Formic CEO: 95% of US factories have no robots — capex-free robotics-as-a-service can change that

There we go. Uh, it's great. But we have been uh keeping our next guest waiting. We will bring him in. We're talking about robots with Formic. Uh forco is the website. Let's pull it up. Full service robotic automation for Z capex. That's what we like to hear. We love when capex is free. Welcome to the stream.

How you doing today? Welcome to the anti-capex club. Good to meet you guys. Thanks so much for having me on. It's a great club. Yeah. Fantastic. Uh, can you introduce yourself? Give us a breakdown of the company, where you're at, and what you're building? Sounds good. Yeah. Uh, my name is Salman. I'm the CEO of Formic.

Uh, we're building the robot army that is helping American manufacturers automate. Uh, you can see some of them on the screen behind me. These are our robot babies, as we affectionately call them.

Uh, we have hundreds of them across more than 100 factories now in the US that are using our robots to produce all kinds of products. So, we're making everything from metal parts for aircraft and automotive, plastic parts for lawnmowers and golf carts, uh, as well as things like food and beverage.

We we make all the all the nachos for Chipotle and, uh, uh, dog food and cat food and laundry detergent. Uh, you name it. You know, our robots are helping to make it. You make nachos robotically. We are indeed. Yeah. The last guy I heard Eric was on, you guys were talking about bring uh, bring manufacturing back.

Like, it's real. It's happening. Uh we're really excited to be part of it. Yeah. What do you What do you have to say to the naysayers that say we'll never make it in America? The haters. The haters. Look, uh I lived in China for 25 years and I saw the industrialization of that country.

And it's like it's not a two-year process, right? Like industrializing is a a 20-year journey at the minimum. Uh and and and probably much more. And you know, we do have 250,000 factories in America uh that are making all kinds of products. So they exist.

Uh the problem is uh we don't have the labor force to support them, right? It's not just about cost of labor. There are millions of unfilled jobs in factories today. The typical utilization rate of a factory in America is less than 2,000 hours a year.

Um this the the comparable number in China is 7,500 hours a year, right? Like there's 8,700 possible production hours in a given year. The vast majority of American factories are sitting around collecting dust 75% of the time. Uh, and it's not because like there isn't demand for their products.

It isn't because there isn't raw materials available, right? Like it bo basically just boils down to the fact that there aren't people who want to work in these factories and do these backbreaking jobs day in day out.

Um, and so until we automate a bunch of them, we're going to have a really hard time uh re-industrializing. So, you know, I'm optimistic, Jordy. I think there is a lot of opportunity to re-industriize and and bring more manufacturing back, but it's not going to be uh in a year or two, right?

like we're we're helping make that happen faster, but it's going to take some time. Uh like a a really dumb criticism that I don't agree with of this company uh that might be completely wrong would be like, hey, this is just uh this is just financial innovation.

You're just you're you're you're swapping capex to opex that allows you to grow really fast. I think that could be good for industization. Um but uh how much of that is key to the business? And then what else are you doing? Uh I imagine that you're not uh you know mining the metal to make the robots.

Uh you might even be purchasing the robots from other people and then designing software suites and integrating. Um but you know Palunteer goes to oh it's just a consulting shop or whatever.

Uh what is the criticism of your company of like of like where you're starting and then why is that the right path to start and then grow out from? Yeah, I think it's a really good question. Um I think it starts from like the basic premise that uh people most roboticists are working on the wrong problem. Right.

I I was a VC for 10 years. I was investing in robotics companies. I I funded like 50 different robotics companies and saw all the different ways in which they they had challenges. And what I realized is like look, any roboticist will tell you uh getting a robot to do a job is 10% of the work.

Uh 90% of the work is like how do you do error handling and how do you solve all of the [ __ ] that happens after you deploy that robot. Yeah. And I think that's the part that's generally underappreciated, which is like we'll see a robot doing a backflip in a cool YouTube video. But once you try to robot, you know, 99.

8% uptime, um, you start to encounter all kinds of issues, right? Joint failures, maintenance challenges, programming issues, your infeed, like your product is just has a lot more variation than you expected. Um, and I think it's generally underappreciated.

And so the reason that factories in America don't automate today is not because you can't get a robot to do this job. The problem is like all the surrounding infrastructure doesn't exist.

And so if you're the factory that focuses on making chocolate chip cookies, uh, and you're really good at the best chocolate chip cookie recipe and baking, like you're not also super good at deploying robots and managing them. Uh, and so what happens is like they're just kind of overwhelmed. They don't know what to get.

They don't know how to get it. They don't know how to manage it. They don't know how to maintain it.

Um and so uh what we realized is like you have to build all the surrounding infrastructure like robotics has this last mile problem where if you can make it easy to scope, deploy, manage and maintain the robot uh then you can drive mass adoption. Uh and uh so we built a bunch of software that does that.

We do uh we use we use computer vision and LAR scans to do kind of full site evaluation in an automated way. This used to take months to walk through the facility and measure every single thing to figure out what you can automate and and what kind of robot you need. We do that in minutes.

Uh then you need to figure out what robot to to to build and deploy. So you need to basically simulate different robot arms doing that job with different grippers, different conveyors, different safety scanners, different fencing. Like you need all this stuff uh to make a robot useful.

Then you need to actually program the robot to do the job. And so we basically built AI that automates the process of programming the robot. Um and then lastly, once that robot's deployed, you have kind of all the ongoing management of that robot, right? So how do you do teleoperation? How do you do error handling?

How do you do preventative maintenance? um how do you gracefully recover from all the different types of errors that you're going to encounter? Uh there's a ton of infrastructure that needs to be built. And so like 90% of roboticists that I meet are working on how do I make the robot a little bit smarter.

Uh and nobody is working on this giant problem which is the reason the robots don't aren't deployed. Right? I can tell you from the hundreds of robots that we've deployed today, less than 5% of my deployment cost is programming the robot to do the thing.

That's like it's like it's it's it's completely, you know, delusional how many people are spending their time trying to make the robot a little bit easier to program. Like that's inconsequential to me. Like I need to focus on solving the other 95% of the of the cost of deploying robots. Uh so I don't know.

Can you talk about uh can you talk about landing on CPG and food and beverage broadly? You you said you invested in 50 plus different robotics companies. I'm I'm assuming you saw a bunch of, you know, success and failure and I'm I'm I'm assuming that was super intentional to land here.

Um, and then I want to ask, you know, potentially, you know, other categories that you're excited about. Yeah, I'll be honest, it wasn't that intentional. We started out uh going really broad. We were doing things in in metal fabrication. We were doing things in plastic injection molding.

We're doing things in CPG and a bunch of other industries. Um, ultimately, we narrowed back into CPG uh for the for a couple of reasons. Number one, it's the actual it's the industry where there's the most amount of utilization, right? So, the typical CPG factory actually runs 20 to 24 hours a day.

Um, in things like metal fabrication, like like parts, factories that make metal parts for aerospace and automotive, they typically run, you know, eight hours a day of of production, but actually only like six hours a day of actual uh parts being made.

There's like two hours a day where people are setting up and tearing down. Um so so high utilization means automation has a much much bigger short-term impact.

Number two, it's the industry where there's actually a lot more of it happening domestically u in metal and plastic manufacturing and other stuff like a lot of it is still being imported today from from um from China, from Mexico, from Canada, from other places.

Um but because CPG products are generally low value and low margin, it's not worth it to ship them across the ocean. So a lot of that is actually happening uh domestically. Um, so there are a lot of factories that can use automation. And third, those are the jobs that are the most backbreaking, painful jobs, right?

Like a m you can afford to pay a machinist 50 bucks an hour or 60 bucks an hour because that's like a high value, high margin product. Uh, you can't afford to pay somebody who's packing boxes more than 15 bucks an hour. Uh, and so and they're still picking up super heavy, you know, cases of drinks and things like that.

Exactly. Yeah. It's backbreaking. Can you take me through kind of a market map or uh like the the the the topography of robotics because I've watched how it's made. I've seen, you know, a little scooper. I don't even know if I'd call it a robot. It's more just like a machine. Then there's, you know, two axis gantries.

I'm seeing a lot of robotic arms, a lot of six axis stuff. Uh and then you get into the humanoid world.

uh where's the biggest opportunity in terms of or where's where's the most need to deploy and like what type of robot is the most important to just roll out in America right now um yeah I think like articulated sixaxis robot arms are still the vast majority of use cases right um I think if you look at what humans are doing in these factories like they're typically using their arms not other parts of their body so so I think it's just an indicator of like that that's the most common task.

Whether you're talking about uh assembly, whether you're talking about uh quality control, you need kind of vision obviously.

Uh whether you're talking about um packaging, whether you're talking about even kind of primary food handling and sorting in the metal fabrication world, you need a lot of machine tending, which is, you know, picking up a blank part and putting it into a CNC machine. We have a bunch of robots that are doing that.

All of those are just kind of articulated arms uh that can do it. So, I think there, you know, there are use cases for the humanoid form factor to kind of do new new types of tasks.

Uh we also have some gantry robots that we've deployed that do certain types of applications where it's like very high payload um uh work but uh you know the to be honest like the form factor is kind of irrelevant because in all these types of robots that you're talking about you have a couple of motors and some motion control right like it's actually uh if you can abstract away the software enough like all of those things are are essentially different variations on the same on the same problem.

I I want to talk about the capex thing. Um, what's the damage uh for a company if they're not working with you to uh buy a bunch of six axis robot arms these days? Uh, can you give me like an order of magnitude for how much these these robots even cost? I've seen a couple of Yeah.

Or maybe even some of the history some of the history. I mean, we've seen so many companies try to automate, you know, bring in robotics, right? like Nike has a history of of trying to do this in in Latin America and just failing miserably despite having seemingly competent partners.

So, um yeah, I'd love some additional context there. Yeah. Yeah. So, um uh I'll start I'll start with the second question first and then I'll move to the first question. So, uh the vast majority of robots in America today are deployed in uh very very big companies.

So, General Motors, Proctor and Gamble are two examples of like the biggest buyers of robots in America. They buy like a couple hundred robots a year. Um, and they have very large in-house engineering teams who can use those robots to to to do different kinds of tasks.

Um, the the problem is 99% of factories in America are small mediumsiz businesses. And the logic follows like if you know if you take a single General Motors plant uh that makes cars uh there's 4,000 small factories that make the components that go to General Motors that get assembled into a car, right?

So there's one guy who makes the windshield and one guy who makes the screws and another guy who makes um uh uh you know the tires, right? Like every single one of those smaller factories is not automated at all today. Like 95% of them don't have any robots. So um the the cost is a part of the question, right?

which is yeah it costs between a couple hundred thousand to a couple million dollar uh to build a fully automated uh robot work cell. Mhm. The the interesting thing though is if you break down that cost uh about 30% of that cost is the hardware itself, right?

Like the robot arm and and and the conveyor and that kind of stuff. 70% of the cost is all the custom engineering that you typically have to pay for uh for integrators and consultants and this and that to come in and try to customize it to your process.

Um and if you're a small factory like you don't have the capacity to kind of get rid of that 70% custom engineering cost uh because you don't really know uh anything about robotics. Uh and so the reason that we built all this kind of infrastructure tooling around it is that we're cutting out that 70%.

it costs us significantly less to go and buy and build and deploy those robots uh than it would be for them to try to do it on their own. Uh can you walk me through uh who the best uh robot builders are in the world right now?

What is the what does the supply chain look like in terms of constructing, building, R&D robots? Uh who are the major players globally? Who are the upandcomers? I I I don't even know if I could name a single robot maker.

Um but I'm interested to hear kind of like what the landscape looks like, who the who the major players are. Yeah. So the the biggest robot companies in the world are all not American, right? So Fanic uh is Japanese. They're they're one of the biggest ones.

Uh Cuka used to be a German company that got acquired by a Chinese company. uh ABB, um Yasawa, uh you know, Mitsubishi, you know, there's a bunch of these kind of very very large industrial companies that make robot arms.

Uh the the thing to note though is that like buying a robot arms is effectively kind of useless on its own. Uh you need a full robot work cell to make that work. So you need PLC's that often come from like Seammens or Rockwell. You need sensors that come from companies like K. Uh a programmable logic controller.

So it's like the it's the thing that controls all the peripherals and safety equipment. Um yeah and then uh you need like conveyors, you need safety scanners that are often made by like Kian Cognex and a bunch of other company. You need grippers which is the end of arm tool.

Uh so uh today like the vast majority of people who deploy robots basically hire a consultant that goes and buys all these different components for them and then assembles it and then deploys it for them. Yeah, I saw this at Hrian.

They have a they have a six-axxis robot arm and they have this like invisible wall that if you put your hand through it'll just light curtain. Yeah, light curtain. That's what it's called.

How uh why has automating 3PLs proven to be such a challenge you know there's been a bunch of you know Amazon had their uh internal efforts.

I think there's like six river systems which it was maybe Shopify acquired them but it it just seemingly is uh such a difficult task and you would think that you know grabbing items and putting labels on them and you know shipping them should be uh it seems straightforward but uh clearly not. Yeah.

I mean I'll I'll I'll note that like we don't know a ton about 3PLs because almost all of our work is in manufacturing. logistics is a much much smaller industry and like we're not in it today. Uh I will say right like logistics has to deal with a lot more variability than manufacturing.

Um because if you're if you're running a 3PL like any kind of product could come into your warehouse any time. Uh and so you need systems that can deal with all of them and there's just there's just like physics based constraints on that, right?

like if you're picking up uh a you know a a 1 oz bag of something versus a 50 50 lb uh box like you just need fundamentally different physical hardware.

Um and that's why I think a lot of this kind of humanoid stuff is is very strange because like if you're picking up a 50 lb box with with one of those robots like it's it's very likely to tip over pretty easily. Like you need and if you want to move at any kind of speed, right?

Because if you take a 50 lb box and you move it uh at you know 5 miles an hour uh your effective load during that acceleration is like double or triple uh that right so it's like 100 pounds of load that you have to be able to handle which is just kind of impossible.

Um if you move to manufacturing like you have a lot more consistency because every production line is kind of making the same part uh or the same product at relatively higher rate. Um and so the the technical complexity is lower but the operational complexity is higher. Interesting.

Um on on the note of humanoids, uh are you worried about depreciation with humanoids? I think this is a concern if if humanoids end up, you know, working well but then following the same path as like EVs because uh there's like high, you know, there's high wear and tear on cars, right?

They're just like going down a road all the time and that's challenging. And you can imagine humanoids are walking around with these heavy uh loads and then uh simultaneously like all of that interaction just feels like um I worry about the economics of humanoids.

If you buy a rob a robot for 50k and the next year it's worth 10k. It's like well you almost could have hired a human for that kind of uh cost. But I'm curious what your point of view is. Yeah I think it depends on what um what the task is, right?

Like if you're using that humanoid to do an occasional task, right, like pick up your laundry and and put it into the basket for example, like you don't worry so much about wear and tear because it's low frequency.

If you're talking about manufacturing use case, right, like these robots behind me, all of them operate 24 hours a day. Uh and so at that cycle rate, like you're doing maybe 50,000 cycles a day. Uh at that rate, like the wear and tear you get on all the components is very very high.

Um and uh like a six- axis robot has six motors. Uh a humanoid, most of these humanoids that we've seen have something like 30 to 50 motors, right? And so just you're just talking about like the wear on the motor, like we have to go out and lubricate these joints and and and and replace motors on a pretty regular basis.

Um I can't imagine if I had to do that for 50 motors instead of instead of three, right? Yeah. Uh, speaking of motors, um, I saw that in the latest, uh, Boston Dynamics demo, they said that they switched to electric motors.

Is that a meaningful shift in robotics or is that something that's unique to their approach to building humanoids? Um, yeah. So, most of the robots that we use use servo motors. So, they're all they're all kind of electric motors. It's pretty it's pretty standard.

Um, some of some of the new humanoids are using uh linear actuators. uh which is a newer technology and I think the Tesla humanoid is doing that. Uh it's uh like the exact kind of wear and tear profile on that is not clear and a lot of them are actually building their own uh their own motors from scratch.

So um like like the Chinese like I think this is a place where China has a huge advantage compared to like a lot of the American startups uh uh uh like unitry for example which I'm sure you guys have seen those videos like they have uh built an entire team to build uh actuators and a whole team that's building their own custom LAR and a whole team that's building their own kind of custom sensing across you know for most of these American companies it's kind of inconceivable to go and try to invent your own motor and try to figure out how to build a humanoid Uh, but in China, it's possible to be much more vertically integrated.

That's so odd because you have to imagine back in the dawn of the automotive revolution.

Like Ford was certainly building motors and wheels and tires and seats and the frames and stuff like we did have the ability to have a vertically integrated supply chain at one point in America and then we just decomposed it so much that uh it's been it's been a struggle to get back to vertical integration.

Uh very very tricky. the management consultants got hold of us. They did. They did. Um uh yeah on on on humanoids. Um I don't know if you've been following what uh physical intelligence has been doing with some of the endto-end models this idea of generalization.

Um are there any are you looking at any software approaches that are more uh I don't know big data or or uh you know um deep learning driven to handle more of those edge cases to automate uh the the last 90% of the work that you're doing uh and what are you what are you following on the AI side of things these days?

Yeah. Um so I think these vision language action models like what physical intelligence is doing is are are very promising. M uh we're running trials on some of those vision language action models for some of our use cases.

Um the problem today is that they're just generally too slow uh for us to be able to use them in a kind of commercial setting, but I think that'll get better in the next few years.

Um so there are like you know in the common trajectory for us is like we'll we'll start out in a factory with one or two robots and then the factory will come to us and say hey can you help me like automate all these other tasks on my production line? uh and today some of those are solved and some of them are not.

Um and so as we go uh you know as as these vision language action models get better and other kinds of uh techniques around you know deep learning get better we'll start to go and address uh more and more of those tasks. Um, yeah. I mean, this is great. Yeah. Yeah. Continue.

I was just going to say, you know, like I I think that the the thing to note here is like we're just at a very different place compared to uh like more established automation economies, right? Like in China right now, uh last year there were about 400,000 robots uh deployed.

Uh in the entire US, we had about 30,000 uh robots deployed. less than a tenth uh of of the amount of robots deployed in a place where the labor cost is so much higher that you could theoretically justify a much higher use case of robots.

And I think a big part of the reason is like China just has so many more engineers that they can put towards the problem of design and deployment. Plus they subsidize a lot of the kind of robotics industry uh to be able to go and make it easy for people to deploy. Yeah.

I think in the US we have this this challenge where uh there's a lot more small and mediumsized businesses and we don't have the historical infrastructure or the engineering capability for all these plants to figure out how to automate on their own.

Uh and so like building the set of tooling and helping them figure out what are the latest things that are happening in kind of AI and machine learning and and computer vision that we can apply to the different use cases in your facility is I think the only way that we can we can move faster than China in the long run.

Is there anything that you want to see from the US government to speed up the deployment of robotics in America? Trillion dollar trillion dollar contract for I mean yeah I mean you could imagine subsid uh subsidiary sub subsidies or incentives or you know a strategic reserve of robots.

Uh but you could also just look at like deregulation but I don't know what your what your take is. Yeah. I mean, so I'm excited to go to the Hill and Valley Forum next week and talk about that a little bit, but I like frankly I I don't think there's much that's necessary, right?

Like I think the the incentive is there, right? There's plenty of financial incentive for these small mediumsiz factories to do it. To be honest, like the biggest barrier to adoption of robotics in America is like it's a cultural change, right?

Like these factory owners, a lot of these are third generation businesses, right? They've been passed down or they're kind of private equity owned. Like in both cases you end up with these businesses that are very riskaverse. Uh they're afraid to adopt new technology. They don't have the internal capabilities to do it.

Uh I think from the government side like things that are being done around educating that that group of of manufacturers and helping them to deploy automation is will have a very big impact like similar to what China does actually in the US.

We've had subsidies for small mediumsiz factories for the last 10 years to buy automation equipment. right there. There's accelerated depreciation. A lot of states have matching grants where if you buy a $500,000 piece of machinery, the state will gr grant up to up to 50% of that cost.

Um, and so like we've had plenty of these incentives in place, but adoption is super low. Like when we talk to the states, most of them say like we can't give out these grants. Like nobody's using it. Uh because people don't know what to buy. They don't know how to buy it.

uh and once they buy it, they're like they can't handle the responsibility of this new piece of equipment. Uh and so there's tons of grant dollars just sitting around uh not getting deployed uh because there's this kind of there's this there's this barrier at the last mile. Makes sense. Very informative. Thank you.

We'll uh we'll see you at Hill and Valley. Yeah, we'll see you next week. Awesome. Look forward to it. Appreciate you guys having me on. Have a great time. Cheers. Bye. And we want to tell you about Poly Market. Go to polyarket. com to get the real news, get the information, uh figure out we got a new market up there.

Uh 1 billion chat GPT users in 2025. That's great. Uh it is currently at a 64% chance. Um and feels likely to me. Uh I the way that