Bram de Zwart (3D Hubs): "Wij willen een miljardenbedrijf worden"

A lot has happened. We changed the company's strategy a bit, but the vision remains the same. We believe that automated production technologies like 3D printing make it possible to produce products much more locally, much closer to the end user. This reduces transport, you only make products when they are needed, there is little inventory, and the world becomes more efficient with fewer resources. But the strategy has changed: we used to focus on consumers and hobbyists, but now we focus entirely on industry. We are adding more automated production technologies alongside 3D printing, like CNC milling and injection molding. So quite a bit has changed. 3D printing is still an important technology. Let me explain: at the beginning, I remember very well from three years ago that the starting point was that there are many designers who are very good with 3D design technology, but not everyone can afford such a printer. So we became the platform between the designer and the people who need those parts. How does that work in a B2B environment? That is the peak we are moving towards. Do people still have their own printers? Not always. What is your role now? First of all, they can have 3D printers. Our best customers are companies that also have 3D printers in-house, but they need a different type of 3D printing. 3D printing is actually a name for six different production technologies, each with its own characteristics. So sometimes they need a different material or technology, sometimes they have too little capacity. Sometimes it's a very large company where one department doesn't even know what another department can do. A company might still prefer to outsource 3D printing. And importantly, we no longer only offer 3D printing, but also milling and injection molding. It's often about much larger orders, much larger quantities, and much more serious machines that many companies don't just have standing around.

What we do is we have hundreds of production partners connected on four continents. These are certified companies. In this way, we can ensure that production takes place as locally as possible and that capacity is immediately available, so the customer doesn't have to wait until a producer has time for their order. Additionally, we have developed a platform that automates all the processes that take place beforehand, such as price calculation. Traditionally, an engineer would look at how much machine time, personnel time, and material is needed, and do complicated calculations that could take days to weeks before the customer gets a quote. We use machine learning to fully automate this, and in 98% of cases, within seconds, the customer gets a price for production. And we guarantee that price. It's one of the many processes that are traditionally very complicated, very manual, and take long hours, which we can now do with machine learning within seconds. We only started applying this in the past few years, about two or three years ago. We started with standard algorithms, gathered a lot of data, and then attracted machine learning specialists.

Yes, with the help of technology. It works with neural networks that do pattern recognition. They extract all the parameters from the customer's designs and CAM files, and input them into the networks together with the prices that suppliers have given and all the production specifications. Because we have so much data—we have produced millions of parts—we can now guarantee the price in 90% of cases. Sometimes we still make a mistake, but we learn from that too. During training, it sometimes goes wrong if we don't have enough data for that specific geometry and production specifications.

That's correct. If you take a step back, the big development is that many more technologies can be used to produce things automatically in many more ways than just 3D printing. The technologies we now offer alongside 3D printing have existed longer than 3D printing, are much larger markets, and are much more proven. But what you see is that because there was so much hype around 3D printing, a lot of software development was done there. Those other markets are huge, and there are very few software companies like us that go in and say, 'Hey, we can create a marketplace and automate all those processes, lowering costs and delivering parts faster.' So 3D printing is an important technology and very complementary to milling. You use 3D printing for smaller production series or when you need a lot of design freedom. But when you go to larger quantities or very high precision, you quickly move to milling or injection molding, which also work on the basis of CAD files. So it's essentially the same basis. We are still busy decentralizing production in the world, making it happen locally, with less inventory, and through a large automated platform.

I think that could still happen, but that development, like every hype, follows the hype cycle. It just takes longer than many people expect. I don't rule it out, but we are much more aware of the proven value of 3D printing in industry than with consumers.

The most important thing is that we were perhaps looking too narrowly at the 3D printing market. We should have focused on industry and more production technologies that we could automate sooner. That has been our main lesson. But luckily we did it in time.

We have a board of directors, which includes investors. That provides a good reflection moment every three months to review the strategy and look at the business from a distance. That's where it came up: 'Hey, maybe that market isn't the one where we can really make a huge impact and decentralize production in the world.'

It's a combination of making a financial forecast and thinking, 'We think we can grow revenue this fast in the coming years, and we need to invest about that much to achieve it.' You also look at what growth is in the market and what a reasonable valuation for the company is, so you can raise that amount without excessively diluting the shareholders. So it's a mix of the financial forecast and what is appropriate for the phase our company is in.

We will continue to invest in the platform. We now have 70 people. Airbnb has about a thousand, and the platform works. The production industry is larger than the accommodation industry, so it wouldn't be crazy to invest as much as Airbnb in development and automation. We want to add more production technologies, integrate more into the workflows of both our customers and production partners. There's an endless list of ideas.

You often find them through your network. They are parties where people I trust have worked and had good experiences. Before a production partner can send an order directly to the customer, they must first send three sample orders to our office, where we inspect the parts for quality. So we physically sit between them for the first orders. We also set the production standards very precisely in an agreement that we sign with them. We discuss these standards with the production partners, and they agree to adhere to them. This way, customers can trust that they will always get the same consistent quality. Additionally, we have a team of five people in China who visit production partners and connect new ones. That takes a lot of time and money. In the old model, where we were still working with hobbyists and 3D printing, we had 30,000 3D printer owners connected worldwide. Now we have reduced that to a few hundred, which allows us to control quality much more strictly from the client's perspective.

It's cheaper and faster. Because we automate many processes that were previously done manually, like price determination and checking if the design is producible, we save time for both sides of the marketplace. That saves time and costs. Additionally, we have built a smart routing system that ensures an order can almost always start production immediately at a production partner. You shouldn't underestimate that. If you, as a customer, search for production partners online, you might have to wait a week or two before you know the price and whether your design is producible. And when you place an order, you have to hope they can start immediately. That's not the case with us because we have a large network and therefore capacity. We can guarantee very fast delivery times and very low costs because we know we have a whole production network in the background that can deliver.

Sheet metal working is on the list. Laser cutting of sheet material. There is a lot of demand for that. In principle, we can add any technology that can be controlled on the basis of a CAD file and where we can predict the price, delivery time, and producibility.

We could be profitable now. We make healthy margins on the orders we handle. But we choose to invest in growth. Why not? Some people say, 'Be satisfied with 30 people, a nice team, and a fantastic margin.' But I want to build a huge company and have an impact on the manufacturing industry worldwide. We want to ensure that production happens locally and with less inventory. You can't do that with a company of 30 people. We need to build something bigger.

Those are two questions at once. We are now scaling a company from 70 people to many more. You hear that this comes with its own challenges, like ensuring everyone always knows exactly what to work on and staying motivated. There are challenges there. As for my strength, the most important role of a CEO is to ensure the company is successful. I think I can adapt well to what the company needs at each phase. I have been good at gathering the right people around me who know things much better than I do. Ultimately, the CEO must ensure the company is successful, regardless of what he does or doesn't do.

I think I could have executed the strategy shift a bit faster. I had already signaled early on that it was necessary, but I could have taken the step sooner to say, 'Okay, we're going to take a big risk, partially leave behind something we became very big with, and move into the new direction.' In hindsight, I could have been a bit faster. But it was necessary. We wouldn't have survived in the old model. We could have built a small, profitable business, but we had already raised twelve million in financing, and the market we were in wasn't big enough to build a billion-dollar company. And that's what we're going for.

Maybe it's because I haven't yet found a reason why it couldn't happen. Maybe there's something in me that wants to go bigger. I think we also miss that in Europe. If you look at the ten largest technology companies in the world, they are all Chinese or American. There is no European company among them. I want to be like those giants.

If you look positively, just the market for small-volume outsourced production, without assembly, just single parts, is 40 billion dollars. But the total manufacturing industry worldwide is 12 trillion. So that estimate I just gave is a very small, constructive, positive part.

It's about automating many of the processes. If you look at a milling shop, about half of the total production cost is labor. Within that labor, we have already automated half with our processes. Ultimately, you could automate even more, leading to fully automated factories that are directly connected to our platform. Our platform could then send orders directly to machines and monitor production quality. Often there is still a local party in between, but what we do is automate all of that away: validating designs, translating files to the specific software of that production facility, and so on.

That's a bit more difficult. Ultimately, automation throughout human history has led to us no longer living in caves and having much better survival chances. It would be very arrogant to say that in 2019 we have reached the perfect state and don't need to automate anymore. But I do see a transition period where there will be job losses that cannot be immediately compensated. That is a difficult phase. But I think the ultimate result of artificial intelligence, automation, and robotization is that people worldwide can produce more with less labor and thus enjoy greater prosperity. The question is whether that prosperity is distributed fairly. That's a question I always ask. There are two aspects: can people retrain quickly enough to remain relevant, and is prosperity distributed fairly? I think we have a role to play in ensuring that happens.

That's very outdated. Everyone used to say that, but look at this MacBook—it's a pretty nice product, and it's largely made in China and Taiwan. Through our platform, it's very easy to have a design made there. But I try to keep production as local as possible. However, we ultimately want to offer the customer the fastest delivery time and lowest costs. As more automation takes place in production, and we are the only party doing this on this scale, the labor component will become smaller, and it will become increasingly logical to produce very locally.

Yes, we have an office in Amsterdam and one in the US. We are now opening an office in Chicago with about fifteen people. We have also hired the first people in Germany and France. So we are internationalizing quite rapidly.

I studied industrial design at Delft, so I came into contact with 3D printing and milling early on, in 2002. I designed something and had it milled by hand. Now, most of our people are software developers and mechanical engineers. So we are all engineers, really.