In Podcast M4Edge, Xometry CEO Randy Altschuler Talks Shop

Marco: Hi, I am Marco Annunziata and welcome to M4Edge. Michael, what's the long name again? 

Michael: The short name is M4Edge and the long name is is Macro Micro Michael Marco and Startups at the Edge. I'm Michael Leifman and today our guest is Randy Altschuler of Xometry. In case you're wondering, that's Xometry with an X.

Marco: Xometry combines two business models that have come to define our new industrial age. One of them is all the rage lately: AAS, or as a service, as in SAAS, or software as a service. In Xometry's case, MAAS, manufacturing as a service.

Michael: XAAS, anything as a service, is the generic term. The idea is that instead of buying a good, you purchase the services the good delivers when you need them. Some examples have been around for a long time, like renting a car, but the trend is now becoming widespread. Xometry allows companies to leverage existing manufacturing capacity instead of getting new parts to get the parts they need rather than having to invest in additional manufacturing equipment of their own. But other than in rare cases, Xometry is not the manufacturer, but is a match maker between customers and suppliers.

Marco: And here is where the second business model comes in: the platform. Xometry provides a platform; the means to link people or companies that want something produced with companies that are able to produce it. It's like a hyper-efficient market—an economist's dream. It is not too dissimilar from what Amazon does, linking buyers and sellers, or Uber and Lyft, linking riders and drivers. But there's more. 

Michael: Hopefully there's more because if you guys only dream of hyper-efficient markets, then you're not looking for life. (Laughs) Part of Xometry's secret sauce is in its machine learning algorithms. That is, Xometry is not Ebay for manufacturing with bids and offers. Xometry figures out what the thing should cost. Its data and algorithms come up with the prices. So, what does it mean for the economy? Lots.

Marco: Probably a lot more than I can dream of. Xometry enables greater production through better utilization of the existing manufacturing capacity. And by connecting large numbers of users, it can yield greater efficiency and lower prices. 

Michael: And by increasing capacity utilization on production lines, we are maximizing the overall productivity of capital and labor. But these business models are also changing something else fundamental: the geography of production. The platform is a service model that theoretically knows no borders. So, if you only had access or thought you only had access to a local, narrow geographic area, or were only aware of certain producers, suddenly you have access to a broad network and supply chain, or a broad customer market.

Marco: This is one of those things that has the potential for exponential change in the economy, and major players are noticeable. In the days since we recorded the interview, BMW announced that they've designated Xometry as a Tier 1 supplier.

Michael: Oh yeah, and did we mention they do 3D printing as well? There's a lot here. Xometry is definitely at the edge. Welcome to M4Edge and enjoy the episode.

Marco: Absolutely. Randy, thank you so much for joining us on the podcast. We are very excited to talk to you. Both Michael and I have a background in manufacturing so we are very excited about what you are doing, but why don't you start off by telling us and our listeners what Xometry does, how you came up with the idea, and what is the purpose or what problem are you trying to solve? Or as I normally put it, why on earth are you doing this?

Randy: Yes, and thank you so much for having me on the podcast. I'm very excited to be here. We have this unique opportunity in the United States and around the world where hundreds of billions of dollars of manufacturing are being done by small manufacturers, and they're spread out across our country here and across the world. For these manufacturers, the long tail of the internet hasn't quite reached them. So, we've described them as being landlocked. They're sometimes second, third generation businesses. They have a dozen to 20 customers they've had for a long time, all in a local area, and that is the heart of their business. When those customers are busy, they're very busy. But when those customers aren't busy, they aren't busy either. They have a lot of free capacity. These companies have a small amount of marketing and sales efforts. Maybe they have a website, but often they have almost no online presence. And their capability—their skill—is largely hidden from all of the other potential customers around the country or around the world. We describe it as having "stranded capacity." And the internet, which has allowed so many small retailers to access customers everywhere, doesn't touch these manufacturers. On the flip side, if you're a customer, there's no central best place for you to find a manufacturer to make your parts. You too have your own supply chain, but it's limited to your own experiences. As new manufacturing technologies get introduced—it could be 3D printing, a new kind of molding, and all sorts of innovations happening—you don't really know, "Is that the best way to make my parts?" What are the decisions, what are the tradeoffs of these different technologies because you have your usual set of suppliers who have their usual set of skills. You don't really know in an automated, digital way how to access these new technologies you're reading all about. So in some ways, business is very analog. So with Xometry, we're trying to bring it into the new age and create a platform that connects buyers and sellers of manufacturing. And our proprietary technology helps create pricing and an entire ecosystem that enables those buyers and sellers to interact and create custom parts. 

Michael: So can you talk more about what a typical use case would look like? How does the process work? Maybe if there's a customer whose story you feel comfortable telling, take us through the idea of someone reaching out through you to a supplier on the other end.

Randy: Sure, so a good example would be BMW. BMW will come to us to make parts in their production line, parts that are used as service parts in a car. These are typically small-volume parts that have quick turn times. They don't hold a lot of inventory of these parts. So they'll come to our platform, they'll upload 3D CAD files— 

Michael: 3D CAD file? That's three dimensional computer-aided design and drafting file. It's the software encoding of the instructions for how to produce something: what the geometry of the part is, what the materials should be, the processes, the dimensions, the tolerances, et cetera. It's the computer engineering drawing for how a part should be built.

Randy: And our technology, which we use artificial intelligence, will produce a price and a lead time for them. And we'll also tell them and show them, in real time, the tradeoffs of different manufacturing technologies because it turns out there are a lot of different ways to make a certain part. Some may be cheaper, some may be faster, some may give you tighter tolerances so it's more exact. Some may have different types of surface finishes, some may have certain material properties. With our platform, they can instantly see all the pros and cons of these different technologies in real time. So BMW can come to our site and upload these files, choose that technology they want, and see the delivery date. There's also what's called post-processing—things they need to make the part usable in a production environment or end-use environment—and then they can buy. You can either purchase with a credit card or if you're a company, you can pay with a purchase order. You never have to talk to anyone on the phone, you don't have to go visit the factory, you don't have to think about how this guy can do that and this guy can do this; it's not a mix and match. No, it's all there for you in one place, like when you want to shop for things on Amazon, anything you can think of is there and you can press buy. It's not hypothetical, it's real. It's a real window into the pricing and lead times and capacity that's available in manufacturing today in the United States. Boom, you buy, and then with our algorithms, we match that order with the best suppliers available. And we use these matching algorithms to pick the right supplier and there's lot of characteristics that factor into that. And then we create a price for them and we marry those two prices. Xometry is the person who is connecting the buyers and sellers, and the transaction is made, and then we monitor the production of those parts. (12:28)

Michael: I'm going to tell a little story that I told you, Randy, when we first spoke on the phone. Going back before XAAS, or everything as a service, was anywhere in the common parlance and long before Xometry was started, I read a New Yorker article in 2004, 2005 about a company called Office Tiger. I was obsessed with this story—it was so cool—it was about a company that basically invented the idea of business process outsourcing. It was something I was really struck by. Fast forward to 2006 when you and I first spoke and I did a little bit of googling around—who was this Randy Altschuler guy? All of a sudden this article comes up from 2004 and lo and behold, you're that guy. It was sort of a cool moment for me and I'm wondering, Office Tiger is similar to Xometry, right? You're linking buyers and consumers through a platform in a way that hadn't really been done before. And so, how much of that background of Office Tiger played into your view of how or why you built Xometry? 

Randy: You know, the concept behind connecting buyers and finding the optimal solutions for customers and, in the case of Office Tiger, it was about skills, it was about location, it was scale. So it was a different set of considerations but it was still the same theme of, how do I make things a better way? In this case, it was a more of a service model, but it was, "how do I provide these services in a better way?" The mission of Xometry at the end of the day is for both the buyer and seller to find the optimal match. So for the buyer, "What's the best way to make this?" And not only "What's the best way?" but "Who's the best one to make it for me?" And likewise for the sellers, a big challenge for manufacturers is they're being asked to quote and do work that doesn't fit into their sweet spot, because the customers aren't educated enough. So if you're a manufacturer, you like to cut metal. You actually like to manufacture it. You don't want to waste time quoting parts or on administrative stuff, so if we can knock that out for you, that's really great. And a similar concept with Office Tiger is the sense of, and in this case, with removing any boundary of nationality and saying, "Hey, we're going to find for you anywhere in the world, whether it's in India, whether it's in Salt Lake City, whether it's in Poland, we're going to find that optimal solution for you." I think that's a common theme—that we're becoming more and more of a globally linked country and economy. I think that theme resonates with Xometry and there's a lot of efficiency there. There's a lot of inefficiency because traditionally, you're not finding the best matches between the buyer and seller. 

(15:00)

Marco: And Randy, you were saying earlier that the key to building credibility is to show you can get it done, especially since you're not doing it yourself. So how does the vetting process for the manufacturing companies work? How do you find them and how do you make sure that they have the capacity and the quality? 

Randy: What's exciting in terms of finding them is that they've found us. We have a lot of people on staff who come from a manufacturing background and, you know, there are facebook groups of machinists—they're all over and it's a whole community. They love to watch videos of people cutting parts and actually machining. I mean, there are tens of thousands of them who do that. 

Michael: Sounds fun.

Randy: Yeah, my son watches other people play video games on his iPad. He watches other people do it. So since it's been coming to us by word of mouth, the vetting process is really important, particularly because a large portion of our customers are aerospace, defense, automotive, industrial, and medical devices. I mean, really important stuff that's in the sky, on the ground, it's in your body. So first here is the online vetting portion, there's the interview portion, then there's a test, which Xometry pays for. So we'll give you test parts and we'll pay the supplier to do that. And then there's the portion of time where all the work will, before it's sent to the end customer, it'll come back to the Xometry facility and it will be tested and be approved before it goes to the end user. So that's an extra cost, but we don't feel comfortable until the partner—we call our manufacturers, "partners"—until they reach a certain point of credibility—they have a certain score we generate—we don't feel comfortable that they can direct ship. And then we also tightly control how much work that partner can take and for what kinds of work. It's all done in an automated way using an algorithm since we do too much volume to do that sort of thing by hand to get lost. But there are a strict set of rules that dictate who can do what when and as you graduate that process—sometimes they never graduate—but if they graduate through that process we're more comfortable with saying, "Hey, it can go directly from you to the customer." Or, "You're good enough to do this very precise part or this very quick-turn part"—you know, all these different gates they need to go through.

(17:19)

Michael: I recall—I don't know if it's still part of the Xometry business model—but I recall you guys also had your own manufacturing facility in case you couldn't match a supplier with a consumer's product demands. Is that still a part of how you do things?

Randy: Yeah, so we consider it more of our lab. When we started our business, to get the data going—we needed to capture some data—and we need to, as I was alluding to before, have some manufacturing experts. We built our own little machine shop, we built our own 3D printing facility, and we learned a lot from that. But that also makes us more credible with our suppliers. We can say, "Hey, I have actual machinists here. I know what you're going through." So we still have that facility. It's an extremely small part of our network and it's less about backup more than about credibility, testing, learning, and eating your own dog food. If we're going to be rolling out a new feature to our suppliers, would we, as a supplier, want this feature? Now if our guys hate it, why would we expect our suppliers to like it? So that helps us a lot. And I think there are other folks who aredoing different elements of the space, and I think having the manufacturing chops is important. I think it helps influence your product development but it's also important for the credibility you get with your suppliers and your customers. 

Michael: Are there other things you think distinguish Xometry from the other players in the space? I've noticed you're no longer alone.

Randy: There are a couple of things from the customer side. We want to be your one-stop shop. And then on the supplier's side, it's about—this I can't really talk about yet, because we haven't introduced it—it's about things we can do to make our suppliers better. So part of our pitch is to go to BMW and say, "Listen, you may even have one of our partners on your network, and they may already be a supplier to you, but working with Xometry, they're better. If you access them through our platform, they're going to do better for you than if they did it on their own. I can't detail exactly how we're going to do that, but that's part of our mission to make our suppliers better suppliers.

Michael: Quality-wise or price-wise? Can you divulge any of that?

Randy: We want to make them more profitable, we want to make them faster, and we want to make them more accurate. We want everything that's good for them and the customer. This is a rare marketplace where both sides can win. Often, in a marketplace, you're pushing down a supplier's price to satisfy the buyer. In this case, there's so much inefficiency in the market, and there are so many tools we can bring to bear, we can actually make it cheaper for the customer and more profitable for the supplier. 

Marco: Let's stay on the concept of this being a win-win situation for both manufacturers and the customers, which I like a lot. You started out the conversation by pointing out that, from the point of the manufacturers, they have spare capacity that they can better utilize through Xometry. Talk a little more about the users and why they choose Xometry as a platform. Is it mostly for prototypes, is it to complement their supply chain, do they use it as an additional form of their supply chain? I'm interested because especially when you see companies like BMW or GE, which have large-scale supply chains, you wonder what is it that Xometry can offer them?

Randy: Yeah, so we're not a contractor manufacturer. When you look at a Fortune 500 or a Fortune 50 company—and we have a lot of those in our customer base—they're going to be doing contract manufacturing. They're going to build a one-to-one relationship that their integration into their supply chain is going to be seamless. We're not going to disrupt that market that's very efficient. We don't want to add noise to that. We don't want to add friction to something that's become as frictionless as possible. There's always innovation but it's pretty darn good, and there's lot of tools out there, lots of software tools and tried and true methods to make that better. But there's a huge subset in the United States of the 80 billion dollar segment of parts that are ordered in low volume that are not end-use production parts. And it's that market that's incredibly inefficient where the procurement person at BMW sends out an email, even if it's an automated email, to 35 suppliers and says, "Hey, here's the specs of what I want, come back to me before XYZ date with your response." They gotta call people, follow-up, answer questions. They gotta do all this work, and after a bunch of days and a bunch of costs, they'll finally have some bids. We are effectively an instant RFQ with our system. You don't need to do all that. Our algorithms are instantly telling you what the best market price is today, and if Xometry is successful, our best market price is better than the best price you could ever get even if you do get that 30-person RFQ. And we do that in a given day. That's really important to the customer. And for the procurement professional, who's our friend, their ability to scale and their ability to get better lead times and solutions without doing all the manual leg work that's required today, even if they have a fancy ERP and procurement system—

Michael: That's an Enterprise Resource Planning System. It's a system that allows manufacturers to track in one place all of the prices, bids, quotes—everything in their manufacturing supply chain—all the material, all the labor, all in one place.

(23:00)

Randy: —but that's really not germane in these instances—that's a win for them. 

Michael: I want to go back to this. I've been sort of itching to ask this since the beginning when you said you have this AI algorithm that comes up with the price. So it's not actually a bidding system, right, it's not as if you've got manufacturers on one side, saying, I can do it for X and another guy saying I'm going to do X minus one, or whatever. It's the algorithm itself doing the bidding. 

Randy: It couldn't be instant if that were the case. So I don't have time to pull my network and say, Hey, Michael just quoted these parts. What do you guys think?" My customer will be waiting around. So I need to find a way to give you an instant price. My algorithms basically are a instant poll and say, "Hey, I know what suppliers have done these parts, kinds of parts like this before for what price, I know who likes to do it at this price. I know what's the right price. You know what I can get this done at, what the lead time will look like, etc." That's the goal I'm going to return to the customer. And that's good of the magic of Xometry because using AI, we've gone very wide if you actually tried to do a cost plus model; then you would never be able to scale the size which AI could do so quickly. It would be extremely slow—we're trying to figure out tool paths and exactly how it was made and, on the flip side, if you wanted to actually get real bidding data, it wouldn't be instant. So the beauty of AI is to mimic that bidding process in real time. 

Michael: How did you get the data to train the model to begin with, since so little of that building data is ever disclosed? Very little of that as public. How did you how did you train the model? 

Randy: We just built and are building the data. It's just running parts, getting data—we acquired a company called MakeTime, one of our competitors based in Lexington, Kentucky a couple months ago and so they accumulated a lot of data. So it's just through different means. But it's all data that we or MakeTime have accumulated ourselves. 

Michael: I assume it continually trains when a guy accepts a bid or whatever. I assume that the model reinforces its knowledge?

Randy: It does. So we need more—the more data we have, the more we train our models and the smarter they get. So, a lot of it is about scale and and the number of transactions we can run. Those are all incredibly important data points for us. Think about the more people are on Google, the better and the more accurate the search results become, right?

(24:30)

Michael: So, sorry to harp on this a little bit, but if you've got either an unusual product or a new product, what's the track record on how accurate you are based on the expectations of the consumer?

Randy: Oh, we'll eat it. We have a portfolio approach to our pricing. Sometimes, to your point, Michael, we'll see something we've never seen before—the system won't have seen it and it will misidentify or won't really know how to handle it. It will spit out a price and a lead time. Now, if the price and lead time, unfortunately, are too expensive and too long, the customer won't order so we don't win their business, which I always like. But that's better than what usually happens, which is it's too good and it underestimates what the true cost is and/or maybe it says it's going to happen faster and that's when we have to eat it. We say, "Dave, you come to the site and you order something that's $1,000." Well, actually, our software hasn't seen this before and it's misclassified it. It's really a $4,000 part. I mean, I'm eating $3,000, you know? Big tasty sandwich. Not tasty, but—

Michael: [laughs] Right, right. So you take all the downside risk, but because if the price is too high, it doesn't get bought. You know, you get the upside risk.

Randy: That's right. So you see the margin on what we do. It's focused in a nice healthy place but there are tails on both sides. Both, some parts where we get really oversized margins, but there's a downside as well, where, hey, we don't have enough data. We haven't learned about this yet, we've got it wrong and it misidentified it. As time goes on, and we do more and more of those parts, and as our system gets smarter and smarter, we're truncating the tails of both sides and it's about getting narrow, narrow books. In the end, we really want to have predictable range for everybody, both their buyers and their sellers, but that takes time. And that's our moat too though. That's why it would be very hard for another company. We talked about competitors, but they don't really use AI to come in and suddenly be able to do that, because you have to have a lot of data and we've had five years plus we've had five years of MakeTime—MakeTime has been around a similar period of time as we have. We get the advantage of all that data we've generated together.

Marco: You should be aware of that whenever we mentioned artificial intelligence, Michael gets excited and suspicious. He has these dystopian nightmares bubbling up his mind. So if we can steer the conversation away from the algorithm for a second, let's talk more about the products or any you were mentioning earlier: aviation, healthcare, defense. Can you tell us a little bit more? Is it mostly intermediate parts, is it mostly industrial products, is it end-user products, or any sector-specific, sector-stronger users than the part of the platform?

Randy: Yeah. So it really ranges and is full spectrum from a quick prototype to an end-use product that a consumer will see. So we're more on on the B2B side. So I'd say that's what distinguishes us. That's why we tend to do higher complexity work because we do a lot in aerospace, defense, automotive. We do more stuff, from assemblies to all post-processing. There is just a lot more work in our products and a lot of our tools are helpful for a professional.

Michael: So let's begin to broaden this a little bit. We just touched on some of the sectors that you produce for. Geographically, you guys are only in the United States only at this point, correct?

(29:00)

Michael: Do you see a future where Xometry is global—where you're able to tap into a more global supplier and consumer base?

Randy: Absolutely. You know, many of our customers are multinationals and are asking us to be in Europe or asking us to be in Asia. And so, absolutely. We need to be where our customers want us to be. For us, the challenge will be building those local networks. And that's important, not only because of the transportation costs and lead time costs, but also because every market's a little bit different. The precision expectations in Asia, in Japan, may be very different than they are in Canada—just for example—but those markets can be somewhat different. And those subtleties are important as we grow global—and eventually Xometry will go global—our platform has to recognize that. It's also a reflection of the kinds of businesses in each of those areas as well. 

Marco: So if I hear you correctly, Randy, you're envisioning going global with a kind of an islands approach that is building local networks of manufacturers to solve the national, regional customers as opposed to having a deeply interconnected global chain. 

(30:22)

Randy: I think it's fine to have a global chain. And I think as we go global, we'd be happy if somebody in America wanted access to a provider in Asia or likewise for an agent to access a provider in the United States. I think there's so much local capacity available and skill available that—and particularly those we're not doing contract manufacturing—I think a lot of that value is much closer than you think it is. And there are advantages to being closer. 

Michael: When you and I had talked earlier regarding some applications of this sort of business model to developing countries, you brought up the important point that you need, for this sort of dispersed business model and production model, you need reliable power, you need roads, you need ports. It's not that you can just build. You can't plop down a factory in the middle of a forest somewhere. So, talk a little bit about that if you still, presumably, feel that way and how you think that limits you or what you think the opportunities are to either build out countries that don't have that infrastructure but have cheap labor costs. How do you see that developing?

Randy: Yeah, I can use an example right here in Maryland. In Maryland, a lot of the power lines are above the ground and, I'm from New York City originally, so, different infrastructure there. So when there are bad storms, our manufacturers lose power. As you mentioned in the beginning, we have our own little manufacturing shop where we'll actually lose power, and that's here in Maryland, in, by the way, Montgomery County, one of the wealthiest counties in the United States. So I was recently speaking at a panel about manufacturing in Maryland and I know the state's pushing hard to get manufacturers to come here and I said, "First order of business, make sure everybody has power." And particularly, when you're thinking about things like 3D printing, getting a power backup is ridiculously expensive. You can't do that. And if you lose power for seconds you can lose hours and hours of build—of manufacturing time on your 3D printer. (Snaps) It's gone. So uninterrupted power supply is critical. Even here in Maryland, it doesn't happen and it's unfortunate. 

(32:50)

Michael: Kind of following the same theme, maybe less hypothetically, but from your current crop of customers, is the pole of extra production what gets the factory that has the slack online or is it the cheaper labor costs in one county or state versus another—what's really at the fulcrum here? Is it the capital costs, is it just the availability of the production line, or is it the labor costs?

Randy: For our customers, it's about getting the parts. They need those parts, and because most of our customers are established businesses, as we alluded to earlier, they start with us because their supply chain failed. Their go-to vendors couldn't get it done, and that's how they first find us. "I can't get it done." They find us on the web. "Okay, let me try this solution." So it starts with a problem, and when we can prove to them that we can get that done, that's when they start coming back to us over and over again and at some point they say, "Wait a minute, maybe it's not just for stuff for my own network projects, maybe it's for more core stuff." For our customers, price is always important and lead time is always important but reliability is certainly the single most important thing, and getting it when you need it. And that's the beauty of having a marketplace versus just doing all manufacturing yourself. You can scale in a way that nobody else can if they're trying to build it all themselves. So with Xometry, you have virtual access to 2,500 manufacturers. That's pretty awesome. Now, the question is how to really deliver on that and how do you make that accessible? That's what Xometry really tries to do but if you can do that, then that's a very appetizing solution for a customer.

(34:30)

Marco: And there is something you mentioned earlier, which I found very, very interesting and very encouraging. When I first found out that your supply chain, your manufacturing base is entirely the US, as an economist I had an immediate negative reaction. I thought, "Oh my god, this is something that will actually encourage more protectionism. It will encourage the breaking down of global supply chains." But our conversation tells me this is not true. You were pointing out you're not contract manufacturers so global manufacturers still need their global supply chains, but at the same time, something you mentioned earlier is that you're creating more demand, more work for manufacturers dispersed across the United States. And therefore, creating more certainty, more demand. And so, possibly more production, more jobs, and more investment. Should we see Xometry as something that can actually strengthen the fabric of manufacturing and job creation across the US?

Marco: One related question is the issue of what happens to capacity utilization. The first point you made is you're creating a system whereby the installed capacity of manufacturers in your supply chain gets greater use. Low capacity utilization is inefficient, but it also provides a cushion to the extent that as capacity utilization increases, there is more the risk that if something goes wrong, there's a spike in demand or a disruption due to a hurricane, you might have more significant problems in the manufacturing chain. How you see this? How you see the impact?

Randy: I don't think that's necessarily the case. In fact, I think it's the opposite. As our suppliers are seeing more business and we're effectively there as SG&A, their sales and marketing engine, they're beginning to buy more equipment; they're beginning to add more capacity. I have to give my manufacturers lot of credit. As I mentioned in the beginning, some of these are second or third or even fourth generation manufacturers. They know you can't run your machines to 100 percent capacity or even 90 percent. I mean there's that comfortable number to take into account. The inevitable problems that occur just because things happen and you can't control Mother Nature. So what I think those manufacturers realize is they're getting more business. "Boy, I probably need to invest more." And we give them the confidence that they can do that, because even if they're in Texas and they've lived and died with oil and gas industry their whole lives, and now they say, "Hey, wait a minute, I can get a whole new stream of business for medical device people in Boston or from a defense contractor in Maryland because Xometry's delivering it. I can go out and buy more equipment."

(38:10)

Marco: So this enforces the point you made earlier that it is a win-win situation. So on the one hand, the users have more confidence and more certainty that they can get their parts when they need them with the right quality, and on the other hand, the manufacturers have now more confidence and more predictability in a higher level of demand and this can, therefore, give them the confidence to invest more and create more capacity.

Randy: That's right. And one of the other benefits we have in our network—so recently we had all this flooding and terrible weather in North Carolina—we were able to take the work that was being done by our North Carolina suppliers and switch it to people in other geographies. From customers' perspective, that was seamless. They didn't see any problems. And from our local partners' perspectives, the burden of having to do work when your shop was underwater, and where it was going to be risking your life to go on the road, but you need to do that, was taken away as well. So it was really a benefit as much as anything can be a benefit when there's natural disaster. But it was a benefit for both the customer and the supplier. With a network where you can distribute work to different places, that enables that to happen. Otherwise you're stuck with that one provider and that's a bad thing for both of you when things go bad. 

Michael: One of the questions I want to make sure I get in is sort of what I'm going to call the the philosophical question of this podcast. When you look at the manufacturing market 15 years from now, 20 years from now—some point that, for a startup, is a very, very far future—how do you see the manufacturing world? You think production as a manufacturing service becomes the dominant form of production? You think it's still sort of niche-y thing? Where do you see this going? How do you see the networks being built out?

(39:54)

Randy: I think it's going to become more and more interconnected. So I think networks are going to be more and more important, and they may be in the shape of what Xometry has, they evolve—I think they always have to evolve—but I think technology is linking that long tail. The internet is reaching manufacturing now and it will forever change and it will—I think in a very positive way—will interconnect us. I also think there's going to be, unfortunately, because of climate change, I think there are going to be some other ships. I think manufacturing's going to have to be more aware of its impact on the environment. I think that's going to dictate what methods of manufacturing technologies themselves start emerging and which ones get throttled back. I think it's going to be different world altogether because of those sorts of trends and manufacturing is going to be front and center of that.

Michael: Can you foresee a moment when Xometry works into its algorithms, the sustainability of the process?  What the emissions are?

Randy: Absolutely. Listen, there is no way that having a ten dollar package shipped overnight via Amazon from somebody far away is good for the environment. There's just no way. And again, our job is to empower and grow all these local businesses, but I think certainly location, carbon footprint, and the impact have to be factored in. It's a true cost. And we've all been in such a hurry. We haven't factored that but it's coming home now. We've got to deal with it.

Michael: This has been great. Randy. Thank you very much. 

Marco: Thank you so much. Good luck.

Randy: Okay. Thank you guys. Good luck.