Expert interview: Xometry’s Greg Paulsen On the Rise of the Manufacturing As A Service Business Model
27 June 2019
Manufacturing as a Service (MaaS) is a business model that could have a tremendous impact on the global manufacturing industry. Put simply, MaaS platforms work with a network of manufacturers to offer production services on demand. The benefits, supporters say, include delocalised, agile manufacturing; enabling service providers to maximise their production capacity and offering clients access to a wide range of suppliers at competitive rates.
One company championing the MaaS model is US-based Xometry. Founded in 2014, Xometry has developed a platform that hosts a network of machine shops and 3D printing service bureaus, allowing customers to procure high-quality parts on demand.
We speak with Greg Paulsen, Director of Application Engineering at Xometry, to discuss the factors driving the growth of Manufacturing as a Service, what the future holds for this business model and how companies can make the most of additive manufacturing.
Can you tell me about Xometry and the services you provide?
Xometry is a Manufacturing as a Service platform that links buyers and suppliers through a single site.
The initial problem we were aiming to solve was the opacity of procuring custom manufactured parts.
Previously, I had a similar role where part of my task was to get rapid prototypes made. I’d have to bundle up my 5 or 10 files in a zip file and ask our suppliers to fill out a spreadsheet to understand how much it would cost. I’d wait days to get a response, and sometimes wouldn’t even get a clear answer after this time.
What Xometry does is use AI and machine learning, which enables us to interpret the 3D files and get pricing right away on multiple technologies. Instead of waiting for an RFQ, we provide instant pricing and manufacturability feedback straight on the site over various processes, including six 3D printing technologies and almost 60 materials.
It’s somewhat analogous to Amazon where you go to buy things, only that you go to Xometry to buy parts.
But we still need to get those parts made. And this is the other problem of procurement: you find your favourite shop and they start getting all of your work. But then all of a sudden that work is cannibalising on all of the other jobs in that shop because they have a limited capacity.
Xometry has a large network of manufacturing partners that provide CNC machining, sheet metal, as well as additive manufacturing services.
Essentially, the process works similar to Uber – we say, “Hey, there’s an SLA job where we’re going to pay you $1,960 and we need these 17 parts in seven business days, do you want to take it?”
Since we have this large network, shops can take orders to their capacity and choose whether or not to accept.
Otherwise, you can say, “Right now I’m out of that capacity”, and that allows you to actually manage capacity dynamically based on whatever their situation is at the time.
Our perspective is that we actually don’t have competitors; we have potential partners. We have service bureaus, large and small, that have different technologies, and they get to see these opportunities.
It’s a win-win because the shops are getting work without having any need to do marketing for it. We’re getting fulfilment and quality parts made. And finally, the customer has a one-stop location to get their parts ordered over many manufacturing technologies.
What impact is the Manufacturing as a Service business model having on the way in which parts are manufactured?
Manufacturing as a Service is an exciting business model for manufacturing as a whole, but particularly for additive manufacturing.
This model somewhat resembles the Internet of Things, where small devices now communicate with other small devices creating their own mesh. The same things are happening with services.
There are some jobs where a portion of those jobs may be the perfect fit for one service provider, but the other part of the task may not be the best fit. The pricing may be high on that portion, or the bureau may just not have the capability to do it.
By spreading the work over a network, you’re able to get those people who are best in class to do that work. So your overall quality increases from your output, and you also get more competitive prices.
For example, a CNC or DMLS job may be a perfect fit for somebody, but it may be more expensive at other shops. So we tend to price towards the ones that have a better fit.
Overall, pricing becomes more competitive, yet the shops still get to take the work that is the best fit for them.
From the demand you see through your platform, what proportion of that is for additively manufactured parts versus traditional methods like CNC and injection moulding?
In terms of number of parts, additive manufacturing is king.
We get a lot of piece part work, particularly when you look at the high throughput technologies like laser powder bed fusion, which includes selective laser sintering (SLS) or HP’s Multi Jet Fusion (MJF). With these, you see a significant amount of parts, either sets of parts or individual parts.
Of those, I would say probably between 15% and 20% are some level of end-use parts for low-volume production needs.
While there’s a lot of output in that industry, the costs are lower. When you look at revenue, methods like injection moulding, CNC machining and sheet metal still have larger order values, just because they are a more expensive process.
It’s an interesting balance because if you look purely at the number of parts ordered, then that is larger on the AM side. But if you look at where the revenue is typically higher, then that will be the traditional technologies.
Either way, our goal is to put them all on the same platform, so customers have just another tool in their toolbox.
How would you describe the current state of the additive manufacturing industry?
It is definitely still a little wild west.
One issue is that most engineers and designers are used to traditional approaches: injection-moulded parts, machined contours and so on.
They are used to this design, methodology and output. When they get an injection-moulded part, they don’t expect a stepping surface finish or matt roughness to it.
With additive, the cosmetics are different: you’ll see grow lines, or you’ll see features and detail resolutions that may not be achievable with certain processes.
So I think understanding what the strong points of additive are, especially per process, is a barrier to entry for many people.
But no matter what process you’re using, it’s all about educating everyone on design for manufacturability. We spend time creating guides on design for manufacturing for all processes that we offer.
Oftentimes, customers will come to us with an idea of producing a part with a certain process, for example, CNC. But then we find out that the form, fit and function of that part and what they need it for really would be more cost-effective to do with 3D printing.
Similarly, they’ll come to us with an idea of injection moulding, and we’ll find out they’re so early in the design process, that there might be multiple iterations, and that it’s not cost effective to go with that process — additive manufacturing would be a better choice.
So instead of coming with just an idea, we want to help them make an educated decision on what process they need, based on the end use application of the part.
That’s been a big part of my career for the last decade: teaching where additive can be applied, as well as the strengths and trade-offs.
On the other hand, I’m seeing the technology being used for end-use production. That’s partly because engineers and designers are seeing the value of no tooling needs, mass configuration abilities and just-in-time inventory.
I also think customers are getting used to seeing more additive parts. Ultimately, it’s becoming much more acceptable in the manufacturing marketplace.
It’s also really interesting to see the growth and diversity within the industry. I’m excited because there’s work for everyone right now. There are more and more manufacturing opportunities because we’re also in the world of customisation. Custom manufacturing is becoming more and more popular and we haven’t seen it slow down.
How can companies identify the best use cases for additive manufacturing?
It’s very challenging! But what I can say is that it starts with the design first. Often, we look at the customer’s requirements and we help them narrow down their choices that way.
All printers require that 3D CAD file. So the very first thing that we need is the 3D design, as that will be the lens through which we try to figure out what process to use.
Certain printers have different size considerations. Usually, if a part is under nine inches, any process could be considered. If it’s between 9 to 14 inches, you may be better off with a technology like SLS, FDM or SLA.
Over 14 inches, I’d lean more towards FDM processes, but SLA still can go up to about 25 inches. Over that, you have to choose FDM if you want to print your part all in one piece because that’s the only technology with which you can repeatedly make a part that’s over 25 inches. So sometimes it’s based on the size constraints.
Other times, it’s a matter of what you need from a fit or function point of view. For example, if you want parts that need to flex and to pop back into place, only a few materials and technologies are good for that repeated ductility. So for repeated wear, I may suggest some of these powder bed processes like Multi Jet Fusion or laser sintering.
Sometimes it’s also the question of aesthetics, where technologies like SLA, Polyjet and Carbon’s DLS tend to build smoother and more aesthetically pleasing parts. Sometimes that’s the most important aspect for a customer.
If you could make three predictions for the future of 3D printing, what would those be?
First, I think hybridisation is going to be more important.
Second, feel that software-driven design is going to be more important for the engineer. We’ll see more software tools that help engineers design parts better for a given process.
I also think software-driven build setups like orientation, pre-deformation will be part of that. It’s so important for repeatability and managing expectations of what’s going to come out. These developments will help to reduce the number of iterations needed, especially if the goal is printing for production.
The third is that the Manufacturing as a Service model will become more widespread. Manufacturing as a whole is still operating as though it were the early 80s.
Today, you need to be able to know what type of digital traceability you have in your own internal systems in your shop. Can you say what you manufactured two years ago, to the exact time and day? Most companies probably don’t have that level of traceability right now.
So it’s important to be able to digitally keep records of everything, including the builds. And being able to distribute that across a manufacturing ecosystem is very important overall.
What advice would you give to service providers offering AM services?
From a service provider perspective, there are a few things that are very useful.
First, it’s important to think about how to increase your throughput and how many deliverables you can put out per week or per day. One way to do this is by splitting your batches.
I’ll use FDM printers as an example, just because they’re unique in the sense that you’re running one material at a time, like yellow ASA, or Ultem 1010 or Nylon 12. In order to get more parts per day, it’s actually more useful to have an army of smaller machines than a few super large machines.
So if I stack a job that takes seven minutes to run with a job that takes 24 hours to run, then that seven-minute job is going to take at least 24 hours to get done. By being able to split your batches and run smaller jobs that you could then turn over more quickly, you’re going to deliver more per week.
So you need a mix because sometimes you need some larger parts. For example, we have some parts where the build time is 60 hours, meaning that it’s going to occupy the machine for several days.
But in the meantime, your smaller machines are going to be outputting. You’ll find this too for metal printing. It’s amazing how many parts are under four inches, and if you have a small set of smaller platforms, they will be pumping out parts while you’re working on the larger projects on your larger platforms.
My advice is to understand that mid-sized platforms like powder bed fusion, Multi Jet Fusion and SLS are really important. You’re usually filling those up with as many parts as you can.
But again, being able to keep the machine running, having multiple exchangeable trays, for example, is the most important thing because you want to see how many parts you can get out per week.
You need to think about how you can set up your builds and production line in a way that everything that’s happening in the machine, your post-processing and shipping, all take roughly the same amount of time. That’s going to be key if you want to achieve a higher revenue per week.
The need for repeatability with 3D printing is a key talking point. What are your thoughts on that?
I’m very passionate about that.
Let’s say I have a design for CNC machining and a drawing that says that everything will have +/- 0.127 mm (0.005”) tolerance except for one area that needs to be +/- 0.050 mm (0.002”) tolerance.
In machining, your print can dictate some of the tolerances of the part. A machinist will be looking at that print and adapting their code or how they’re moving their tool to hit a certain tolerance. So essentially they can stop and measure the tolerance at certain points.
In 3D printing, it’s not the print that dictates the outcome of the part — it’s the printer itself.
So if I take that same part and produce it in selective laser sintering, I have high repeatability on the outcome of that part. There’ll be some variation, but very little. Even across different platforms or across different services, you’re going to have a relatively good expectation of the outcome.
However, it doesn’t necessarily mean that it’s going to hit that print tolerance. Usually, internal holes tend to shrink down a little bit more, so I may actually do a 3D CAD offset to compensate for this.
But a lot of times if you’re fine-tuning a part, especially a part for production, the first one is where you’ll do a fit check. Any modification that needs to be done will probably mean going back to the original CAD file to manipulate it.
What I want to see is software that does this automatically. I want software to essentially become a virtual machine where it can simulate a build and the outcome like shrink, warp or any deformation. This is so important, particularly in metal AM.
It’s very easy for me to shave off some nylon or drill it out if I need to change a feature, but metal is metal. You need a machine shop to be able to manipulate it.
So being able to virtually create a part and then pre-deform the part, so that when it gets built, the stresses and post-processing forms a part closer to CAD, I think it’s very important moving forward.
In your opinion, what has been the secret behind Xometry’s success?
We actually base our pricing models on the marketplace. It’s our machine learning algorithms that build the pricing, not us creating a price and then dictating it blindly.
We take into consideration what the marketplace is accepting. For example, how much does a typical shop charge for this for a certain part, or what is the typical customer price?
We’re creating a competitive marketplace that is actually giving fair pricing out to manufacturers.
It’s an entire ecosystem that makes it successful. We’re only as strong as our manufacturers. And we are empowering small businesses — many of our machine shops have less than 20 employees.
We’re empowering these people by being able to give them work on demand at a click of a mouse. We’re even bringing them up to the quality standards that we’re trying to achieve to create a consistent experience for our customers.
Since we have this quality, we’re able to bring them up to that level with our own internal digital traceability, our job board system where they can upload things like inspection reports, any progress photos and track the steps in which they’re manufacturing the parts.
We have all that system in place to help go beyond what they can do as just the shop alone.
What does the future look like for Xometry?
We recently received $50 million in funding, which is extremely exciting. We have investors like Highland Capital, GE Ventures, BMW Ventures, and recently with this $50 million funding, Dell joined the game as well.
With that, we’ve been expanding different parts of our marketplace and looking at how can do more for the manufacturers.
We can provide them with jobs, yes. But can we also help them get raw material? In the case of CNC, can we help them get the tools that make that job, like end mills and cutters? And for additive, can we help them get other additive materials to help them make better products for customers?
For this, we’ve just opened Xometry Supplies. They have over 50,000 tooling SKUs on that, so tools like mills, cutters, raw materials, like aluminium, are available.
The aim of that is to help nurture both the external people that are coming in and ordering from there, as well as our internal supply chain for our manufacturing partner network.
So you’ll start to see these ecosystems building up as Xometry becomes even more of a platform for manufacturing transactions.
To learn more about Xometry, visit: https://www.xometry.com/