Xometry (XMTR)
A marketplace business of questionable quality
Note: these are notes from one weekend of work I did in June. This is more of an action plan for the key questions I’d want to answer in a full diligence than a final or conclusive report.
Overview of business
Marketplace
The marketplace is Xometry’s core business connecting allowing buyers to order just-in-time parts from a network of suppliers through the platform. Buyers are quoted a price, and Xometry then handles the process of finding the supplier, pricing the work, and getting it delivered. Thus, Xometry takes on risk if the project costs more than they quoted or if the work quality is bad.
The marketplace is ~80% of Xometry’s business with ~30% gross margins.
Supplier services
Xometry Supplies sells supplies (materials, tools, etc.) to sellers. They also offer advertising, cash flow management (including factoring), and a job management product called Workcenter.
Supplier services is ~20% of Xometry’s revenue with ~80% gross margins.
Recent context
The business decelerated significantly in Q3 ’22, going from growing mid-teens QoQ to 8% in Q3. In Q4, QoQ revenue actually declined, and management went from guiding to adjusted EBITDA profitability in 2023 to adjusted EBITDA profitability only in Q4 ’23. They are now guiding to 23-25% revenue growth in 2023, down from 75% in 2022.
Overall thoughts on marketplace businesses
From my observation, the best setup for marketplaces when you have a fungible product / service, with fragmented supply and demand and low order size but inconsistent ordering patterns in terms of time or place.
As such, marketplaces struggle along the following axes:
Not fungible (customization required) – hard to price (e.g., Angi’s fixed price initiative)
Large order size (tends to correlate with not fungible) – too much incentive for people to find a way to circumvent the marketplace take (e.g., Opendoor)
Consistent ordering pattern – incentivized to move off the marketplace once you find a supplier you like (e.g., SMBs use Upwork / Fiverr for one-offs but it would make no economic sense for a BCG / McKinsey to move their in-house slide production / research group to Upwork given the consistent volume)
Non-fragmented demand or supply – no need for marketplace, go straight to the large supplier or customer
As it pertains to manufacturing, I’d segment the market into (1) one-off, prototype manufacturing, and (2) scale manufacturing.
Prototype manufacturing meets all the marketplace criteria except being fungible. Xometry claims to have solved the pricing problem with their AI solution. If we believe them, that segment of the market is fit for a marketplace model.
However, scale manufacturing fits none of the first three criteria. Thus, my ingoing hypothesis is that it is not suited to a marketplace model. Furthermore, I strongly suspect the vast majority of manufacturing TAM sits within the scale bucket.
Key questions for Xometry
As such, I’d focus diligence of Xometry’s marketplace business on the following key questions.
Do buyers only use Xometry for prototypes, or do they scale up volumes (number of orders) and quantity over time. In other words, does Xometry only play in the “prototype” segment of the TAM or have they shown an ability to move into the “scale” piece?
Most of the commentary I see on Reddit suggests that Xometry gets used for:
One-off, small runs of custom parts where quality is not critical, but local shops aren’t able to fulfill the order for some reason (too tight turnaround, needs custom materials, etc.)
Small engineering firms that mostly do one-off prototypes, so they don’t feel the need to invest in relationships with specific shops or machinists
Even on prototypes, most engineers seem to prefer working with machinists with whom they have long and personal working relationships.
For larger quantity orders, most companies have a list of approved vendors that procurement sends designs off to. There is probably favorable pricing built into these “approved” relationships. One seller wrote the following on Reddit.
I generally don't see things with a quantity much higher than 50 and that's pretty high for them, lots of 1-2 pieces.
This does not bode well for Xometry’s move into large volume production.
However, Xometry presents strong buyer cohort metrics in their investor decks – for example, their 2016 cohort spent 10.7x on the platform in ’22 vs. ’16. Every other buyer cohort has also grown spend significantly over time. How do we square this with the anecdotes about one-off production from Reddit?
The company says that 50% of marketplace revenue comes from their top 200 buyer accounts, and that Xometry still has a small share of wallet vs. those top 200 accounts’ total manufacturing spend.
What’s likely happening is that in the large accounts, even one-off or prototype volumes can add up over time as a greater % of engineers in the organization move onto Xometry. I’d love to talk to some of those large customers. If I see evidence that they’re moving scale production to Xometry, it would certainly make me more interested. But logically, that wouldn’t make sense to me, even if Xometry were to offer volume discounts — just like it wouldn’t make sense for BCG to move its back-office slide production teams to Upwork.
Further work I’d want to do: talk to large customers and see if they’re increasing the number of large orders through Xometry. Talk to Xometry formers and see if the they were seeing more large quantity orders.
What is the prototype TAM?
I’d like to do a bottoms-up build of prototype parts TAM. Maybe it’s large enough to support significant growth even if Xometry can never crack scale production.
However, Xometry’s decelerating growth even with huge S&M spend, at what should be the beginning of the S-curve (revenue is a small % of management’s stated $2tn TAM) suggests that the prototype TAM is not as greenfield as I’d like.
Further work I’d want to do: Bottoms-up estimate the TAM, and try to find the revenues and growth profiles of all major competitors in the market. Try to understand what prototype TAM is and how much of it goes through “traditional” channels vs. digital channels. It would be encouraging if our estimate of TAM is much larger than the revenues of all key players, and if all players were growing quickly with new logos.
Is Xometry able to acquire buyers and sellers profitably? (e.g., are they differentiated vs. competition?)
Xometry’s largest competitor is buyers going directly to machine shops. There are also “service bureaus” that have self-service online ordering. Finally, there are online brokers who allow you to upload a design and have a group of shops bid on it.
Some of the large brokers and bureaus include Protolabs, PartsBadger, Plethora, Fictiv, Stratasys, Materialise, UnionFab, and before Xometry acquired them, ThomasNet. There are many more. This is a very competitive market, and it’s unclear how differentiated Xometry’s marketplace is. One indication is the amount they spend on S&M.
This portrays Xometry far more favorably than it should – marketplace revenue has been effectively flat since Q3 ’22, and in that period, Xometry has invested 52-68% of gross profit into S&M.
This puts Xometry’s profile much more in line with mid-tier marketplaces than best-in-class ones. In the last 9 months, Xometry’s sales spend efficiency is actually worse than its legacy competitor Protolabs. It’s also telling that even after they started “right sizing” the business for profitability in Q4, Q1 ’23 S&M as a % of GP was still 57% — my guess is revenue would collapse if not supported by heavy S&M spend.
While Xometry is more macro-exposed than some of these other marketplaces, it is also much earlier in its S-curve than Airbnb or Etsy, if you are to believe management. The heavy sales spend is an indication that it doesn’t have the same “market pull” and unique network effects that best-in-class marketplaces do.
This is backed up by Reddit commentary that doesn’t paint Xometry as a head-and-shoulders better way of manufacturing parts than alternatives.
I use Protolabs more than Xometry. The quality from Xometry can vary significantly because of the fact that a different shop gets the job each time. Protolabs at least I know what I will get.
Further work I’d want to do: run a survey to get a data-driven view of if customers see Xometry as one of many equivalent options for manufacturing, or if it has some unique capabilities.
Do sellers shift a larger % of their business on platform every year, or do they do the bulk of their volume off-platform and use Xometry to fill spare capacity?
I couldn’t find one Reddit comment from a supplier raving about Xometry. Reactions ranged from lukewarm to very negative. Many complain that there are only a few “good jobs” that can actually be done profitably, and they get taken quickly. The commentary strongly suggests Xometry is a platform machinists use to fill excess capacity, not as the core of their business.
Our shop does work for Xometry regularly. We use it as kind of a fill in while we have machine time open. Do one offs and production runs… Need to keep the job board up and keep an eye on it because good jobs go quick.
This is negative for Xometry because if sellers aren’t increasing the % of capacity they fill through Xometry over time, it will become very expensive for Xometry to continually acquire new sellers and increase supply on their marketplace to keep pace with demand. And if supply doesn’t increase, prices will increase for buyers, increasing the cost of retaining buyers on the platform.
It is telling that while the company discloses cohort metrics for buyers, they do not for sellers. Based on the commentary I read, I don’t think it would look pretty.
This will not get better as Xometry pushes to increase its marketplace gross margins to their 35-40% long term target. For Xometry to improve its gross margins without raising prices for buyers, it needs to cut into prices it offers sellers, which hurts seller retention.
Further work I’d want to do: find sellers who do significant volume on Xometry (there are some based on case studies from Xometry’s website!), and understand why they do it. I want to see if only a niche group of small sellers will do significant volume on Xometry or if the broader universe, and particularly large sellers, will.
Other misc. questions / concerns
Xometry’s ‘22 numbers are boosted by the Thomas acquisition. Thomas had 1.3M registered users and 500K sellers. Xometry bought it a year ago and still have only has 2.5K active sellers. I’d want to dig into this more.
I haven’t focused as much on supplier services. That business accelerating could be a bull case but it’s capped out at ~$20M of revenue per quarter for a year. As such, for an initial screen, it’s not worth spending much time on.
Cyclicality. Some of the decel last year was due to exposure to cyclical end markets like autos, O&G, and retail. Xometry currently burns $50M per year. They have a high OpEx base that they’re struggling to bring down. Burn could really spike in a recession.
Levered to US manufacturing. Is there a bull case to be built around more manufacturing coming back onshore?
Overall thoughts
To conclude, Xometry doesn’t seem that interesting for the following reasons:
Unclear how large its actual TAM is – prototyping / one off vs. scale TAM
Intensely competitive environment and unclear how differentiated Xometry is
No evidence that wallet share is increasing among sellers, and limited evidence that it is growing among buyers
No evidence of notably high NPS among buyers or sellers
I haven’t spent much time on valuation because I’d want to get comfortable with these business quality questions first.