Kristin Myers: Good afternoon from the floor of the NYSE. This is Startup to Stock Exchange. I’m Kristin Myers. Now, every factory in the world has the same problem. A machine breaks, the line stops, and it costs a fortune for every hour that it’s down. And almost always, that machine was making a noise for weeks beforehand. Somebody just had to be standing there to hear it. Well, my guest grew up around that problem. His father was a maintenance technician at a paper mill, and then at the age of 23, he started a company to fix it. Now, just seven years later, it is worth $1.4 billion. First, let’s take a look at what he’s built. The road from an idea to the opening bell starts right now.
When a computer crashes, you reboot it. When a machine on a factory floor breaks, an entire production line can come to a stop, and that downtime can cost companies millions. Igor Marinelli understood that problem long before he became a CEO. His father was a maintenance technician at a paper mill, giving Marinelli a front row seat to the people responsible for keeping industrial machines alive.
And at just 23, he decided there had to be a better way. So he built Tractian. The company puts sensors directly on industrial equipment, listening for things like vibration and ultrasound, and uses AI to figure out what’s wrong, why it’s happening, and what a technician should do before that machine fails.
And seven years later, this is no longer a small experiment. Tractian technology is running across more than 2,000 facilities, with customers including Whirlpool, McKesson and Cummins, and the company says annual recurring revenue has now crossed $100 million. And now the AI boom is creating an even bigger opportunity.
Tractian is using physical AI to turn data from real world machines into predictions, helping factories, data centers and other critical infrastructure fix problems before they become expensive failures. And investors are buying into that vision. Tractian just raised $130 million, valuing the company at $1.4 billion.
But now comes the next test: [unclear: “contraction”] go from monitoring machines to becoming the intelligence layer that keeps the world’s physical infrastructure running. And what has to happen for a $1.4 billion company today to become a $10 billion company tomorrow? From Startup to Stock Exchange, Tractian founder and CEO Igor Marinelli is next.
Kristin Myers: And joining me now is Igor Marinelli, founder and CEO at Tractian. Igor, thank you so much for joining us today.
Igor Marinelli: Kristin, thanks for having me.
Kristin Myers: I think you are actually our first unicorn to come on the show. So thank you so much for being our first unicorn, and congratulations on hitting unicorn status. So we just mentioned a little bit about your background. Your father was a maintenance tech at a paper mill. What sort of problems did you see him deal with?
Igor Marinelli: Well, first, at a personal level, you can imagine the amount of birthdays that he missed, right? All the 3 a.m. alarm calls that he had to take and go back to the plant. Multiple weekends, overtime, double time. That definitely takes a toll on the personal side. On the business side, the impact that unplanned downtime causes is, first, monetary. It is millions of dollars that’s lost. In fact, if you take a look at the size of the problem of unplanned downtime, we’re talking $1.4 trillion. And that is a figure that I think is even underestimated by Siemens.
But then you take a look at the sustainability side. That is more electricity that goes into redundancy. Because if you’re afraid that your systems are going to fail, you’re going to buy more of them and you’re going to run more because you don’t want any collapses and outages of backups. Rightfully so, right? And more water consumption, too. If you think about a data center, which is a hot topic in the AI boom, everybody’s talking about the GPU. But if you zoom out, the electricity is getting to a transformer, which is going to a battery pack, a diesel generator, chillers, which is refrigerating, cooling, evaporators, and all of that is a waste in electricity, in water, to make sure those operations meet 100% of their SLA.
Kristin Myers: So I’m sure your father experienced this, as do most techs. They know the machines so incredibly well. So what did he know that perhaps wasn’t necessarily written down or in any sort of manual, that really influenced and impacted his work inside of the factory?
Igor Marinelli: Actually, one of the things that he told me, and I thought it was kind of unbelievable at first, is that he was using a screwdriver to go to the assets, to the actual equipment, and trying to hear the sound of a gear inside. And I was joking with him that, you know, that must be some sort of Mozart’s ear or something like that. But it turns out it’s kind of true. Technicians nowadays still do it in a more sophisticated way. They go out there with ultrasonic equipment.
What they’re trying to do when they do those manual inspections is ascertain: can I trust this asset to run one more day without any problems? One more week, one more year, right? And there’s the pressure to deliver results, of course. And you know what companies end up doing? They actually will perform maintenance a lot more than they should. Either things are breaking down, and that’s a reality, or you’re buying more gears than you should. You’re changing more of that asset than you should. You’re spending CapEx more than you should. You’re changing HVAC more than you should.
And I’m pretty sure that problem has already impacted you. If you go to this building that we’re in right now, there’s probably a very, very big HVAC system here. If this thing goes down on a hot day in summer, here it is no problem, right? But our customers, they lose goods. Logistically, they can have delays. So the impact is much broader.
Kristin Myers: So we started this seven years ago, when you were 23. Most folks are still trying to figure out what they’re going to do with their lives after college. What did you believe about this market nearly a decade ago that was actually right, making it such a good time for you to start your company?
Igor Marinelli: Yes, I thought when I got into this, and I had a deep desire to solve my father’s problem first and foremost, but I didn’t know it would be that hard, right? There are levels of complexity, but it’s very complex. What I didn’t know is how much the machines, they come from the OEM, which is the original equipment manufacturer.
Kristin Myers: Okay.
Igor Marinelli: How much they come already in sort of bare bones. They have very little automation, very little sensing. And things were not built for maintenance managers to anticipate themselves. And then as I got into it, I was like, oh, that sort of makes sense. Some of the companies that are listed here that manufacture that equipment, it is convenient for them that you have a big aftermarket, that people are buying parts on the schedule from you, that you try to provide a service to go to their plant if something goes down. But they cannot wait 72 hours for a technician to show up, right?
Kristin Myers: Right.
Igor Marinelli: So it turns out to be much more complex. And we turned out to have to manufacture our own hardware to do asset retrofitting. We had to go into the software and eventually, of course, the AI, which is the most fundamental piece to add predictability for that.
Kristin Myers: So you mentioned a second ago about your father using a screwdriver to hear. A lot of technicians now have a lot of tools at their disposal to hear what the machine, the noises that it’s making. So walk me through one machine. What does the technician actually do about it when they hear that particular sound?
Igor Marinelli: Yeah. So the tools that are existing today, apart from friction, which we’re changing, what’s existing today is just a faster horse. Remember, [unclear: “reward”] saying, if you ask people what they wanted, they would say a faster horse, they wouldn’t say a car. So all those tools and those gears, they’re basically faster horses. They’re better screwdrivers to listen and pinpoint, but they still carry all of the complexity of utilizing them. You cannot go to hazardous areas in order to inspect. You don’t have trends over time. It’s not learning your asset.
So what Tractian does is it completely revolutionizes that. And there’s an interesting thing: some of our customers, they’re not only having that revolution when they sign up with Tractian, but sometimes it’s also their first real use of AI altogether. So you compress time, right? The technician, without Tractian, you would have to essentially hire external providers. You would have to have a big team to manage 1,000 equipment. With Tractian, you can have a team of two or three people managing it.
So it’s all about having the observability, and our product, fundamentally, if you distill it down, is information. So maybe a version of Bloomberg for those that are in finance, right? You’re trying to access your future information and what you’re going to do with your asset. And in order to do that, there’s a lot of complexity, but then the product, the final format, is quite simple for the user. What is the remaining useful life of this asset? Can I increase the equipment life cycle? You do load balancing, so you know when to turn off a chiller, when to turn on the other chiller, so you maximize it. You have access to this information.
Kristin Myers: So walk us through this. A client signs up for Tractian. How exactly does it work? What are they receiving on their end? What are they seeing or hearing? Walk us through that process.
Igor Marinelli: Yes. So first it starts with a digital twin, right? We did last year 2,500 customer visits on-site. So we are on the road a lot, because we’re working closely with customers, and we’re building their digital twin, like their virtual factory. So we understand the complete functionality of the equipment and the utilities. When we understand the context, then we start generating information, which we call insights.
So imagine that. Have you ever heard of the music app called Shazam?
Kristin Myers: Yes.
Igor Marinelli: It’s guessing what music is playing. In a certain scenario, our models, our AI, is doing very similar things. Is it lack of lubrication? Is it misalignment? Surcharge? All of those are fingerprints that we do. And we deliver that information for our customer, and he has the ability to enrich it. How will he enrich it? He can upload the data sheet of the equipment. He can upload drawings. What he can do as well is his tribal knowledge. Essentially, there’s always that technician that will look at an asset and say, oh no, this one works this way because of XYZ. So he’s able to, just like you interact with ChatGPT, interact with the system. The system will learn.
And you can imagine a food and beverage manufacturer is very different than a pulp and paper one. They have very different needs. What’s a problem for one can wait for another, right? And so they use that on their desktop. They use that a lot. A lot of our users actually use the mobile app. There’s a huge portion of the usage which is our agent systems directly on their text messages. So it’s the circle in the loop that, for as long as we get to you with timely information, we’re doing our job.
Kristin Myers: Okay. So, Igor, don’t go anywhere. We’re going to take a quick break. More with Igor Marinelli in a moment, including why you cannot sell this over Zoom. Stay with us. Startup to Stock Exchange continues after this.
Welcome back to Startup to Stock Exchange. So why does this matter so much? Well, take a single production line in a large plant. When it stops unexpectedly, the cost runs into the tens of thousands of dollars an hour, and in some industries, far more. And most of that downtime is avoidable. A bearing wears, a motor drifts out of balance. The vibration changes weeks before anything fails. Now, the traditional answer was to maintain a schedule: service every machine every so many hours, whether it needs it or not. Which means you’re either fixing things that aren’t broken or missing the one that actually is. The pitch is simple for sensors and software: stop guessing and start listening to the machine.
Continuing our conversation now with Igor Marinelli, founder and CEO at Tractian. Igor, just before the break, you were talking to us about essentially how the system is listening to the machine and understands what noises mean, misalignment, what one noise might mean. And it’s customizable for every single plant and every single factory. But I’m curious to know, how does a factory or a plant learn to trust this system? Because it sounds like it gets better with time as it learns the machines over time in any single plant or any single factory. So I imagine there might be errors at the beginning which might make someone want to say, forget it, I’m just going to [unclear: “turn it on”]. So how are you getting folks to trust this technology?
Igor Marinelli: Yes. So first, it starts with people. We visit every single customer before we do a sale of the product, and we analyze the risk of their assets, and we guide them on where they could start the journey with us. Ideally, you can start at 100% of coverage, so all your assets are covered by Tractian. But we understand if you want to start with the most critical systems, perhaps there’s a particular type of equipment that you’re more accustomed to and you want to try that. We’re completely flexible there.
Then second, understanding that, okay, I call it the stubborn intelligence. There is a layer of stubborn intelligence. No matter what you tell the AI to do, it will still follow its beliefs, right? It’s almost like when you’re interacting with ChatGPT and you ask ChatGPT to do something that it will not do. There are these barriers as well, and we have safeguards. So, for example, we are never, ever going to miss an outage of any kind. We don’t allow the user to say, hey, this is not a problem for me, if it performs in a certain pattern that is dangerous for the asset.
But after you remove that safeguard, then there is the asset memory portion. The more you interact with it, the more there’s that, like, yes, you’re right. In the beginning you can have a false alarm, and then you say, hey, it was a false alarm. But why? Because I think of XYZ. It’s interactive, and the next time it will reason again and again and over and over. So the assertiveness, right now customers start at 88%. So, yeah, one in every ten alarms they’re going to have to learn with the asset. But then that quickly shifts to over 95% of assertiveness.
Kristin Myers: So as we mentioned, you’ve been doing this for almost a decade now, and you really started it before the current wave of AI. So what has actually changed in the product with this new advent of AI technology that now exists?
Igor Marinelli: This, what we’ve just talked about. Because now the customer, the key user, is the center of attention. Before, with the classical approach to artificial intelligence, just like in medicine, you have an X-ray and the AI would say it’s a tumor X, Y, Z. That was AI back then, and that was how we started in AI. And then users sometimes would not adopt it, because they would say, look, it is just not learning with my reality right now.
Fast forward, still in the medicine comparison here, they can say, oh, because this is a smoker or this is a diabetic person and this and that. You can add all this context. And keep in mind that the average age of an equipment today in America is 24 years old. And then you get into a factory, you have assets that they just bought, you have 24 years ago, and sometimes World War II age, which I’ve seen myself, right? So that context is key. What’s completely changed now is the user can quickly feed the system and it becomes truly your buddy. It is working for you around the clock, 24/7. And it’s a partnership that you do with our AI.
Kristin Myers: So I mentioned at the start, coming back from the break, that this can’t be sold over Zoom. You did 2,500 customer site visits. You mentioned that you are on the road quite a lot. Why can’t this be sold over Zoom? Why is this an in-person product sell?
Igor Marinelli: Yeah. Look, I mean, America is a vast, vast market, right? There is America here, in New York Stock Exchange, and there is America, America, right, which is in, like, this city that you don’t know the name of in the middle of Georgia, in the middle of Alabama. And it is very important that we accelerate the process of giving access to those remote locations, access to technology. They are not adopting it faster because of lack of intention or lack of will. They’re just not adopting it because nobody really went there and helped on how to do it, right? That partnership is key.
So there is the process of removing all complexity for them, because traditionally, if you want to do something like this, you have to call a vendor that calls a distributor, and all their interest is to sell you something. They want to sell you one more sensor. We don’t sell sensors. It’s a service. We are together with our customer, monitoring it for them. And yes, we are efficient because we have way fewer people doing it. But the people who are doing it, they’ve seen it before. For our customer success team, we’ve only hired people that have actual in-plant experience. They had to have at least five years of working in an industrial facility to see what that actually feels like, because they’re going to increase adoption, they’re going to remove complexities for our customers, and they’re going to allow them to make decisions. And that is the only way to make a distribution happen.
And I remember, for investors, this point can be quite unpopular. “Igor, you’re spending so much on traveling to see customers, Delta, United Airlines, going back and forth.” But look, would it be any better if you were spending that digitally with Google ads? Every dollar that you put in seeing your customer is more worthy, right? And it actually proved to be quite efficient. You can say, I’m going to spend $15,000 here to show these ads digitally to a bunch of people, or I’m going to spend $5,000 and take two people out, all travel and expenses covered. And now the customer has seen our product and what we’re capable of.
Kristin Myers: So who’s actually writing the check when you get there? Is it the plant manager? Is it procurement? Is it the CFO of the company? And does that change what they all want?
Igor Marinelli: Yeah, that’s another thing that we also debunked, because traditionally, enterprise sales had to be done exclusively with the C-level at the top. And we’ve realized that the plant managers are the ones feeling the pain, right? And if they are feeling the pain, they also have a discretionary budget to solve their problem. So absolutely, it’s a must that we start a relationship at a local level. My father climbed all the way to be a plant manager in his career, too. And he used to say a plant manager is the highest authority of any individual P&L but the lowest on an org chart, on the corporate org chart. But you are the highest in that local P&L, and you have a lot of autonomy in making those decisions.
Of course, once that starts working, that plant manager is all of a sudden a hero for corporate, and then corporate’s coming to their plant, like, how did you make that happen, and why didn’t you let us know sooner, right? So then there’s that other portion, which is HQ. You sell to the headquarters, but you have to have users that love your product by then.
Kristin Myers: So how long did it take you to get your first Fortune 500 company? Because I imagine you had to start getting some pretty big clients to hit that unicorn status. How long did that take?
Igor Marinelli: Oh, yes. Our whole journey is seven years long at this point. From the day we started until the first customer, that took us two years, in building the technology and making sure everything was right. But from that first customer on, it has been quite a slingshot. And so I think it was three years ago we got our first Fortune 500. And now we have names, again, that are listed here, that are very important for the economic development of the country, and we partnered up with them in all their operations.
Kristin Myers: So you have $100 million in annual recurring revenue. That’s where you’re at now, $1.4 billion in valuation. What do you think it’s going to take to get to $10 billion?
Igor Marinelli: Yes. I mean, every new investment round comes with resetting expectations, right? So I think the moment that I was probably the happiest in my life was when I received that first angel check. That felt amazing. The bank account was 125.
Kristin Myers: Okay.
Igor Marinelli: And that felt amazing to the bank account. That was infinite money, right? But as you know, the stakes keep getting bigger and bigger, and so is the expectation as well. An investor that will chip in right now is expecting that he or she is going to write a much bigger check. We will have to 10x the value of the company, and there’s no secret. We do that by getting out of here, getting back to work, and delivering to our customers.
Kristin Myers: Last question that I’m going to ask you. You work with so many companies that are listed here at the New York Stock Exchange. You are sitting with me now on the floor of the New York Stock Exchange. The bell is right over there. Have you thought about going public, and when do you think that might happen?
Igor Marinelli: Yes. Look, definitely, right? This is very iconic. We are still early in our trajectory. I think we have to hit certain milestones first, but definitely. The New York Stock Exchange would be a dream for the company, because this is where our customers are. This is where our partners are. This is a trusted environment of people doing business together. So definitely, it’s on our radar.
Kristin Myers: All right. Thank you so much, Igor Marinelli, founder and CEO at Tractian. Thank you so much for joining us. And don’t go away. Much more from the New York Stock Exchange right after this.
Welcome back to Startup to Stock Exchange. Seven years from nothing, Igor Marinelli started Tractian at the age of 23, and today more than 2,000 factories are running on it. That’s 700 employees, offices in Atlanta, Monterrey, São Paulo and Poland, and $100 million in annual recurring revenue, more than $300 million raised at a valuation of $1.4 billion. Unicorn status, and the customer list reads like an industrial index: pharmaceutical companies, appliances, bottling, engines, aircraft, none of which existed in 2018, when the pitch was a 23 year old telling plant managers that his sensors could hear what their technicians [unclear: “put in”].
That’ll do it for this episode of Startup to Stock Exchange. Tomorrow, we’ll be interviewing Alex Hauck, CEO of Intake Breathing, the company using magnets to revolutionize how the world breathes. We’ll be back at 12:30 p.m. live from the New York Stock Exchange. I’ll see you then.
