AI Entrepreneurs Q&A: How Every Industry Is About to Be Transformed by Humanoids w/ Vinod Khosla & Brett Adcock | EP #160
Vinod KhoslaBrett AdcockPeter Diamandis
- Figure’s bet is that humanoid economics improve when one general-purpose hardware platform serves most human work. Adcock rejects customer-specific hardware because high NRE, maintenance burdens, and changed observation spaces impede transfer learning; standardization is the path to “twenty thousand dollar levels,” while Figure’s near-term scale target is 100,000 units.
- Deployment should arrive unevenly: indoor workforce work first, then homes, outdoor jobs, and space, though not strictly sequentially. Adcock places this broad rollout within the next 10 years but distinguishes demonstrations from scaled integration; Khosla expects farm capability within five years and considers 1% penetration “the job done,” with broad adoption taking longer.
- Khosla sees fusion’s bottleneck less in physics than in siting, transmission, and politics. He is “a little less bullish” on fission because permitting may outlast developing fusion from scratch, yet sees an “easy path” to 5,000 fusion plants by 2050; 50 MW Realta units could occupy substations, while Commonwealth is building a 500 MW reactor.
- AI and robots can deliver abundance while widening income disparity enough to make unmodified capitalism unworkable. Khosla forecasts job displacement, productivity, and GDP growth together; rather than prescribing one UBI model, he says countries must decide how to share abundance, pairing techno-optimism with “care and caring.”
- The opportunity can be enormous even if most funded companies die. Khosla estimates perhaps 80% of AI investments will lose money, but exponential winners leave aggregate gains positive, with three to five winners per field and perhaps half a dozen robot companies; Adcock calls humanoids “winner take most” and says 95% of such deep-tech efforts go bankrupt.
- Figure treats security and civilian positioning as constraints on the product, not public-relations add-ons. Its 2022 mandate rules out military work because the “Terminator vibe” could damage the larger commercial market; Helix kitchen demonstrations ran offline with weights on an onboard GPU, while firmware enforces actions the robot “ultimately can never do.”
- Healthcare exposes the gap between model capability and regulated physical autonomy, even as AI could multiply scientific capacity 10–100x within five years. Khosla expects healthcare expertise to become free “well before 2030,” but unsupervised robot surgery to remain more than 10 years away—possibly 15 or 20 depending on the FDA—while AI scientists make research 10x cheaper and 100x more abundant.
1. One chassis is the route to $20,000 humanoids
Adcock says Figure—under three years old, with a few hundred engineers—chose the minimum number of companies to ship robots to. Its “magic number” is 100,000 robots, which it has across its first two groups; the focus reflects limited bandwidth even as Figure works in healthcare, construction, other industries, and homes.
Customization creates “high NRE scenarios,” reliability and maintenance burdens, divergent product paths, and weaker transfer learning when embodiment or observation space changes. Adcock instead wants one platform that increasingly moves, touches, and feels like a person while performing most human tasks end to end.
The economic objective is “twenty thousand dollar levels,” with foreseeable development coming through product revisions rather than adjustments to the main chassis. Khosla invokes Henry Ford—“any color as long as it’s black”—while allowing that specialized robots may emerge for superhuman jobs such as lifting a car without a jack.
2. Capability arrives before society absorbs it
Adcock’s rollout is “not quite sequential”: workforce robots appear in vast numbers first, followed by homes, outdoor work, and space on uneven, overlapping timelines. Figure’s immediate task is shipping 100,000 indoor units, lowering costs, building the data pipeline, and achieving full autonomy with no human intervention; homes and outdoors are part of the next 10 years.
Agriculture illustrates the timing gap. Adcock expects indoor environments first because weather makes outdoor operation harder; Khosla expects farm capability within five years but distinguishes that milestone from adoption: at 1% penetration, he considers the technical job done, while societal diffusion may take much longer.
Adcock’s dream is a humanoid flying an Archer aircraft amid a pilot shortage, and he believes robots will eventually do “everything a human can. Physically.” But demonstrating a robot driving or farming and integrating it into society at scale belong on different timelines.
Space appeals because it is “such a pernicious place for humans”: Adcock imagines mechanical embodied agents and autonomous specialists on Mars and the Moon, though Earth comes first. Khosla argues exploration—not resource extraction—will motivate expansion, and recalls telling “Sam” that a self-replicating AI probe would leave Earth in 15 years, only to hear: “You’re way too conservative.”
3. Fusion can localize generation, but permitting remains the risk
The named fusion efforts include Commonwealth, Helion, and Taifun; Khosla first refers to another effort as “Rialta” and later calls the 50 MW builder Realta Fusion, amid what he calls emerging “sovereign fusion technology.” He stresses that major technological transitions remain exposed to regulatory forces and political pressure, citing the Screen Actors Guild’s AI agreement as an example of rules that could drive customers out of business if followed.
Despite investing in TerraPower with Bill Gates, Khosla has become “a little less bullish” on fission because community objections, environmentalist objections, and NIMBYism may make siting slower than developing fusion from scratch. By contrast, he sees an “easy path” to 5,000 fusion plants by 2050, “maybe more.”
Reactor size changes grid economics: Commonwealth is building at 500 MW, while Realta targets 50 MW—small enough, Khosla argues, to place where a substation sits today and eliminate some transmission needs. More lines will still be needed, supplemented by superhot geothermal and, if cheap energy permits, transmission tunnels underground.
4. Abundance widens inequality and concentrates venture returns
Challenging Kathy Woods’s historical claim that technology creates more jobs than it destroys, Khosla says displacement will happen. His earlier forecast remains abundance, productivity, and GDP growth alongside increasing income disparity—the outcomes economists love coupled with a distributional problem that policy must address.
Khosla does not offer UBI as a universal prescription. He says policy must determine “how do we plan to share the abundance more broadly,” with different countries choosing different paths because “the current capitalist system unmodified” will not work; dystopic elements, he says, will also be societal choices. His formulation is techno-optimism with safety-minded “care” and “caring for those left behind.”
His venture lesson is staged fundraising: pressure teams to test assumptions incrementally, while his own analysis found that companies raising less money were more likely to produce more absolute dollars. Diamandis offered a different timing view—raising a lot early and scaling later once there is a path—while Khosla agreed that capital is needed at the scaling stage. Khosla’s conclusion is that complacency is innovation’s enemy.
On AI commoditization, Khosla expects perhaps 80% of investments to lose money, yet more money to be made than lost because of exponential winners. He foresees three to five winners per area and perhaps half a dozen humanoid companies—probably fewer than today’s car companies in 20 years; Adcock’s harsher version is “winner take most,” with billions required for manufacturing and data and 95% of such deep-tech efforts failing. Khosla therefore “only care[s] about 2030,” not today’s headlines.
5. Regulation slows physical autonomy as AI accelerates science
The robot-care question was framed by critical healthcare-worker shortages and boomers entering the healthcare system; Diamandis recast it as whether robot surgeons would become the baseline. Khosla expects healthcare expertise to become free within a couple of years, “well before 2030,” but interventional medicine such as catheter procedures will move more slowly. Unsupervised robot surgery is more than 10 years away; whether it takes 15 or 20 depends chiefly on the FDA.
AI safety remains “a real problem to worry about,” not something Khosla dismisses despite his optimism and respect for warnings from Yoshua Bengio and Geoffrey Hinton. He expects major explainability advances within three to five years—and within two to three years, claims that models are merely black boxes or parrots may begin to fade.
Figure’s 2022 mandate excludes military applications because humanoids already carry a “Terminator vibe,” and Adcock sees the civilian market as much larger. Cybersecurity is architectural: Helix kitchen robots ran offline with neural-network weights on onboard GPUs, while Figure works to deny kernel root access and encode actions the robot can never perform in nonvolatile firmware.
Khosla’s most aggressive forecast is the number of scientists multiplying 10–100x within five years through AI PhD chemists, biologists, and physicists. Agents would generate and test hypotheses without human intervention, making research 10x cheaper and 100x more abundant. Diamandis imagines AI models designing experiments and robots running them lights-out; Adcock agrees more generally that agents on both the physical and digital sides will increasingly do what humans do.
Full transcript
You’re going to see robots first in the workforce in vast numbers. You’re going to see them in the home, and you’re going to start seeing them in space. This will all happen.
Brett Adcock.
Space would be fantastic for humanoids. It’s just such a pernicious place for humans.
And Vinod Khosla.
The future of work, surgery, and space travel.
Do you see robot surgeons in the future being the baseline?
I think the beauty in a humanoid is that we really want to be general-purpose.
Customization generally means more cost.
What is the role of universal basic income in the economy of the future?
I do think we will have to deal with that issue through policy. I don’t think the current capitalist system, unmodified, will work.
Now, that’s a moonshot, ladies and gentlemen.
This is your time. We’ll take questions at the mic, and we’ll take questions on Slido. Let’s begin right here.
I’m Mike Wandler. I’ve got a company called Evercore Energy that I just started to innovate energy, and I was really focused on microreactors for heavy industrial and mining. The fusion really changes that. I hadn’t heard that it was going to be available in the next 5 years. What companies should I watch, and should I pivot my business model?
Look, there are a lot of fusion companies, but there are a good half-dozen efforts that are highly credible.
Commonwealth, Helion.
There are others. There’s Taifun, there’s Rialta, and, in fact, every country has its own effort. Just like there’s sovereign AI, there’s sovereign fusion technology coming out.
My only comment would be that, in all these areas of massive change, you have to look at regulatory forces and political pressures.
Right.
The Screen Actors Guild fought the use of AI, and their customers will almost certainly go out of business if they comply with that agreement. That’s what will happen.
Regulation matters. In fission specifically, no matter how safe you are—and we are investors in TerraPower with Bill Gates—what will happen is that it will take longer to site a power plant because of community objections, environmentalist objections, NIMBYism, and all that than it takes to develop a fusion technology from scratch.
That’s the situation. That’s why I’ve become a little less bullish on fission reactors. I would love to see it, but I don’t see it happening rapidly, while I see an easy path to building 5,000 fusion power plants by 2050. Maybe more.
Amazing. Let’s go with John.
Hi. My name’s John. This is for Brett. How soon do you think there will be an opportunity to have robots in the field for the trades? I’m in an auto repair business, and I want to send a technician out.
Secondly, will there be an opportunity to beta-test and partner with a company like yours using the expertise we have in that area?
Yeah. We made the decision pretty early on to pick the minimal number of companies to ship a certain number of robots to. For us, the magic number is 100,000 robots. We have that now with our first 2 groups.
We’re relatively young—we’re still under 3 years old—and our team is a few hundred engineers. The truth is, we just don’t have the bandwidth a lot of times to do a lot of extra work outside what we have.
We’re doing a lot of work right now in other areas, like health care, construction, other industries, and obviously the home. I think there’s a mechanical need for a mechanical human to be doing those tasks, and we want to understand and look at it more. Hopefully, it’s an area we can spend time in over the next few years.
I hope it is, too. I would love it. Thank you.
Yeah.
Amazing. All right, Steve.
Hi. It’s Steve Strickland. First of all, to both of you, thank you for the wealth you’ve created for many in this room and certainly around the planet.
For Brett, when you start reaching smaller businesses and placing humanoids, will you allow customization of the chassis to meet the needs of the particular user?
I think the beauty in a humanoid is that we really want to be general-purpose. When we start customizing and making things really specialized, that’s when you get caught in the problem of high NRE scenarios. It becomes really hard to keep things reliable, maintain them, and design them. Then we have multiple different product paths.
The AI system also becomes really challenging if it’s not doing well with transfer learning across those different tasks, because the embodiment is different or the observation space of the robot is different.
I think our vision—my vision—for this is that we have a single hardware platform that can do everything a human can. We’re moving our roadmap to be more like a human. We move like a human, and we can touch and feel things more like a human over time.
Our roadmap is drawing us down that curve. We won’t do every single thing in the world, but we want to be able to do a majority of what humans can do with 1 hardware system. That’s how we’re going to get the cost down to $20,000 levels, and that’s how we’re going to get the intelligence up to allow it to do that end to end.
In the foreseeable future, we do not plan to make any hardware adjustments to the main chassis, but we do plan to make product revisions so it can operate more and more like a human over time.
Okay, thanks.
Yep.
Vinod, on Slido here: If electrical generation is taken care of, will there still be a need for additional investments in transmission lines, distribution, and transformers? I’m thinking about the need for reasonably close-to-room-temperature superconductors and such.
We will need transmission lines, but let me suggest the following: There will be different-size fusion reactors. We have Commonwealth Fusion, which is building a 500-megawatt reactor. We have Realta Fusion, which is building 50-megawatt reactors.
Now, you can put 50-megawatt reactors in what is today a substation. What does that do? It eliminates the need for transmission.
We will need more transmission—I’m not saying we won’t—but my bet is that siting more reactors in more locations will be an alternative we have. I’m pretty optimistic that will happen.
Yep.
I want to add a comment on Brett’s answer to the last question.
Yes.
Customization generally means more cost.
Yep.
For small businesses, Henry Ford said, “You can have any color as long as it’s black.” The point was to reduce the number of SKUs, reduce the cost, increase the volume, and that’s the way to make them really affordable.
Now, there will be specialized robots that will do other things and have more capability, especially when you go beyond human capability. When you get to superhuman capability—being able to lift a car without a jack—you may need more specialized operations.
I think it’ll look very much like the auto industry. It’ll diversify some, but the highest-volume things will be the cheapest and most applicable. The world is designed around human form factors, so they’ll fit in many more places without needing to change the environment. People will change some environments.
Let’s go to mic 5. Mark?
Everybody, I hope you’re enjoying this episode. Did you know that we’re likely to see as many as ten billion humanoid robots by twenty forty, and that Brett Adcock, the CEO of Figure, anticipates they’ll have robots in our home in the next two to three years? How about Max Hodak’s new form of BCI called biohybrid interfaces that could offer millions of connections between your neocortex and the cloud? Then there’s Michael Andregg, whose efforts at Eon is focusing on uploading the human connectome to the cloud by twenty thirty. These aren’t science fiction scenarios. They’re serious efforts underway today. I’ve distilled the most powerful insights and roadmaps from this year’s Abundance twenty twenty five Summit into a comprehensive report that will transform how you see the future. Get your free copy of the Abundance twenty twenty five Summit summary at diamandis.com/breakthroughs. That’s diamandis.com/breakthroughs.
Hey, Peter. My name’s Mark Donovan, founder of the Denver Basic Income Project, and my moonshot is to create affordable living as a service—housing, food, energy, data, and transportation—for $250 a month by the end of the decade.
Vinod, you started by asking, “How will we pay people in the future?” I haven’t heard anyone answer that yet. Yesterday, Kathy Woods said that technology will create more jobs than it displaces, which has been true historically, but I’ve been assuming this will not be true with AI and humanoid robots.
What do you think about that, and what is the role of universal basic income in the economy of the future?
For a longer dissertation on that, I wrote a 25-page piece on AI: Will It Lead to Dystopia or Utopia? It’s on our website. You can Google it.
Which one did you choose?
Clearly, utopia.
Okay. Just checking.
I do think there are dystopic elements—
Yeah.
—but those will be societal choices. We can go down certain paths and not go down others, but there’s a 25-page explanation of that. I’m sorry, I forgot your original question.
Wait, wait. UBI—is it coming? Is it important?
Job displacement. Are more jobs going to be displaced—
Yeah—
—and what’s the role of UBI?
I addressed that question extensively in that paper: what happens when, what happens in the next 5 years, what happens in the 2030s, and what happens in the 2040s. I cover all this in this paper.
Job displacement will happen. In 2016, I wrote a paper on this. I think it was in Forbes or Fortune—I forget. It was a long, 5,000-word treatise on that. I do a lot of writing, and my view was that AI would lead to great abundance, great productivity growth, great GDP growth—everything economists love—and increasing income disparity.
That was about 8 or 10 years ago, when I wrote that piece. I do think we will have to deal with that issue through policy, and that’s why I think societies will make different choices in different countries. It’ll be a country-by-country choice of what we allow and how we plan to share the abundance more broadly. I don’t think the current capitalist system, unmodified, will work in that environment.
Interesting.
I always say I’m a techno-optimist like other people, but with care and caring. Care being focused on safety, which is also going to be an issue, and caring for those left behind.
Thank you, Vinod.
Thank you.
Nabihah.
Hi.
How do you pronounce your name properly?
Nabihah.
Nabihah. Thank you.
Hi, Vinod.
One of our portfolio companies.
Yes.
It’s great to see you, Vinod. Thank you for supporting my journey. It’s been awesome. I would love to get both of your thoughts on how space exploration and colonization are going to evolve with humanoid robots.
I’m getting pretty excited about it. I always wanted to be an astronaut, but that hasn’t happened. How do you think this space is going to move in the next 3 to 5 years?
Very cool.
Oh, yeah. Thanks. We actually wrote about this when we started the company—or, sorry, I wrote about it.
Space would be fantastic for humanoids. It’s such a pernicious place for humans. Think about colonization and exploring the stars. It would be great to send mechanical, embodied agents out into the world, or out into outer space, and have them really help us there.
I think you start thinking about a really exciting future where we’re living off-planet and how that would work. I really hope that, even in the near term, we’re helping to deploy robots there. There have been robots developed with NASA that are up at the ISS and have been up before, so there’s pretty good precedent here.
For me, all of this comes down to where it is on the adoption curve. You’re going to see robots first in the workforce in vast numbers. You’re going to see them in the home, and you’re going to start seeing them in space. They’re not quite sequentially, but you’re going to happen to see them in certain places first, in higher quantities, and over time, this will all happen.
We’ll have robots colonizing planets. We’ll have special-purpose robots helping out there fully autonomously on Mars and the Moon. It’s going to be extremely exciting. It’s going to take some time. We have to work on this on the ground here on Earth first, but we would love to play a part in that.
Yeah.
I don’t know what you think, Vinod.
I would agree, with one caveat. People think we have to go to the Moon or into space, or mine asteroids, to get resources. I don’t think that will be the reason we do it. It’ll be because we love to explore, and I think that’ll be the reason to go to space.
I don’t think Sam would mind me telling this story. I was just talking to Sam, and he asked me a question I hadn’t thought about: When will the first self-replicating AI probe leave planet Earth? I said, “15 years.” His answer to me was, “You’re way too conservative”—not something people call me very often.
Yeah.
So it’s a fun fact.
Amazing.
Wow.
All right, let’s go to Orland on Zoom. Orland, if you’re there, what’s your question?
Yes, hello. This is for Brett. I’m in the agricultural business. When should we expect humanoids in agriculture, working on farms in the rain and sun?
Was that working?
Yeah, Brett. When do you expect to see humanoid robots working in the fields, in farming, in all weather conditions?
Totally. I actually grew up on a farm—a third-generation corn and soybean farm in Illinois—so it’s kind of close to my heart.
I think it’s a similar type of through line as the last response: robots will start indoors. It’s better for IP. It’s easier because of weather reasons and other things, and we’ll start seeing robots from there go outside. There’s tons of decent work outside that we would love to do. I would say it gets harder there than probably indoors.
All this is happening in the next 10 years. We’re seeing robots inside the workforce and outdoors. We’re seeing robots in the home. Hopefully, we’re seeing some replicating robots out into space. That’d be really cool to have. That’d be incredible.
For us, we’re focused right now on trying to get to 100,000 units shipped. That’s our near-term goal. We need a certain amount of momentum here from the business to start getting costs down, building the data pipeline, and getting robots out in the world, working fully autonomously with no human intervention.
For us, that will be indoors. The next chapter in our book will be the home and outdoors. I think all that’s happening in the next 10 years.
I would add the following: I think in the next 5 years, we will demonstrate that robots can work on farms. The capability will be there, but there’s a big difference, as we are talking about. This generally applies to all the technologies we’re talking about.
When we demonstrate the capability and it has 1% penetration, that’s when I consider the job done. But there’s a societal-change matrix that can take a long time for full adoption.
Mm-hmm.
Keep that in mind with respect to all these technologies, including farm robots.
It was about thirteen years ago, I had my two kids, my two boys, and I remember at that moment in time, I made a decision to double down on my health. Uh, without question, I wanted to see their kids, their grandkids, and really, you know, during this extraordinary time where the space frontier and AI and crypto is all exploding, it was like the most exciting time ever to be alive. And I made a decision to double down on my health. And I’ve done that in three key areas. The first is going every year for a Fountain upload. You know, Fountain is one of the most advanced diagnostics and therapeutics companies. I go there, upload myself, digitize myself, about two hundred gigabytes of data that the AI system is able to look at to catch disease at inception. You know, look for any cardiovascular, any cancer, any neur- neurodegenerative disease, any metabolic disease. These things are all going on all the time, and you can prevent them if you can find them at inception. So super important. So Fountain is one of my keys. I make that available to the CEOs of all my companies, my family members, ’cause, you know, health is the new wealth. Uh, but beyond that, uh, we are a collection of forty trillion human cells and a- about another hundred trillion bacterial cells, fungi, viri. And we, you know, don’t understand how that impacts us. And so I use a company and a product called Viome. And Viome, uh, has a technology called metatranscriptomics. It was actually developed, uh, in New Mexico, uh, the same place where the nuclear bomb was developed, as a biodefense weapon. And their technology is able to help you understand what’s going on in your body, to understand which bacteria are producing which proteins, and as a consequence of that, what foods are your super foods that are best for you to eat? Or what foods should you avoid, right? What’s going on in your oral microbiome? So I use their testing to understand my foods, understand my medicines, understand my supplements, and Viome really helps me understand from a, uh, biological and data standpoint what’s best for me. And then finally, you know, feeling good, being intelligent, moving well is critical, but looking good, when you look yourself in the mirror, saying, you know, I feel great about life, is so important, right? And so a product I use every day twice a day is called One Skin, developed by four incredible PhD women that found this 10 amino acid peptide that’s able to zap senile cells in your skin and really help you stay youthful in your look and appearance. So for me, these are three technologies I love and I use all the time. Uh, I’ll have my team link to those in the show notes down below. Please check them out. Anyway, hope you enjoyed that. Now back to the episode.
Emeka.
Thank you. This is highly inspiring. This is phenomenal. My moonshot is to advance the adoption of sustainable habitat for human beings all over the planet.
I have 2 questions, 1 for each of you. In terms of robotics, do you currently or are you planning to have some kind of policy on military applications? Where does that go? Because once you have humanoids, I mean—
Great question. What’s the second question for me?
And the second question, in terms of fusion and power grids and stuff like that, my moonshot is sustainable habitat for humanity. Is there any thought given to what might happen if we have a solar flare or some kind of serious event that could wipe out grids as we have them today through massive EMP or whatever? How do we protect our future power grids—
Great—
—from that kind of—
I’ll ask for short answers. I want to try and get a few more questions in. I know you have a very clear answer on the first—
Yeah, we won’t do anything military-related at all. It’s in our company mandate. I wrote a line in 2022. I think it severely hurt the kind of civilian commercial market that we’re in. We always have a little Terminator vibe going on with humanoids, so you really want to stay clear of that in our mind, and the market’s much bigger on the nonmilitary side.
Vinod, fragile grid?
Yeah. I’ve sort of already answered that. Between superhot geothermal, which applies to most of the Western United States, and fusion, large reactors, small reactors, I think we’ll be in good shape.
We still have to build more transmission, but my bet is we’ll—with energy becoming cheaper, you’ll be drilling tunnels under the ground—
Ground, yeah—
—to do transmission lines.
Protect yourself against that. Fantastic. All right, let me go to Cole on Zoom, and then we’ll go to Zi next after that. Cole.
Yeah, Brett. Again, under the general-purpose guise, I guess, one of the most common things that humans do is drive vehicles, so it’s a big ask to ask all agriculture and businesses to go to self-driving vehicles. Does Figure plan to have these humanoids be able to drive a vehicle and do all those things that go along with that?
Yeah. One of my dreams is to have my robots fly an Archer aircraft. That’d be— We need pilots, right? There’s a pilot shortage.
I think one thing that Vinod did a good job on is there’s the ability for the robots to do it and demonstrate it, and there’s the ability to actually integrate this into society at scale and where the focus lies. Those could be on different timescales, so we can definitely show a farming robot, perhaps even a robot driving a car down the street or something like that.
What we actually plan to do in terms of where production’s heading, where the clients are at, where we’re shipping into—I mean, on a long enough timeframe, I think these humanoid robots will do everything a human can. Physically.
Great. Zi.
Like, I think so.
Yeah. Hey, Brett. Your talk has been really inspiring, and I was really excited about the future of humanoid robots. And now I thought about something quite disturbing, which I wanted to ask your thoughts on.
You see, in the future, humanoid robots are obviously going to be connected to the grid, right? So what’s stopping a malicious hacker from hacking the system and, instead of making the robot do dishes, making the robot grab a knife and stab its host to death during their sleep? That was quite disturbing, and I would love to—
You’d love to—
See how—
You’d love to experiment with?
No, no, I’d love to hear your thoughts on how you prevent that, sorry.
Yeah, no, it’s great. It’s like you just watched a Black Mirror episode and came in here wondering what happens here.
Yeah.
We treat cybersecurity and local security on the robot as extremely important. We have a whole team around it. It’s not like one solve here to say, “Oh, we’re going to do this and leave everything in nonvolatile memory in firmware, and it’ll be fine.”
Yeah.
But we also do that. That will help. I think, in general, our robot doesn’t need to be connected to a network. The Helix video you saw of the robots in the kitchen doing it—those are fully embedded neural-net weights that live on the GPU. The robot wasn’t actually connected to any network and doesn’t need to be connected to any network.
We can basically see through cameras, we have battery power, we could run those weights through a GPU onboard, and we can output actions and torques to the motors. So I think, in general, we ultimately want to make sure nobody can get root access to any kernels in the system.
We want to set everything in terms of what the robot ultimately can never do in nonvolatile memory in firmware, which we do today. Ultimately, it comes down to the robot not being able to do certain actions people command it to.
So, trust him. Let’s go to Kartik.
Hey, guys. Thank you so much for your wisdom. Brett, you’re a moonshot venture builder. Vinod, you’re a moonshot investor. Similar to the question that’s actually sitting up there, voted up, what’s your advice to moonshot ventures and their founders and their teams in terms of building a roadmap when the underlying technologies or even the convergence of the technologies doesn’t quite exist yet?
I do believe companies are more successful if they raise in stages because it puts more pressure to achieve milestones. The more comfortable you feel, the less you will explore the options and look for better ways. Now you can—
Do you think companies can be overcapitalized, though, and become lazy?
My experience—I did this analysis a few years ago—the less money a company raised, the more likely it was for us to make more absolute dollars. I’m not even talking about rates of return. Why? Because they spent a lot more time examining their assumptions and testing them in small increments.
So this is very counterintuitive. It’s the opposite of the SoftBank strategy.
Yes.
Yes.
It’s like stressing a vine to get a fine wine.
Yeah. It’s like stressing a vine to get fine wine: detect problems early, test everything. It’s not to say you can’t do it if you raise a lot of money, and hopefully Brett’s doing that. But it generally tends to be that people get complacent, and complacency is the enemy of innovation.
Yeah. And I would say raise a lot of money early, later to scale when you have a path that—
Yeah, but there’s a time you need money to scale—
Yeah—
—and that’s when you need it.
Mark.
Thank you. With all the money being invested in AI, what’s your thinking about the fact or the possibility that it becomes kind of a commoditized service, as opposed to something for which people are willing to pay the amounts of money that are going into it?
Don’t listen to the press you read, right? If a multitrillion-dollar economy is to be created, you’re going to see massive opportunities. Take Brett’s opportunity. If there’s an industry larger than the auto industry in the next 2 decades, then it doesn’t matter how much you invest; the opportunity is large.
I would say the following: with what I see in the market, most investments, most companies will lose money—the vast majority, maybe 80%. But more money will be made than lost because of the exponential nature of the winner’s outcome.
By and large, I believe the press likes to write short-term headlines, so you have to ignore them—
Mm-hmm—
—and focus on… This is why I say I only care about 2030. I don’t care about what the stock market’s doing today. It’s largely irrelevant to 2030.
Is it a winner-take-all market?
I think this will be a fairly diversified market, and there’ll be many winners. There’ll probably be 3, 4, 5 in each area, with a power-law distribution of outcomes. But there will be more than 1 winner.
How many robot companies do you imagine?
I could see half a dozen, given the size of the market. Probably less than the number of car companies today, if I’m looking 20 years out.
Brett, how about you?
I think I agree that it’s kind of winner-take-most here. Billions of dollars need to be invested. You need to make it at a very high rate, which in manufacturing is very difficult. You need to do a lot of data collection. The robots are—it’s just a hard business.
You look at any deep-tech group, like eVTOL, satellites or whatever, space, AVs.
You just don't have a lot of groups out there that really make it. And Vinod's right: 95% of all these go bankrupt through this whole process, so it's tough, but I would say a few will make it.
Let's go to one last question here.
Thanks, Peter. This question's posed to both Vinod and Brett. The twin forces are critical healthcare worker shortages, as well as three generations of boomers coming to the healthcare system. The robots we see are primarily for social engagement purposes. There's a question online about working with elder loneliness. My question is in regard to bedside care and direct patient care with invasive procedures—for example, nurses, physical therapists, occupational therapists, and speech therapists. Do you see robots being a viable option in the future to replace humans?
Can I twist that a little bit? Do you see robot surgeons in the future being the baseline? I do.
Yeah. Look, healthcare expertise will be free in the next couple of years, well before 2030. Interventional medicine, such as doing a cath procedure, will probably take a little bit longer, and the biggest pole in that tent is the FDA, so I mentioned regulatory earlier. I expect we're more than 10 years away from an unsupervised robot surgery. Whether we're 15 or 20 years away depends on regulation.
But I want to end on 2 ideas. When we talk about technology development, there are 2 things to worry about. One is safety. We haven't talked about it, and I think in the next 3 to 5 years we will make massive leaps in the explainability of models. I suspect in 2 to 3 years you'll start to hear that this is no longer a black box, no longer a parrot—all the stuff you hear.
The characteristic of our technologies is that when you identify the problem, somebody starts working on it, and there is a real safety risk. Really valuable people like Yoshua Bengio and the people I respect, like Geoffrey Hinton, have surfaced the question of AI safety, and I think it's a real problem to worry about. But I do believe the optimistic part of me says I'm already seeing business plans explaining how these models are coming to their conclusions, and so that will hopefully get solved. So that's one on the caution side.
Let me give you my most optimistic view of acceleration. Whatever number of scientists we have in America or on the planet, within 5 years we'll multiply that by 10 or 100 times. Why? Because we will have AI chemists, AI PhD chemists, AI biologists, and AI physicists. The acceleration ahead of us from AI expertise is not only in the delivery of care, like a doctor, but in research, where an AI scientist is generating hypotheses and then testing them themselves without needing any human intervention. It's 10 times cheaper and 100 times more abundant in the amount of research. I don't think people have started to imagine that acceleration of science and technology.
Yeah. It really is a rapid intelligence explosion and science explosion. Brett, I imagine that we're going to see these AI models designing experiments, the robots implementing them and running them almost lights-out, 24/7, and accelerating all of this. Do you agree?
Yeah. Listen, both on the physical side of the world and the digital side, these agents will do more and more of what we do.
Yeah.
That's the trend.
So that's a great optimistic note—
Let's end it there.
—to end this session.
Let's end it there. Give it up for Vinod and Brett.