side, even though Figure absorbs the maintenance cost.
• 2:59
  [Guest] Which brings up something most coverage is missing. A piece in Robotics and Automation News titled "The Rapidly Falling Cost of Humanoid Robots" tracked average humanoid hardware costs dropping from around eighty-five thousand dollars to roughly twenty-five thousand in two years. Their key finding was that price decline alone didn't predict adoption rates. What predicted adoption was who bore the liability when something went wrong.
• 3:19
  [Host] So the bottleneck isn't price. It's risk allocation.
• 3:22
  [Guest] With a subscription, Figure keeps hardware risk but the customer absorbs productivity risk, workflow disruption risk, and the cost of restructuring their labor force around a technology with no long-term track record.
• 3:32
  [Host] So the real question isn't whether the robots work. It's who pays when they don't.
• 3:36
  [Guest] Exactly. And that's where the staffing industry gets hit hardest. According to the American Staffing Association, temp labor is a hundred-and-sixty-billion-dollar industry in the US. Staffing agencies have always absorbed that liability gap — they guaranteed a warm body would show up trained and insured.
• 3:50
  [Host] And their value proposition just became Figure's value proposition.
• 3:53
  [Guest] If Robot-as-a-Service scales, those agencies lose their reason to exist.
• 3:56
  [Host] And what about the workers those agencies place? We're talking about people who depend on temp work as a primary income, often without a four-year degree. Where do they go?
• 4:06
  [Guest] Warehouse pickers, packaging line workers, basic assembly roles. These jobs have traditionally absorbed people displaced from other sectors. If robots fill those positions, the landing pad disappears.
• 4:14
  [Host] What strikes me is that this is a business model breakthrough more than a technology breakthrough.
• 4:19
  [Guest] I'd go further. The real breakthrough is liability architecture. Subscriptions convert a scary capital bet into a predictable operating expense. That's what made cloud computing take off.
• 4:27
  [Host] Right, AWS didn't sell servers. They sold the absence of server risk.
• 4:31
  [Guest] But subscriptions also split risk in ways that aren't obvious until something breaks. Figure keeps the hardware risk. The customer keeps the operational risk.
• 4:38
  [Host] And the displaced workers?
• 4:40
  [Guest] They keep all the economic risk with zero upside. That three-way split is the actual innovation, and nobody in the deal is representing the third party.
• 4:48
  [Host] Once you've restructured operations around robot labor, unwinding it means rehiring, retraining, rebuilding institutional knowledge that walked out the door. The lock-in is real.
• 4:55
  [Guest] And Figure pushes software updates to the fleet continuously. Your robots improve every month without you doing anything. The hardware is a delivery mechanism for continuously improving AI. That's the competitive moat.
• 5:05
  [Host] This tracks with what Patrick Day-glon describes in Unscarcity as physical labor becoming a subscription line item instead of a human commitment, the automation of bodies catching up to the automation of minds. The unscarcity dot ai article "What Happens to Your Twenty-K Robot When Robots Cost Two Hundred Dollars" digs deeper into how subscription and lease models make robot labor accessible before purchase prices collapse. The book's on Amazon, audiobook's on Spotify. Head to unscarcity dot ai for articles and the forum. This has been Minds, Bodies, and Terawatts.