UPDATE: Tl;dr – Digital Ocean tried to make their CAPEX numbers look bigger than they are, which we have to respect.
Digital Ocean’s S-1 is out and they use the word cloud 212 times, which definitely suggests they see themselves in the cloud business. Lets see how they stack up on my favorite dimension: CAPEX.
Observations from the filing:
- “Capital expenditures as a percentage of revenue” is one of the company’s five key business metrics (unclear why they even bother with the other four). But they make you compute the actual CAPEX numbers yourself (which is a tell they’re being slippery). Between 2018 to 2020, the arithmetic they want you to do says they spent $116M, $94M and $121M.
- However, they pad that CAPEX number (with the careful legal disclaimer “We are not aware of any uniform standards for calculating these key metrics, and other companies may not calculate similarly titled metrics in a consistent manner, which may hinder comparability.”) by including capitalized software development and the purchase of intangibles. Adjusted “Hard” CAPEX (i.e. the actual data center infrastructure that CAPEX obsessives care about) looks more like $93M, $64M and $102M. So about 20% lower and the cumulative spend over the three-year period drops from $331M to $259M.
- They did less than $5M in capital leases a la AWS and Microsoft in 2018 but none more recently. Seller-financed equipment (presumably servers) has also fallen significantly from $49M in 2018 to $4M in 2020.
- The bedrock “Capital expenditures – property and equipment” line – independent of any other financial gymnastics — actually looks pretty good in an up-and-to-the-right way: $38M, $54M and $98M.
- They depreciate servers over five years, versus four for the hyperclouds (I’m not actually sure about Microsoft’s interval).
- They have 14 data centers in co-lo facilities around the world.
Here is Digital Ocean on the standard chart we use for other clouds (and clowns):
Here is that chart showing the difference between “Reported” and our quite recently invented adjusted “Hard” CAPEX:
I’m always on the lookout for new cloud players. Given the size of the market, there should be a lot more niche players like Digital Ocean. The question is whether any of them can transcend niche status and challenge the hyperclouds. Digital Ocean’s CAPEX, at least, says no. They spend a substantial percentage of revenue on CAPEX, but the absolute quantities are not going to change the competitive landscape.
Digital Ocean has focused on developers and moved beyond VM hosting to offer a variety of storage, managed database and PaaS services. They even offer “Managed Kubernetes designed for you and your small business”, so have a strong sense of humor. But it hard to make enough margin dollars as a low-cost provider serving developers to really invest, especially given the breadth and resources of the hyperclouds they are going up against. The founders are gone (they resigned from the board this month) and the CEO is a finance guy. A 25% growth rate doesn’t keep up with the hyperclouds who are all growing faster on an order of magnitude or two greater revenue.
And this is a little cruel (and my motive for that will be revealed very shortly), but here is Digital Ocean’s CAPEX in comparison to the hyperclouds. That yellow line you may mistake for the x-axis is Digital Ocean’s CAPEX spending (to reuse a joke I once made about Oracle’s CAPEX spending).
