Do we really need 400 humanoid robot companies?

Seth Winterroth: There aren’t hundreds of millions of unique embodied AI agents operating in physical-world industries right now.

And so the question that we all need to ask is, why is that? My strong belief is that there’s still a lot of friction in the development and deployment efforts necessary. Even given the advancements we’ve had in technology, it’s still really hard to put these systems together in a way that they can be deployed efficiently and operate with five nines of reliability in incredibly complex application spaces.

Speaker: My guest today is Seth Winterroth. He’s a partner at Eclipse, one of the leading investors in robotics and automation. Eclipse has backed companies including Genesis AI, Wayve, MiND Robotics, and Foxglove. Today, we’re going to talk about where robotics is really headed and whether humanoids are, in fact, the most important form factor.

John: Hey, Seth. Great to be chatting with you today. I’m super pumped to talk about where investment is going, what we’re seeing in robotics right now, what we’re going to see in a few years, and some of the pitfalls, maybe dangers, and opportunities.

As I was prepping for this call, I was thinking of the fable—it is a fable—of this Chinese character for crisis. It means danger and opportunity. That’s not true, but it’s a wonderful fiction, and it’s kind of reminiscent of where we are right now.

We see literally hundreds of billions of dollars going into robotics, so much into humanoid robotics as well, right? We see nations like Japan and China investing massively and supporting their local ecosystems. We see that starting in the U.S. as well. There’s so much money pouring in here. There’s obviously something going on, but it also is a great danger, right? Not everybody’s going to win.

Talk about how you see the current investment landscape in robotics, and specifically humanoid robotics.

Seth Winterroth: Very nice to be with you here today, John.

As you know, I’ve been a partner at Eclipse since we founded the business almost 12 years ago, and we’ve made an applied robotics investment in every single fund going back to the very first fund that we raised in 2015. Prior to that, I spent the early part of my career at GE and really had a front-row seat to working on applications of robotics technology in industrial settings across manufacturing and industrial inspection.

So I’ve spent my entire career working on these applied robotics, applied AI, physical-world applications of AI-type challenges. For somebody, frankly, that’s been at it for 15 or 16 years, I couldn’t be more excited about the way that you articulated the landscape dynamics right now.

It’s gone from being a small cadre, a small community of folks that thought this was an important area of technology to be working on, to really mass-market acceptance across whatever mass market you’re talking about, right? Every single major enterprise is talking about their robotics or applied AI strategy.

Capital markets on the private and public side have brought significant capital to bear. Sovereigns around the world are understanding that this is a critical area of technology to focus on for industry, but also for national security. I think maybe the most important thing to me is that world-class engineers, operators, multi-time founders, and first-time founders are looking at the problem space and saying, “There are some really interesting, massive, generational companies to build now that the technology across the entirety of the stack has come of age.”

At Eclipse, we call it the five forces: founders, commercial markets, capital, policy, and talent writ large. We really see all five forces being brought to bear on this category right now, and I think that’s what’s driving a lot of this increased velocity.

John: I like what you said, that the technological foundations for this to happen are just coalescing right now.

Seth Winterroth: Yeah.

John: AI’s had a massive leap in the past few years. That’s obvious to everyone, and now the focus is often on physical AI, which is exactly what’s needed for robotics.

Just getting motors and actuators that are good enough and efficient enough, and that work for long enough without overheating, is getting to the point where that’s interesting. That’s half the cost of a robot, right? There are many more challenges to be solved, but I’m tracking almost 400 companies globally right now that are building humanoids.

Is that too many humanoid robot companies?

Seth Winterroth: The term that comes to mind is maybe creative destruction.

I’m a big student of historical technology cycles, and so it would be interesting for us to go back and look at the early 1980s or mid-1980s and say, “Well, how many personal computing companies were there?”

John: I remember Tandy.

Seth Winterroth: What you see during these times of mass-market recognition of something really important coming to bear on society, from a technology and commercial-opportunity perspective, is lots of enterprising individuals launching their ship into the abyss and saying, “We’re going to go make a play here.”

That’s really exciting. It creates competition. It creates true creative destruction, in that with the companies that aren’t successful, a lot of the learnings then get promulgated into other companies.

John: Yeah.

Seth Winterroth: Ultimately, I think that’s a fantastic thing for the ecosystem.

There will always be winners and losers, but the net-net of it, when you zoom out and see the non-smooth curve, the jagged curve, is that it will still be up and to the right over the next decade. And so then the game becomes, how do you pick the winners, right?

As an engineer or a leader going to one of these companies, or as an investor looking to invest in them, that’s why the game is so fun.

John: The other thing it creates is velocity, because maybe you don’t need to be first, but you can’t be super-crazy late.

We’re seeing people raise—what, Figure’s last raise was at a $39 billion valuation, right? You have to show velocity and progress in order to get that.

You invested in Genesis AI. Talk about what you’re seeing in Genesis, as well as some other companies that have caught your eye that are interesting, fascinating, and investment-worthy.

Seth Winterroth: Going back, you can’t be late, but you also can’t be early, right?

John: Yes.

Seth Winterroth: I think companies need to know what race they’re running.

Let’s put it in practical terms. If you are a developer-tools-focused company, like FoxGuard, for example, whose board I sit on, and you want to become one of the one or two companies that gets established in every new computing cycle and becomes the toolchain of record for that computing cycle’s generation of developers, it’s very difficult to do that on a come-from-behind basis.

You actually need to be a first mover. But if you move too quickly when there is no market, you run the risk of not making it to the right position when the wave comes, right?

John: Yes, yes.

Seth Winterroth: So that’s an example where you actually do want to be the first mover, with the right timing.

On a different example, I led the Series A and I’m on the board of a company called Wayve. Wayve’s a self-driving car company. Wayve was one of the later movers in the last decade. The company was started in 2017 or 2018, well after billions of dollars had gone into Waymo, Cruise, Argo, and Aurora.

What were they in a position to go do? They were in a position to take advantage of the absolute destruction of the legacy robotic stack—the modular map, localize, perceive, predict, plan, control software stack—and take advantage of this new thing, deep learning.

John: Mm-hmm.

Seth Winterroth: So you need to know where you’re at and which race. I would say another example of being a late mover is, look at what just happened with Project Prometheus, right?

John: You’ll have to enlighten me.

Seth Winterroth: What?

John: You’ll have to enlighten me.

Seth Winterroth: This is Jeff Bezos’s new entity in physical AI.

John: Okay.

Seth Winterroth: Not everybody can go out and raise a roughly $10 billion seed round.

John: No.

Seth Winterroth: You need to know the characteristics of the entity, the organization, the company, what’s possible to achieve, and then orient that toward how you want to run a specific type of race.

John: Mm-hmm.

Seth Winterroth: Genesis. One of the strong opinions I have right now—it’s pretty strongly held, but I reserve the right to change it as data presents itself—is that it’s very difficult to build a company with general-purpose intelligence and then transmit that into a physical morphology, close the loop, and create the flywheel of how you improve that intelligence by isolating the edge-case deficiencies and then being able to include that in your pre-training dataset, and to have the right evaluation frameworks, if you aren’t going full stack.

If you aren’t going all the way to the hardware, if you’re not co-developing hardware and embedded software and model and evaluation infrastructure and the deployment apparatus yourself.

There’s another school of thought right now in these general-purpose foundation-model-for-robotics companies that says, “Hey, we’re only going to focus on making the egocentric foundation model as robust and capable as possible. And if we get that right, it will promulgate to any hardware morphology.

“We’ll basically be able to operate as the cognitive models and the LLMs do, and provide API-level access, and then the application developers will take over from there.”

My intentionally cheeky analogy would be that operating in that way would be like trying to build Android when there are no mobile OEMs operating at scale to combat Apple’s integrated approach.

The reality is, the dirty secret in robotics right now is, where are all the robots?

There really aren’t that many outside of Amazon and the traditional industrial folks. You’re starting to see an uptick in defense, certainly, especially on the drone side of things.

John: Yeah.

Seth Winterroth: But they’re not among us. There aren’t hundreds of millions of unique embodied AI agents operating in physical-world industries right now.

And so the question that we all need to ask is, why is that? My strong belief is that there’s still a lot of friction in the development and deployment efforts necessary. Even given the advancements we’ve had in technology, it’s still really hard to put these systems together in a way that they can be deployed efficiently and operate with five nines of reliability in incredibly complex application spaces.

We can prove that they’re safe. There’s still a lot of surface area to that problem. And if you don’t wrap your arms completely around that problem space, you are subject to increased friction when it comes to meeting product requirements from a performance perspective, from a safety perspective, from a reliability perspective, and then ultimately getting to the unit economics necessary to have a great business.

John: That’s really challenging, though, right?

Because going full stack, which is very attractive for a lot of founders—they want to own it, control it, the Steve Jobs model to some extent, right? We see 1X going incredibly vertical. We see Figure going fairly vertical as well. We see different models.

We see Apptronik saying, “Hey, you know what? Our brain’s mostly going to be developed by Google DeepMind and others, and that’s going to benefit other robot makers as well, but we’re going to focus on what we’re doing here.”

It’s such a core challenge because a Bessemer report that came out three or four months ago said, yes, the full-stack providers have the advantage initially, maybe not long term. But you’ve tried to do everything. Maybe you only achieve 75%. You’re okay, but maybe Apptronik, which has been in the space for decades and done very successful stuff in robotics, or maybe others that are focused on one job, get amazing at that.

We’ve seen that in the past, right? You talk about the history of technology adoption. We’ve seen a feature become a product when it was just insane and incredible, and become a platform, right? Messaging is a good example, perhaps.

So it’s really challenging, especially if you haven’t raised the $3 billion that some startups have and you can’t focus on everything. I don’t have an answer for that, but that’s not an easy problem to solve.

Seth Winterroth: Definitely not.

Definitely not. It goes back to what we said earlier about knowing which race you’re running and which race you’re capable of running.

John: And winning.

Seth Winterroth: Well, winning ultimately comes down to that, but also great execution and running the race fast.

With these full-stack companies, if you’re taking a full-stack approach, you also want to be very clear about what is core for you to develop and what actually is now available off the shelf, right? There’s a lot that’s out there. You don’t have to do everything yourself. You don’t have to have that level of ego, right?

You can say, “Hey, we’re going to focus on what’s core and accelerate on that front.”

You do need to be able to raise significant amounts of capital. I would say more capital than if you’re just focusing on a horizontal layer of that full stack, maybe outside of the foundation model. Anybody who’s trying to pre-train large models is going to need significant amounts of capital.

But then you also need to be able to convince the capital markets that you’ve anchored on a commercial application that is of a magnitude, and you’re uniquely positioned to address it with multiple stack functions of improvement in the eyes of the customer that you’re serving, and that there are many of those customers, such that the capital markets feel like the juice is worth the squeeze.

John: Mm-hmm.

Seth Winterroth: I’m not afraid of companies going deep into the J-curve. There are ways in which you evaluate excellent execution even through the depths of that J-curve.

But getting to the other side of that J-curve and then beginning to commercialize—what’s so interesting about these businesses is that your commercialization slope doesn’t look like an incremental one from there, because the nature of the industries that you’re addressing is such that they’ll procure. They’ll do small-volume deployments, prove it works and that they have an ROI, and then they’ll go all in.

John: Mm-hmm.

Seth Winterroth: And so oftentimes what you see happening is, you emerge from this J-curve, there’s a year of POCs or small-scale production deployments, and then your expansion isn’t a 2X or a 3X of that initial commercial footprint.

It can be an eight- or nine-figure expansion because of the scale and magnitude of these customers and the degree to which the solutions you built can change their structural unit economics.

John: Yeah.

Seth Winterroth: That’s what we’re starting to see now in the robotics space, and that gets me very excited.

John: Understandably. I want to talk about that a little bit in terms of timing, because you mentioned earlier the dirty secret, the ugly secret: Where are all the embodied AI units, right? Why are there not hundreds of millions out there?

It doesn’t feel like robotics, and humanoid robotics in particular, is like quantum, though, which is perpetually five to 10 years away. It doesn’t feel like that because we’re seeing the progress. Figure 01 to Figure 02 to Figure 03 has been impressive. Figure 04 is under development right now, and the CEO recently said on social that it’s a bigger leap than any of their other generations. And frankly, they need it.

NEO from 1X is going to be shipping in December. I’ve had a personal commitment to that from the CEO and from the head of product design, so in December they’ll be shipping. Different market, right? It’s the home market specifically. Not saying Figure’s not attacking that market, but Figure’s attacking a broader market base, I think.

Seth Winterroth: Yeah.

John: And what we saw from 1X recently in NEO’s hands, that’s an impressive hand.

That’s a very impressive hand. So it feels to me like, okay, we have robots in homes right now. That’s mostly people who are connected with companies that build them. We’re going to see the commercial shipping start in December.

I mean, you can buy a Unitree or an AgiBot today, obviously, and ship it to your home. It probably has hands like this, right? Unless you spring for expensive ones, and it probably doesn’t do a lot. You can make it dance and do other stuff like that, but I’m talking real robots that do things, right?

I feel like we’re two to three years away. I mean, that’s probably six or seven models, maybe nine or 10 model generations, for a Figure or a 1X, and the leaps that we’re seeing already are impressive.

Seth Winterroth: You have been really focused on this biped and quadruped physical instantiation of a robot.

I would like to zoom us out just a little bit and say that my definition of a robot is intentionally extremely broad, right? It’s any system that operates at the edge, senses locally, and then computes an understanding of what it’s sensing through a connection to the cloud or computes locally, and then takes more efficient action or informs more efficient action to be taken by humans.

Right? And by that definition—

John: Is it mobile?

Seth Winterroth: Not necessarily.

John: Okay.

Seth Winterroth: What is—

John: A dishwasher fits.

Seth Winterroth: A dishwasher fits. We have automation in the home right now, right?

John: Yes.

Seth Winterroth: What is a Verkada or a Flock Safety intelligent computer-vision-at-the-edge security system? Is it just a camera now, or is it an embodied AI agent at the edge informing more efficient human action?

So the highest-volume modern robot architecture coming to market right now at scale is?

John: Robot vacuums.

Seth Winterroth: The automotive—

John: Cameras.

Seth Winterroth: The automotive passenger vehicle.

I was in my Model Y and did an out-and-back from my home to the Bay Area, hundreds of miles, flawless performance, right?

These vehicle architectures that are—

John: Full Self-Driving?

Seth Winterroth: Full Self-Driving. It had a little bit of issues with parking, but the vehicle architectures that are coming out now from Mercedes, from GM, from BMW, from Toyota, are more than capable.

They’re running Thor or Qualcomm 8650s. They’ve got the ECU architecture in place. They’ve got advanced ADAS. They want to run a high-value software workload that enables humans to have an autonomous passenger experience.

John: If only they had hands—

Seth Winterroth: If—

John: Or grippers, or could do things.

Seth Winterroth: Right. But okay, the societal impact of autonomous transportation becoming a full reality over the next decade could be transformational.

So I think humanoids are a really interesting application of robotics. It could be the biggest market of all time. That’s why so many people are going after it.

We still have a lot of work to do to get product-level capability in whatever applications are being attacked, whether it be industrial or consumer, to a place where it delights. And that’s purely on a performance basis.

Then there will be a long road to scale, to good unit economics, to the addition of multiple applications within one physical hardware platform. So I think we have a long way to go still on that product form factor.

And while it’s an exciting one, I think there are lots of other areas where autonomy will touch society at scale with tremendous economic value in advance of those humanoid form factors.

John: Yeah. Can’t disagree too hard with most of what you’re saying. I’ve been in Waymo, enjoyed that. I have a Tesla and didn’t really enjoy Full Self-Driving, but opinions differ, obviously.

It’s always the perennial question: Will I have one machine, or will I have 20, right? And I’m oversimplifying. Nobody’s going to have one, even if you have a humanoid, right?

Do I have a machine that cuts my lawn, a machine that vacuums my house, and a machine that washes my dishes, or do I have a machine that does all those things using other machines?

Seth Winterroth: Yep.

John: And there’s no perfect answer there, and there are going to be shades of gray. Where do you slide on that continuum?

Seth Winterroth: In the consumer robotics application space, I am on the continuum of—

I think that the most simple, most basic product that engenders the first initial delightful experience with the customers, and has a team that knows how to rapidly learn from the customer, translate that into product requirements, translate those into engineering requirements, execute with high velocity, test, and ship—that flywheel is going to be a winning product orientation in this space.

Because I haven’t seen very many consumer products in history that land with the Holy Grail system right out of the gate. I mean, even think about the iPhone. The first iPhone was pretty basic until about iPhone 3.

John: Yeah.

Seth Winterroth: That’s typically the consumer product journey.

And so what I’m looking to see, and I don’t really know what it is yet, is what is the minimum viable set of product features that a system has to be able to execute inside a consumer home environment—

John: Home environment.

Seth Winterroth: —that results in that delightful consumer product experience that we all know, right? We all can think about a product experience we had where we were just, “Oh my God, good luck taking this thing away from me again.”

And I don’t know what it is going to look like in this environment, but I think it’s whatever is simplest.

John: And I think that changes a little bit. I’m going to use a word I hate, which is “phygital,” right? Physical, digital. I know, it’s awful. It’s so ugly. It’s the worst word ever, but it’s the only thing I could think of in the moment. I’m going to use that concept.

Seth Winterroth: Yeah.

John: Because, as you know very well and as our listeners know incredibly well, the concept of a product has changed immensely over the past two to three decades.

A product used to be a thing, and you would buy a thing, they would ship a thing, it came in, the thing was the thing, and the thing did the thing’s thing, and that was the thing. Right?

But now a product is both physical and digital—okay, I’ll use that construction—and it grows and changes over time.

What made the iPhone not just the, “Holy shit, that’s amazing, love it, want it in my hand,” but what made it the thing that created the universe, created the ecosystem, was the App Store.

Seth Winterroth: Mm-hmm.

John: That’s happening in robotics already. We see robotic app stores.

And the extension there is that, for instance, 1X’s NEO will ship with less capability than its hardware enables, and it will add capability through over-the-air updates to do X, Y, whatever list of things.

I think you’re right that the first experience of a new product, whether it’s a humanoid robot or a different type of robot—the new robot dog thing that the inventor and CEO of Roomba is just launching now as well—what makes that work or not work is, “Hey, wow, that’s awesome. Love it.” It does one thing really well, or three things really well, whatever.

But I think the thing that makes it an ecosystem is that it takes those three things and it multiplies them, and it’s nine, and it’s 50, and it’s 100, and it’s 1,000, and then it’s just indispensable.

It’s part of my life. Take it out of my cold, dead fingers.

Seth Winterroth: That’s right. You wake up and it all of a sudden has a capability to perform some function that you didn’t have the day before, and it engenders some small amount of delight within you.

So I think I’m looking to see who gets that minimum viable product with the right product economics.

Consumers are fickle, right? Price points matter, and the economics of that first solution matter. And let’s be honest with ourselves: What is a home robot? It’s really just a consumer electronics device.

John: Yes, it is.

Seth Winterroth: And a consumer electronics startup—that’s startup on X Games mode.

There was this statistic a few years ago that came out: 98% of all consumer electronics hardware startups fail. It’s hard. We could do a whole podcast on why that’s so hard.

John: Where I think maybe it works is that I’m not buying a robot, I’m buying a clean home.

I’m not buying a robot, I’m buying peace of mind. I’m not buying a robot, I’m buying whatever. And you understand that 100%, obviously, as well.

Okay, we’ve got to broaden the conversation. We’ve been talking about humanoids a lot, and it matters, and it’s important, and it’s also the sexiest part of the whole robotic revolutionary era that we’re going through.

It’s not the only form factor. I talked to the CEO of the Association for Advancing Automation recently, and he said, “You know, we’re not going to have 10,000 humanoids in a factory.”

And I believe that 100%, right? We’re going to have some great robots that do things. We’re also going to have a lot of automation. Maybe they’re all robots according to your definition, right?

Talk about what you see doing real work. There’s a $50 trillion market globally for hands, right? Physical work getting done. That’s sort of the TAM that these robotics companies are building massive valuations on top of.

Seth Winterroth: Yeah.

John: What do you see working there?

Seth Winterroth: Historically, Kiva Systems kind of changed everything for robotics. It was the first real commercial modern robotics company, and I like to say it was the acquisition that launched a thousand robotics ventures, right?

And where did that happen? It happened in a constrained supply chain and distribution facility where floors were laser-leveled and you knew where all your inventory was, right? You could map it easily.

And the economic value proposition was to move more things at a lower cost structure, with higher degrees of throughput and operational uptime through the roof.

This was a very clear equation. And I think when you look at the robotics landscape over the last—what, they were acquired in 2012?

John: That’s what I was about to say.

Seth Winterroth: So, 14 years. That’s really still where we see a lot of the value today, in manufacturing and supply chain logistics.

We’re beginning to see transportation. We’re seeing some surgical applications come of age. But what do all those have in common? It’s a relatively constrained environment.

John: Mm-hmm.

Seth Winterroth: A set of control policies that robots are really good at: precision control, decent dexterity, but not human-level dexterity required.

The value of the application affords you the ability to really instrument the sensing suite, so you have superhuman vision of an environment, which obviously helps with the fidelity of path planning and the control policy that you execute.

And so that’s where we’re seeing things come of age.

There’s a great line that any profound technology, over time, kind of just disappears into the fabric of society. And so that’s what we’re beginning to see in supply chain logistics and in manufacturing. We see a lot of automation.

Now, where I think it will accelerate over the next decade is into some of these more—in automotive, for example.

When you go into an automotive manufacturing facility, you see a lot of KUKA Titans. You see them lifting chassis and doing mobile conveyance and all this kind of stuff.

Where don’t you see robots? In final assembly. You still see a lot of human hands there. So I think you will begin to see that shift happen over time.

In distribution, you still see a lot of the individual-item manipulation, the sortation of it, or the each-picking from the container on the shelf happening by humans. I think that will transition over the next decade.

John: Yeah. I’m pretty sure you guys have invested in a lot of that, right?

You mentioned Wayve, but you’ve got MiND Robotics, right? You’ve got Foxglove. You’ve got Third Wave Automation, which is autonomous forklifts.

Seth Winterroth: Right.

John: You’ve got a bunch of different things there. So this has been a wide-ranging conversation. We’ve kind of gone everywhere.

Seth Winterroth: Yeah.

John: And it’s been cool. I’ve enjoyed it. It’s been really great.

We’ve got to bring it to some level of a close here.

I did the timing thing where I think that in two or three years we’ll have robots in our midst, whatever that means. Is it legal to take a humanoid out for a walk with you through city streets? I don’t know. Is there a regulation for that?

What if it kicks somebody in the nuts, like that Chinese robot did when it was dancing, right? Is that a problem? What if Figure’s robot had whacked the First Lady in the face instead of just walking nicely beside her? I don’t know.

But I think in two or three years we’ll have more of those walking among us. What do you see as the next inflection point, let’s put it that way, for the mass distribution of embodied physical AI units?

Seth Winterroth: I would say two things, and you’ll forgive me for talking my book just a little bit.

John: Show it.

Seth Winterroth: In supply chain and manufacturing, a couple of companies got it right in the previous decade. And by right, I mean they drove value for the customers and they generated good unit economics for themselves.

But a lot of GMs of distribution facilities that were pushed by C-suites in the last decade to adopt automation ended up with one good system and 19 shitty robots. And they kind of said, “This isn’t driving better throughput at a lower cost structure. Let me do what I know, which is manage 200%, 300%, or 400% annual turnover.”

I think we’re getting out of that trough of disillusionment from the perspective of the customer.

And so I think you will see supply chain and distribution and many manufacturing applications really see the true value that’s been promised by these types of automation solutions—modern automation solutions, flexible, software-definable automation solutions—over this next decade.

In a way that really maps to true economic value for both the customers and those companies providing solutions. That’s the first thing I would say.

The second thing I would say is that the world isn’t keeping pace with what I think is going to be one of the most transformative things for the world over the next decade, which is autonomy being brought to bear at global scale in the transportation sector.

John: Mm-hmm.

Seth Winterroth: It’s done, I think. That problem is solved. We have optimization. Waymo’s got regulation, and all this kind of stuff.

But last decade there was all this hype and fervor, and our kids won’t have driver’s licenses, and that didn’t quite pan out.

This next generation, I don’t know that they will ever have to drive. My two-year-old, I don’t know that she will ever have to drive a car if she doesn’t want to. It’ll be a novelty.

And so you think about the derivatives of that from a business cost structure, from where people live, how it impacts real estate values. The number of driving-related deaths globally will crater, which is just a fantastic positive externality.

So I think transportation is really going to be an area.

And then, because I’m just so excited about the space, I’ll give you one more. In 20 years—let’s say 15 to 20 years—we will look back as a society on the fact that we let humans perform surgical procedures on other humans as about as medieval and archaic a societal reality as imaginable.

And so those are three areas that I’m really tremendously excited about. And yeah, that’s what I would say.

John: That’s pretty cool. And one thing I’ll highlight, which is included in your second point, is delivery drones. We are getting really good at that.

Seth Winterroth: Yep.

John: And the thing that’s holding us back right now is regulation in different areas.

There are places in the world, some in Europe and many in the U.S., where this has been normal for a couple of years now.

Seth Winterroth: Yeah.

John: We have air traffic control systems that manage overlapping drone armies of delivery bots—flying drones.

And so this is largely a solved problem, as you mentioned, and it’s about the political world, economic world, and regulatory world catching up.

It’s an exciting time we’re living in. There’s so much going on. I want to thank you for this time and for this chat, and we’ll chat again sometime.

Seth Winterroth: My pleasure. Thank you so much for having me.