One-on-One with the CEO of DigitalOcean (DOCN) - Transcript
Originally published on 2-26-2024
Written by Ophir Gottlieb
Lede
DigitalOcean (DOCN) reported earnings on 2-21-2024 and we reviewed it.
I have completed my one-on-one conversation with the CEO and include that transcription below.
As a reminder, all of our CEO and CFO interviews for all companies are available on the Interviews Tab.
One-on-One with the CEO of DigitalOcean (DOCN)
Rob Bradley (Vice President, Investor Relations, DOCN)
Before we get started formally on some of your questions, why don’t you just give Paddy a little bit of the backstory on Capital Market Laboratories?
Ophir Gottlieb
Oh sure.
Hey Paddy, it’s nice to meet you. Congratulations on the DigitalOcean CEO job and your first week.
Paddy Srinivasan (CEO, DOCN):
Thank you. Appreciate it.
OG:
So, I run a boutique research firm, Capital Market Laboratories.
We cover probably 15, it changes, let’s say 15 to 20 technology companies.
We are members of FactSet consensus for some of the companies and we publish on Refinitiv as consensus as well.
This goes out not just to subscribers via email, but through the terminal business.
I’d say we kind of have three different customers.
There’s just straight down the middle institutions. We had about 300 different institutions reading in the last year.
We also have what I kind of call professional non-professionals – people who probably were in the business at one time and are now managing maybe their family’s wealth or maybe a little home shop.
And then we do have a few thousand just straight down the middle retail subscribers.
So, it’s a pretty large audience.
PS:
Got it. Okay.
No, it’s wonderful to talk to you.
Rob gave me a little bit of background, but it’s wonderful to hear it from your side.
So great. It’s great to make the connection and let’s get into it.
OG:
All right, sounds great.
Thank you, Paddy.
Thank you, Rob.
RB:
Wait, sorry, I don’t want to steal your time, but why did you initially, right after we had gone public, so going back three years now Ophir, very early on, Ophir strongly endorsed DigitalOcean.
So maybe before we get formally started, what were the one or two catalysts that made for that decision?
OG:
Yeah for sure. So, we can broadly start with infrastructure, right? It was clearly a place to be.
But then through personal experience as an AWS customer, I knew full well that there had to be a solution that was just for SMBs.
Not just colloquially, but very personally.
It was very personal to me, and I saw the community kind of growing around DigitalOcean even before you guys were public.
And then Yancey took it public, and he transformed the company from kind of an untouchable company to one that actually looked like how one should look.
Even then, I knew about the growth trajectory for SMBs. There’s an accelerant, there’s an accelerant to the downside in a bear market, right?
And there’s an accelerant to the upside in a bull market.
I’m basically bullish on the US economy and SMBs, and I think DigitalOcean with the right leadership just has this remarkable opportunity basically to be the AWS of SMBs.
I really believe that to be.
And there’s been a hiccup due to COVID.
I like to say COVID was the irritant, and the further we get from the irritant, the less irritated we will be.
So, we’re moving through this period, but I thought and continue to think that DigitalOcean is one of the most interesting investments basically for public markets in the world.
That’s how it all started with me and DigitalOcean.
PS:
That’s very high praise, but I largely agree with you on the fundamental premise.
I really believe that.
And that is a conveyor belt, which if you are a believer in the US economy, like what you just said Ophir, I think that ecosystem is thriving, and it has to be served.
And I think if we are disciplined and focused on that in, it’s a phenomenal opportunity.
OG:
And also, I would say, I mean in fairness, the most innovative new companies, developers don’t, I mean all due respect, they don’t really care about IT that much.
That’s just, I don’t pay them a lot of money because they know how to run servers on Amazon.
We need to get up and get running and get going.
And that’s a really, really big, this idea of DevOps is real.
I mean, I’m living it. It’s not just the colloquialism, it’s real.
Developers are becoming the entire tech ecosystem.
PS:
Right.
OG:
So, my first few questions, Paddy, all surround the growth levers for DigitalOcean.
I’ll try to go sort of one by one to avoid any of those weird compound questions.
I’ll touch on all of them or just go one by one.
I do actually cover Amazon and Google and some enterprise software companies like Datadog, Confluent, and CrowdStrike, things like that.
For the ones who have already reported Q4 and guidance for the full year, there has been consistent language that I’m kind of putting air quotes up.
This “cost optimization regime” is either nearly done or actually done.
So Datadog, for example, made a point to say that the very specific cohort of customers that were the first and the largest optimizers have returned to growth in Q4.
Obviously DigitalOcean is dealing with much smaller SMBs, it’s not a hyperscaler.
Having said all that, I understand that the cost optimization pressure has eased, and I saw that because NDR essentially didn’t get worse.
But what I don’t have a good feel for Paddy is whether growth is restarting in the cost cutting cohort or if it’s just less shrinkage.
PS:
Yeah, it’s a good question.
So, we don’t disclose that.
And you have to understand this is my day eight, officially on the job.
So, I’m pretty sure we don’t disclose the independent levers of our NDR number, but there are two truths in what you just described.
One is our curves will look very similar to the hyperscaler curves, but there could be a time displacement.
What I mean by that is when we started looking at our sub levers of the NDR, we also saw similar patterns perhaps with a little displacement from a time perspective.
In terms of the individual components themselves I think remarkably our churn number was not that different across the last several quarters.
It is the contraction or the cost optimization, if you will, and the expansion are the two that where the biggest contributors to our NDR dip and you are right that we are seeing in constant cohorts, we are seeing a gradual reduction in the contraction or the cost optimization.
And I think it is a little too early, I don’t want to go on record to say that we’re seeing those same cohorts now starting to grow again.
I think it is a little early for us to make that statement, but we’re certainly starting to see the cost optimization stabilize and get back to recovery mode.
OG:
I want to talk about another opportunity for DigitalOcean, which is to reach out to smaller hyperscaler customers, say in the $200,000 to a million dollar a year customer range.
So, when I spoke to Matt Steinfort last quarter, he noted that customers that grew up on hyperscalers as opposed to those that grew up on DigitalOcean, they have specific requirements that they may need to be met in order to be able to move their workloads to DigitalOcean.
And there was really good feedback from them that it wasn’t this sort of black box.
In other words, it wasn’t complicated necessarily, it just had to be done.
So how is that product roadmap and sort of hyperscaler, I don’t know, I’m going to call it poaching going, is that a focus of the sales team in some regard to go after those and is there any kind of, I don’t know, anecdotal success?
PS:
Yes, it is a great opportunity for us, and I would say even more than poaching, it is more of the companies that are afraid or don’t want to get into a hyperscaler lock-in the spillover functionality or spillover workloads that they want to distribute.
And even in my previous company, we had deployments in three of the four hyperscalers, and I think that is becoming very much a trend and that is something that we barely participated in at DigitalOcean.
So, it is an initiative that we are starting to ramp up, but I would say that we have to be super disciplined and careful about not chasing the breadth of functionality that will distract us from being great at what our core mission is.
I feel like there’s a lot of headroom in our core mission in terms of both on the core DigitalOcean side to simplify and get some of the core things, but also the AI/ML where it feels like it is super early, and we can replicate some of the playbook that DigitalOcean had in the core infrastructure and platform as a service on AI/ML and data pipeline.
So, I feel we have to be judicious in going after the Hyperscaler poachers or the cloud lock-in avoiders, but we will do that if and only if those spillover workloads fall within the parameters of what we currently already offer, if that makes sense.
OG:
It does, yeah.
It’s always been a tricky balance for DigitalOcean.
There’s no doubt you want to get larger customers, but you can’t blow up the complexity and then totally destroy the value proposition of DigitalOcean to smaller companies.
I’m quite sensitive to the balance. It’s not easy,
PS:
But I’ll give you a couple of examples on the flip side.
So, there are several customers that I have noticed now making calls out to our top 25 customers, and I’ve talked to a handful of them, and one thing I’m noticing is that there are some customers that grew up with us historically, but now have more on the hyperscalers than they have with us.
And they love us.
They want to give us more business, and they want to start consolidating and rebalancing their workloads on us.
I think that’s a great opportunity for us to go after because these are DO native customers that have outgrown us for whatever reason over the years, but still have a substantial presence so much that they’re still one of our top 25 customers and they’ve always had the loyalty and willingness to partner with us and want to keep doing business with us.
OG:
Okay, yeah, that’s really good feedback. Yeah, I don’t think anyone is with DigitalOcean and with a hyperscaler and wants to be with a hyperscaler.
I’ll just put, that’s my view.
PS:
Yes.
OG:
I want to go back to a September presentation with Goldman Sachs.
Obviously, you weren’t CEO then, but DigitalOcean was talking about the early stages in developing new go-to market motions.
But in particular, it’s getting traction in nurturing potential high growth customers.
In other words, I think what was being said is that kind of what you’re talking about, customers who are on DigitalOcean, they have reasonably large workloads on DigitalOcean reaching out to them to tell them there’s more you can do and what do you want?
I know one of those initiatives was, for example, storage.
Yancey gave this great statistic about how much more customers pay if they’re using DigitalOcean storage versus those that don’t.
So just in general, is that a part of your priority as CEO to really get this, I call it the NDR building sales motion?
Is that a part of your priority or is that down the list?
I just want to know what to tell investors relative to what you are focusing on.
PS:
Yeah, so I think I outlined this in my earnings report, but this is one of my top priorities.
So, I have two buckets of priorities.
One is getting the innovation and product roadmap delivered as fast and furious as we can, and there are obviously multiple buckets of that.
The second one is complementing our existing product led growth customer acquisition engine and augmented tastefully with a touch organization, a very small, efficient, scaled up touch organization that can help us with the retail and expansion motion.
So, we’re not going to boil the ocean.
We have 600,000 paying customers as you know, but we are going after the top say, 3,000.
So, we are looking at the top 100 which I’m time personally establishing a relationship with, but we are also deploying our CSM and account management resources on the top 3,000.
And there are a lot of tools these days available to have a fairly generous ratio of a CSM to an account because our accounts are obviously developer heavy, so they’re fairly sophisticated.
They don’t want us to babysit them by any stretch of our effort.
So, we can be very efficient in touching them and driving expansion.
I’m not sure what exactly what Yancey was alluding to six months ago, but I completely understand there.
I have seen some charts which talk about an obvious causation between multi module customers and the ones that are using our platform as a service components obviously tend to be a lot more sticky.
So, for example, we just announced the daily backups as a functionality just a few weeks ago.
I’m looking at the dashboard every day already just to look at how many customers are actually adopting those kinds of features, because that’s a great leading indicator of stickiness on our platform.
We’ll do that both using our product-led growth initiative to start exposing these functionalities from our admin console to administrators and developers, but also through outreach motions from our touch CSM team.
OG:
Yeah, I’ve always imagined a sort of extraordinarily white glove service from DigitalOcean for very certain size customers; it would be interesting.
I think it was in Q3 that DigitalOcean was talking about just sort of this initial 2024 guidance, which was very much in line with what was the official guidance just last quarter, something along the lines of 3% from Paperspace, 3% from Cloudways, and let’s say the rest 5, 6% from the DigitalOcean self-serve business.
Okay.
So, we’ve talked about the DigitalOcean self-serve business, Cloudways is going gangbusters.
You guys did make a point to talk about Paperspace and this supply constraint rather than demand constraint in particular to GPU capacity.
So maybe give me a little more color about what is and isn’t possible in 2024 and what does this supply constraint, how does the supply constraint impact DigitalOcean?
Maybe does it act as a tailwind in 2025 or, I know it’s a little early to talk about that.
PS:
Yeah, so obviously I don’t want to talk about 2025, but I can give you a lot of color on 2024.
A lot of the stuff that Matt and I talked about on the call in terms of the supply constraint are orders that we placed last year that we are getting delivery of this first half.
That’s why the CapEx is going to be front loaded this year is because we are actually finally getting shipments that we started the process late last year.
Now there’s a lag between when we get possession of these cards and we have to then prepare the physical infrastructure, test it, connect it up with our network.
There’s a whole process, there’s a process of certification.
So, there’s a lag between when we actually get the possession of these physical things and make it available for our customer.
The good news is the ones that we currently have, we don’t see a lack of demand for, and we are doing it in a very measured approach and we have a roadmap between now and the summer timeframe where we are bringing this capacity live on a month over month basis, and we’re already talking to many startups that want to buy half of the fleet or they want half of the fleet for two thirds of the time.
So, there’s a lot of time space splicing, and there’s a lot of different opportunities we have to figure out how we want to attract and identify customers that are going to be long-term users of this capability versus just one time fly by night type of startups that are just trying to test something on a model.
They just want a massive surge capacity for a weekend, and we’ll never see them again.
It’s a lot of experimentation and learning, but I think that the universe of startups, as you can imagine that are trying these high parameterized LLM training, there’s a lot of them now.
So, I hope that gives you some color.
We are not really supply constrained for the rest of the year because we are, as late as yesterday, we got confirmation that a lot of our backlog has been cleared up.
We are really looking forward to getting our hands on those things and getting it prepped up for our customers.
OG:
Yeah, it’s such an interesting position right now to be in as an infrastructure provider to know how much CapEx to spend on these GPUs.
I mean, eventually they’ll be used, but like you said, it’s just you do have those lurkers in there who are going to have this massive usage for a week and then you’ll never see them again.
I can’t imagine how hard it’s to model that. Yeah, it’s a really hard problem.
PS:
Yeah, I would love to hear your thoughts because while we don’t want to be left behind in this craziness of the GPU stuff, but on the flip side though, there are alternative headlines that are emerging just in the last couple of days.
Microsoft is building an alternative to GPUs. There was this company called Groq that established a benchmark, which is an order of magnitude faster without any GPUs, I think they’re calling it LPU.
So, we don’t know exactly how this is going to play out over the next several quarters.
OG:
Yeah, one of the biggest funds, I talked to one of their analysts, he actually called me to talk about Groq.
So, in fairness, he introduced it to me just a few days ago.
I don’t think any investment, any sensible investment in GPUs is going to prove to be a weight on DigitalOcean, perhaps CapEx, perhaps lumpiness in CapEx blah, blah blah.
But for anyone who’s legitimately looking at this company as a long-term investment, I think that the impact of large language models, not the building of them, the usage of them, the impact of large language models is going to increase the amount of product and therefore companies that comes.
I spoke with a CNBC reporter saying I think it’s going to be like a 10x in product creation.
And just between you and me, I don’t think it’s going to be a 10x in product.
I think it could be a 100x 1000x or 10,000x increase in product.
I think the amount of compute that’s going to be needed, both just standard infrastructure and also some on the GPU side is going to be larger than anyone can possibly fathom.
PS:
Well, that’s really interesting you’re saying, so the 100X is on the basic platform and infrastructure and database and the bandwidth consumption that the GPU will drag through versus the actual GPU consumption.
OG:
Yes.
So, I don’t have a really good sense of the long lasting demand for, or let’s say the long lasting durable growth of GPUs in general to train generative AI.
Obviously it’s coming and you’re seeing it, but if you’re talking to me about infrastructure in general, like the DigitalOcean infrastructure platform or AWS excluding GPUs, I think the number of products and the company formation that’s coming is going to be, it will break every model I’ve ever seen.
I would just add two zeroes at the end of every model I’ve seen from every fund I talk to.
I just look at it, I’m like “multiply it by a hundred.”
PS:
That’s music to my ears.
OG:
I just don’t think people fully grasp how quickly developers are moving.
I’m a developer too, but just how quickly developers are moving as they’re realizing what power they have.
And by the way, it’s not just driven by the demand or desire to create products, it’s also coming because there’s actually a palpable fear in the developer world.
Their jobs are at risk.
They may have to become entrepreneurs because a lower-level engineer is as of today replaceable with a citizen engineer.
I’m not talking about the best of the best.
The best of the best is something else.
But let’s just say like a beginner to middle of the road developer, citizen developers are breathing right down their neck.
As of today, I’m not even talking about iterations. So that’s just my view.
PS:
So that is fascinating because let me share with you what I’m thinking in terms of our Paperspace to DigitalOcean convergence and you tell me whether this makes sense.
So right now, we are doing this GPU stuff, but the company that we acquired last year, Paperspace, I don’t know if you know how familiar you are with that.
They have an integrated development environment for ML ops called Gradient.
So, think of this as in the way back machine if you go look at the developer experience of DigitalOcean eight or nine years ago, that’s exactly what these guys have built for an AI and developer pipeline and the discoverability of all these different language models and the ability to easily do the vector embeddings and things like that.
What we are doing right now is we are integrating that into the core DigitalOcean experience to do exactly what you were just talking about Ophir, which is once we get those developers come and start consuming these models and maybe write some vector embeds or do some fine tuning, they will get compute hungry and storage hungry and bandwidth hungry, and we want to just naturally give them an on-ramp to the core DigitalOcean platform.
So, start your adventure with us from Gradient and Paperspace perspective, but stick with us for everything else that DigitalOcean offers.
We are converging those two things. I think long-term, what is going to make us look differentiated.
OG:
Absolutely. And if you think about it, Paddy, for everyone that does their really intense training, they’re really only going to be measured as a success as a company if they are in fact eventually on normal infrastructure and compute, that’s the measure of the success.
Blah, blah, blah, blah, blah the accuracy of the model, a business is a success, not because benchmarks are hit.
It’s a success because sales are being made and if sales are being made, then there’s usage.
And if there’s usage, that’s infrastructure. Not GPU, not AI/ML necessarily.
I mean, as you know, once an AI/ML model is trained, it’s trained, not that there aren’t more to come but…it’s trained.
PS:
Then you’re doing the tuning and embedding.
So, do you have any data that you publish with say something to the effect of in five years for every model trainer or model builder, there’ll be a million developers that are going to be consumers of this or some kind of a projected ratio?
OG:
I haven’t fine tuned it enough to say that.
I’m still very much in the whirlwind I think we’re all in and I’m just putting out, I’m trying to get my arms around just orders of magnitude.
And by the way, I think the orders of magnitude question is a really important question.
Don’t get me wrong. I love Nvidia. I used to cover NVIDIA for a long time, but I see straight down the middle compute as a measure of the success of these AI companies because when it comes down to it, that’s what it is, right?
Not everyone is going to be open AI [creators of Chat-GPT].
There’s going to only be so many of those companies, right?
PS:
And it’s like, I’m sure you saw this video from Andrej Karpathy where we talk about these large language models being big kernel operating systems of the future.
How many kernels do we have? They’re like maybe 3, 4, 5.
It’s really interesting what you’re saying.
OG:
Everything else is product and product in the world of the internet runs on infrastructure, normal infrastructure, compute infrastructure, storage infrastructure, platform and service infrastructure.
Okay, so my last question is always the same.
Is there anything I didn’t ask Paddy that I should have asked or anything that you wanted to say, but I didn’t give you a chance to say because I didn’t ask the right question?
PS:
We talked about so much.
No, I think it’s amazing.
Two weeks in, I think my conviction that we can be such a massively larger company than we are, it’s just that much more stronger I would say, because of what I’m seeing right now in terms of some of the operational things that we will absolutely drive a truck through over the next several quarters and then just the generational opportunity that we are sitting in front of.
I’m super excited and I look forward to engaging with you. It’s a fascinating conversation Ophir.
OG:
Yeah, I’m super bullish on your company guys.
I see. Let’s get out of, I call it “after this,” which is after the irritant. Okay?
I call it the “after this” period.
I’m saying this is a 30% compound growing company and for several years to come after this.
That’s my opinion.
I’m not assigning your guidance; it’s just the opportunity is absolutely immense in my eyes.
PS:
I’ll take that challenge up and try to prove you right.
OG:
Yes, I believe you will. Alright guys, have a fantastic weekend. Paddy, happy day eight.
PS:
Okay, thank you so much.
OG:
Bye guys. See you next quarter.
Risk
DigitalOcean has many risks.
- The SMB world is infamous for customer churn as companies go belly up, or simply move their smaller operations to other vendors. A smaller operation is nimbler, and that nimbleness helps DOCN in acquiring new customers, but hurts it when trying to maintain them.
- It simply must be the case that the cloud titans (AMZN, MSFT, GOOGL) will push their “cloud lite” products and the allure of a titan behind the scenes powering the lite cloud could be very attractive.
As for DOCN’s risk disclosure on the subject in its S-1, we get:
Many of our competitors and potential competitors, particularly our larger competitors, have substantial competitive advantages as compared to us, including greater name recognition and longer operating histories, larger sales and marketing and customer support budgets and resources, the ability to bundle products together, larger and more mature intellectual property portfolios, greater resources to make acquisitions and greater resources for technical assistance and customer support.
- Other competitors will arise. As the company states in the S-1:
The markets in which we participate are competitive, and if we do not compete effectively, our business, financial condition and results of operations could be harmed.
- Macro everything is a risk: inflation, recession, a strong US dollar. For now, it’s both inflation and a tech recession.
We have been clear in our CML Pro webinars that we foresee a bumpy first half 2024 as the Fed grapples with the reality the inflation will never wave a white flag with mission accomplished printed on it. It will take the Fed’s faith that inflation has been tamed in order for a cut to happen before a recession makes it all too clear that they have waited too long.
We see the odds of recession as 50/50, but we see more upside than we previously thought once the inflation monster has been tamed.
But, inflation is still the “only” thing that matters, with only in quotation marks because, yes, of course, company specific news and other potentially shocking macro news will matter too.
But, the boot on the throat of the market is the Fed, and the Fed only cares about inflation – recession and banking failures be damned.
- We note that the company deliver substantial slowing down of the core business due to the macro environment and 2024 is seen as yet slower.
- The company’s new CEO needs to prove himself.
- There are many other risks, including those that are nearly impossible to point out until they arise and suddenly, we say, ‘oh, yeah, that could happen;’ like a war.
Conclusion
It’s finding the technology gems like DigitalOcean before they are household names, that can turn into the 'next Google,’ or 'next Apple,’ where we have to get ahead of the curve. This is what CML Pro does.
The precious few thematic top picks, research dossiers, and alerts are available for a limited time at a 30% discount.
Thanks for reading, friends.
The author is long DOCN at the time of this writing.
Please read the legal disclaimers below and as always, remember, we are not making a recommendation or soliciting a sale or purchase of any security ever. We are not licensed to do so, and we wouldn’t do it even if we were. We’re sharing my opinions, and provide you the power to be knowledgeable to make your own decisions.
Legal
The information contained on this site is provided for general informational purposes, as a convenience to the readers. The materials are not a substitute for obtaining professional advice from a qualified person, firm or corporation. Consult the appropriate professional advisor for more complete and current information. Capital Market Laboratories (“The Company”) does not engage in rendering any legal or professional services by placing these general informational materials on this website.
The Company specifically disclaims any liability, whether based in contract, tort, strict liability or otherwise, for any direct, indirect, incidental, consequential, or special damages arising out of or in any way connected with access to or use of the site, even if we have been advised of the possibility of such damages, including liability in connection with mistakes or omissions in, or delays in transmission of, information to or from the user, interruptions in telecommunications connections to the site or viruses.
The Company makes no representations or warranties about the accuracy or completeness of the information contained on this website. Any links provided to other server sites are offered as a matter of convenience and in no way are meant to imply that The Company endorses, sponsors, promotes or is affiliated with the owners of or participants in those sites, or endorse any information contained on those sites, unless expressly stated.
