Q3 2023 DigitalOcean Holdings Inc Earnings Call
Participants
Matthew Steinfort; CFO; DigitalOcean Holdings, Inc.
Rob Bradley; VP of IR; DigitalOcean Holdings, Inc.
Yancey L. Spruill; CEO & Director; DigitalOcean Holdings, Inc.
Jaiden Rajiv Patel; Analyst; JPMorgan Chase & Co, Research Division
Jason Noah Ader; Partner & Co-Group Head of Technology, Media and Communications; William Blair & Company L.L.C., Research Division
Joshua Phillip Baer; Equity Analyst; Morgan Stanley, Research Division
Matthew Ryan Calitri; Research Analyst; Needham & Company, LLC, Research Division
Patrick D. Walravens; MD, Director of Technology Research & Equity Research Analyst; JMP Securities LLC, Research Division
Quinton Amedeo Gabrielli; Research Analyst; Piper Sandler & Co., Research Division
Raimo Lenschow; MD & Analyst; Barclays Bank PLC, Research Division
Unidentified Analyst
William Mandl; Research Analyst; KeyBanc Capital Markets Inc., Research Division
William Kingsley Crane; Senior Analyst; Canaccord Genuity Corp., Research Division
Presentation
Operator
Hello, good afternoon. My name is Jeremy, and I will be your conference operator today. At this time, I would like to welcome everyone to the DigitalOcean Q3 2023 Earnings Conference Call. (Operator Instructions)
I would now like to turn the call over to Rob Bradley, Vice President of Investor Relations.
Rob Bradley
Thank you, Jeremy, and welcome, everyone, to DigitalOcean's Third Quarter 2023 Earnings Call. Joining me today is Yancey Spruill, our Chief Executive Officer; and Matt Steinfort, our Chief Financial Officer.
Before we get started, I want to cover our safe harbor statement. During this conference call, we will be making forward-looking statements, including our financial outlook for the fourth quarter and full year as well as statements about goals and business outlook, industry trends, market opportunities and expectations for future financial performance and similar items. All of these statements are subject to risks, uncertainties and assumptions. You can review more information about these in the Risk Factors section of our filings with the SEC. We remind everyone that our actual results may differ, and we undertake no obligation to revise or update any forward-looking statements.
Finally, we will be discussing non-GAAP financial measures on our call, and reconciliations between our GAAP and non-GAAP financial results can be found in our earnings press release, which was issued earlier this afternoon and in the investor presentation, which can be found on our investor website.
With that, let me turn the call over to our CEO, Yancey Spruill. Yancey?
Yancey L. Spruill
Thanks, Rob. Good afternoon, and thank you all for joining us today. I'm pleased to share our solid Q3 results, representing another quarter with an attractive balance of growth and profitability.
This afternoon, I'll share an update on the search for our next CEO, some broader macro observations that have been informed by customer and partner interactions and briefly recap some of the exciting product announcements that we've made recently. I will also highlight a recent customer win that gives us enthusiasm for our product trajectory before turning the call over to Matt to go over the financial results and updated financial outlook.
On the CEO transition front, the search process is well underway, and the Board is making good progress in our recruitment of a new CEO. There has been very strong interest from numerous leading technology executives, and we are confident that we will hire an outstanding executive with extensive cloud infrastructure expertise.
The Board of Directors is deeply engaged with several candidates, and we will continue to move these candidates through the interview process as the search progresses. It is still too early to communicate a specific time line for the hiring of a new CEO, but the Board is working with focus and deliberate speed to fill this critical role.
Third quarter results were encouraging as the business delivered solid performance from both the top and bottom line perspective. Revenue grew 16% year-over-year to $177 million in our first quarter having lapped the Q3 2022 pricing actions. And as we indicated, we have started to see at the tail end of Q2, we continue to see positive signs through Q3 and in early Q4 that there are abating trends from the growth headwinds from our cohort, that we have seen over the past 12 to 18 months.
Our margin profile for the core DigitalOcean business was very strong for the quarter and enabled us to absorb the incremental cost from our Paperspace acquisition while remaining within our targeted range of adjusted EBITDA and free cash flow margins. The combination of a stabilizing revenue growth rate and the efficiency we have created in our core business, is allowing us to deliver robust free cash flow margins.
As we look to accelerate our top line growth rate, we continue to build distinct revenue growth drivers through product and infrastructure investments. A big portion of our focus over the past several months has been on our ongoing efforts to deliver on our product road map and continue to add critical capabilities that enable our customers to build and scale their own businesses.
Through numerous interactions with customers, prospects and partners, we continue to fine-tune and evolve how we can best serve customers as they navigate the complexities of the cloud markets. The consistent sentiment from our customers is that simplicity is foundational to our value proposition and relying on DigitalOcean to remove the complexities of the cloud is a productivity multiplier for them.
We are focused on enhancing our platform's performance, reliability, security and scalability, which enables us to retain and grow the spend of our customers and positions us to win new customers.
Over the past several months, we have delivered an increase in velocity of product releases and announcements as we focus on meeting our customers' evolving needs. In September, we launched Managed Kafka a new fully managed database service, Managed Kafka is critical tool for businesses whose model involves significant data streaming. However, for small and medium business customers, it often comes with increased complexity.
With our new fully Managed Kafka as a service offering, companies can focus on their development environment and not be slowed down by complex processes. Customers have provided strong positive feedback about Managed Kafka citing that it is allowing them to focus their time and resources on their customer-facing activities, referring to their increased productivity as a game changer.
In September, we also introduced premium general purpose Droplets extending the premium features of enhanced CPU, memory and storage to offer improved performance to a broader way of cloud-native applications. Our first-generation premium dedicated Droplets saw broad adoption and drove strong ARPU growth when launched in the beginning of 2021 as customers expanded their computing capabilities and migrated certain workloads to this premium compute offering. Now we've taken it one step further and expanded the opportunity for more customers to use these enhanced features with this new premium general-purpose offering.
And just a few weeks ago, we introduced scalable storage for Postgres and MySQL managed databases to help customers better utilize their cloud spend by scaling to meet their storage requirements without needing to similarly scale their compute and memory configurations. This introduction is a deliberate effort to better enable customer productivity on our platform.
By introducing scalable storage, we are better serving our customers as they can monitor and optimize their utilization to get the best return on their investment.
On the Paperspace front, we continue to be very excited by the addition of their AI/ML capabilities and the expanded market opportunity that this business creates for DigitalOcean. While we are still only a few months into the integration process, the demand for Paperspace services has been very strong. And we continue to learn more about the market and how customers are leveraging our capabilities to develop and grow their businesses.
To give you a sense for one such use case, I'll provide an overview of one of the exciting customers we recently added to the Paperspace platform. Company is Nomic, which was founded in 2022 to improve the explainability and accessibility of artificial intelligence. Today, Nomic has released 2 products: an open-source AI model, GPT4ALL for all, which is the third fastest growing repository in GitHub history. And Atlas, a tool that allows users to visualize unstructured data sets used to build large language models.
Nomic, selected DigitalOcean excess high-performance compute, along with our intuitive software, customer support and reliability. Their co-founder was quoted as saying, "Our team loves Paperspace. It's far more intuitive than other compute providers. It allows us to spend less time managing infrastructure and more time building great products for our customers." Their testimony to DigitalOcean's combination of simplicity, reliability and support along with the current demand environment that we see excite us for the opportunity ahead.
The RFPs that we are seeing span across multiple verticals for multiple applications and demonstrate the opportunity for new customers to join our platform to not only build their AI applications but also to scale their products while utilizing the breadth of our IaaS and PaaS capabilities.
In summary, we're making good progress with our efforts to bring on a new CEO. And during this transition, our business is seeing stabilizing revenue trends while continuing to deliver significant free cash flow. We are seeing very encouraging signals that our top line growth rate is stabilizing relative to the last 6 quarters of deceleration.
We continue to work to reaccelerate our growth rate through targeted product and go-to-market initiatives. The improved operating leverage we have created in the business over the past year is enabling us to make strategic investments while still delivering compelling free cash flow margins. We remain excited about our near and long-term potential in the large outstanding $100 billion addressable market for SMB cloud infrastructure in which we compete.
As we approach the end of the year, I'd like to thank each and every member of the DigitalOcean team for their commitment to our customers and for delivering these solid results.
With that, I'd like to turn it over to Matt to provide details on our financial outlook results and our outlook for the balance of the year.
Matthew Steinfort
Thanks, Yancey, and welcome to all of you who are joining us on today's earnings call as we review our solid Q3 results.
In Q3, we continued to execute our strategy of accelerating the achievement of our long-term margin targets in the core DigitalOcean business while positioning the company for accelerated future growth through both organic and inorganic investment in our platform despite the ongoing macro headwinds. We have driven these margin improvements by achieving the savings that we had identified at the beginning of this year, which included improvements in our gross margin as we grew into capacity investments made in 2022 and the achievement of identified savings in both headcount and nonheadcount-related expenses.
As we have continued to execute on these savings, we have increased our overall profitability and free cash flow margin significantly. Allowing us the flexibility to make targeted investments in growth.
During the first 3 quarters, we have continued to invest in organic growth by our product road map, as Yancey described. And in July, we invested inorganically in the acquisition of Paperspace, which meaningfully increases our total addressable market.
With our strong balance sheet, our continued focus on improving operating leverage and our commitment to share repurchases, we have made material progress driving earnings per share, increasing 22% year-over-year.
As we approach 2024, our strategy remains the same. We will continue to drive operating leverage while balancing investment across organic and inorganic growth opportunities and share repurchases.
With that context in mind, I will review the highlights of the third quarter. Revenue in the third quarter was $177.1 million, which was an increase of 16% year-over-year and 4% quarter-over-quarter. In our first full quarter that laps the pricing increase we implemented in early Q3 2022.
Contributing to this growth was Cloudways, which grew 11% quarter order and the addition of Paperspace, the AI platform we acquired in early July, which contributed $3 million to our results for Q3 and which surpassed the $1 million in monthly revenue mark in September.
Profitability was very strong as we delivered $75.8 million of adjusted EBITDA, a 43% margin for the second consecutive quarter. Adjusted EBITDA margin improvements are the result of improving gross margins as we grew into the incremental data center and bandwidth capacity investments that we had made in late 2022 and the ongoing benefit of the efficiency improvements that we had identified in Q1 of this year and have been realizing throughout the year.
Free cash flow was also a highlight in the quarter as we generated $56 million, which was 32% of revenue. This 500 basis point improvement from Q2 was a result of both lower capital expense and working capital timing. Given the working capital timing dynamic, coupled with our anticipated near-term Paperspace investments, we expect free cash flow margin in Q4 to be lower than Q3 levels. While free cash flow margins can be lumpy quarter-to-quarter, we continue to be confident in our full year free cash flow margin expectations.
Finally, non-GAAP earnings per share was $0.44, which is a 22% year-over-year increase as we have increased our profit margins while having simultaneously reduced our shares outstanding through our systematic share buyback program.
As expected, net dollar retention declined in Q3 to the mid-90s, coming in at 96% as we lapped the price increase that was implemented in July of 2022. Fortunately, despite the difficult year-over-year comp in Q3, we did continue to see evidence of the ongoing stabilization of market demand that had started to show signs of bottoming in Q2.
As it has been for most of 2022, churn remained stable at historical levels, around 11% to 12% for each of the months in Q3, which is consistent with historical churn levels in early 2022 prior to the market slowdown.
Contraction, which has historically been in the 12% range in early 2022, showed improvement as we progress through the quarter, ending at 15% after starting the quarter at over 16%. We also saw positive signs and expansion in Q3, which was the final metric we said we needed to see for stabilization as it had continued to decline in Q2, albeit at a decelerating rate.
In Q3, expansion stopped declining for the first time in over a year and was fairly consistent at 23%. While we have not yet seen a meaningful improvement in NDR, we are encouraged by the relative stability of the key component metrics in Q3. We expect NDR to improve in Q4 and early 2024, driven in part by the continued strength of the Cloudways business and more favorable year-over-year comparison.
From a customer perspective, our durable customer acquisition and graduation model remains solid despite the challenging growth environment. We graduated 4,000 builders and scalers on our platform in the quarter, and we now have more than 154,000 on the DigitalOcean platform. These customers who spend more than $50 per month with DigitalOcean continue to represent 86% of our overall revenue and remain a key focus of our product, sales and customer success investments.
Average monthly revenue per customer, or ARPU, was $92.06, which was an increase of 6% year-over-year, which, like NDR, faced a difficult year-over-year comp as we lapped the price increase from last July.
Before providing guidance for Q4 and for the full year 2023, it is important to understand where we are in the 2024 planning process. We are working diligently to finalize our plan and budget for 2024 with a focus on delivering double-digit growth with healthy profit and free cash flow margins.
Our strategy will remain the same in 2024. We will continue to drive operating leverage in the core DigitalOcean business while investing to take advantage of the compelling market opportunity we have with Paperspace.
As we shared last quarter, despite the continued market pressures, we believe we have a solid foundation for double-digit top line growth for 2024 with our steady self-serve funnel generating around 5% growth, Cloudways generating around 3% growth and Paperspace producing at least 3% growth. Key to growing faster than this, we'll be getting NDR above 100% at some point in 2024, which we are working aggressively to accomplish.
We look forward to providing more specifics on our plan -- our next earnings call in February when we report our Q4 and full year results.
In terms of financial guidance for Q4 of 2023, we are targeting revenue to be $178 million. For the fourth quarter, we expect adjusted EBITDA margins to be in the range of 36% to 37%. And non-GAAP earnings per share to be $0.36 to $0.37 based on approximately 100 million to 101 million in weighted average fully diluted shares outstanding. For the full year, we are targeting revenue to be $690 million.
Given the strong performance driving margin improvement in our core DigitalOcean business, we continue to project adjusted EBITDA margins to be in the range of 38% to 39% for the full year despite our increased investment in our Paperspace AI/ML business. We project non-GAAP earnings per share to be in the range of $1.52 to $1.54, with weighted average fully diluted shares outstanding for 2023 of 102 million to 103 million.
And finally, given the strong progress we have made on improving margins in our core DigitalOcean business for the full year 2023, free cash flow will be 21% to 22% of revenue, consistent with our initial February guide despite the incremental investment we are making into our Paperspace business.
That concludes our formal remarks, and we'll now open the call up to Q&A.
Question and Answer Session
Operator
(Operator Instructions) Our first question comes from the line of Raimo Lenschow from Barclays.
Raimo Lenschow
And congrats on a great quarter, and it's nice to see the stabilization and improvement. My first question is on Paperspace. Can you kind of talk a little bit about what you're seeing in the customer base in terms of how you see the adoption there? Because like at the moment for us, AI is a lot of like big companies and copilots and stuff like that. Can you talk a little bit about like practical use cases, how like SMBs are using this? And what are you seeing there in terms of excitement about that? And then I had one follow-up on financials.
Yancey L. Spruill
Well, the customer base, I think, Nomic, that we provide an example is a good example of -- there are diverse array of use cases, verticals, people targeting, building large language models for specific use cases. Creating tools that enable people to use AI productively. These are tech-enabled businesses, I don't know that we would call them SMBs because I think they have aspirations to grow very rapidly, supporting a broad set of use cases.
But the good news is the value proposition that we offer and the complementary aspect of the Paperspace platform, which is built around simplicity is making it easy for people to come on to the platform. And I think that's been a distinguishing characteristic for us over the life of DigitalOcean and exciting to see that value proposition be weighted heavily as we move into the AI opportunity.
Raimo Lenschow
Okay. Perfect. And then a question -- quick question on the free cash flow. So obviously, a very strong performance this quarter. And you called out CapEx and working capital a little bit as factors there. But how do you think about that -- those 2 factors that you mentioned there in terms of like, in theory, some of that could be driven forward? Or was it just timing?
Yancey L. Spruill
Thanks, Raimo. We are still staying with our full year guide for this year of 21% to 22%, and that's -- we will have some of the working capital catch up in the fourth quarter. It's -- looking at free cash flow margin on a quarterly basis, it's tough just because it's a lot influenced by timing of payments. So that will normalize in the fourth quarter. And then in the fourth quarter, also given some of the equipment purchases we made when we acquired Paperspace that are coming in, we expect that to hit partially in the fourth quarter as well.
So again, the full year guide of 21% to 22% is still the right way to be thinking about it in the -- it's great that we have the margins size we do this quarter, which is evidence, again, of the core DO business. As we said, we were going to drive free cash flow margins into the high 20s and approach 30 by the end of the year, and we've done that. And we're using that as a means of funding the growth opportunity that we have to drive more revenue in the Paperspace business and yet we're still able to come in at 21% to 22% for the year. So -- and we're really happy with the progress we've made.
Operator
All right. Our next question comes from the line of Mike Cikos from Needham & Co.
Matthew Ryan Calitri
This is Matt Calitri on for Mike Cikos over at Needham. I know you hit on this a little bit, but kind of coming from a different angle. Is there anything onetime or anything, I guess, that sticks out as far as just the timing of expenses that you would call out in the reduction to the implied 4Q EBITDA guide?
Matthew Steinfort
Timing of the expenses related to the EBITDA. The sequential decline in EBITDA. No, it's -- I mean this is -- again, it's -- with Paperspace being on board, that's clearly causing some -- a couple of basis point impact on the margins. It's just a different margin business given it's not at scale. And then as we invest to grow the business into next year, both the core DigitalOcean business, and the Paperspace business, it's -- there's nothing that's kind of onetime in nature.
It's again, the guide for EBITDA of what is 38% to 39% for the year is the right way to think about it. I don't get caught up in the margin kind of the quarter-over-quarter margin fluctuations. It's -- you got to think about it over a longer period.
Matthew Ryan Calitri
Got you. That's helpful. And then in any way, is the customer base showing any potential signs of greater caution given news on the broader macro or digital native pressures in the SMB space?
Yancey L. Spruill
I think what we said is we're pleased with -- churn has been relatively stable now for several quarters as we move through the year. We saw improvement in contraction as we move through Q3 and that continues at a better level than it was 6 months ago. And we're seeing stabilizing signs and expansion. And so I think those are the biggest telltale signs for us as it relates to what our customers and new customers and existing customers are seeing that things seem to be stabilizing.
Operator
Our next question comes from the line of Josh Baer from Morgan Stanley.
Joshua Phillip Baer
Congrats on the good quarter. Was hoping you could break out some of the metrics around Paperspace, the contribution to ARR and just how to think about the contribution to the different customer cohorts?
Yancey L. Spruill
So from an ARR standpoint, the -- the -- I think we said it was 2.8 million in the quarter. It's going to be just a little over 3, we're going to end up around 6 for the year. So from an ARR standpoint, it's going to be north of 12 because that's where the quarter is ending. But from a customer account standpoint, it added about, I want to say, 12,000 customers in total to the customer base. And again, the average revenue per customer there is a lot bigger. So it's closer to -- average is 900-ish. So they tended to be added to the scalers segment and the builder segment.
Joshua Phillip Baer
Great. And apologies if I missed this. Just looking at CapEx as a percent of revenue is pretty low. I understand that can be lumpy and it is a priority to invest in Paperspace. But how should we think about the trajectory of that CapEx spend? And -- and is it a good leading indicator sort of for your demand and top line growth?
Yancey L. Spruill
No, I don't think it's a good leading indicator of the demand because it is lumpy. We were originally tracking to be about 15% for the year. What we said with Paperspaces will be closer to 17% in CapEx as percentage of revenue for the year. And some of that is just -- it's lumpy. I mean we ordered gear right out of the gate when we bought Paperspace and we increased that amount because we're seeing a lot of really good demand and just the supply chain around GPUs, it's unpredictable. And so we didn't get everything that we ordered in Q3, in Q3, and we're going to get -- start to get some of that in Q4 and then we'll start to get some more of it in early next year.
So it really is working capital and the timing of that CapEx purchase, but it's not -- that's why we're saying for the full year, we're still sticking to the 21% to 22% free cash flow and still the target of around 17% CapEx as a percentage of revenue for the year.
Operator
Our next question comes from the line of Patrick Walravens from JMP.
Patrick D. Walravens
Great. And it's nice to see the business stabilizing. So my question is on with Paperspace, how are your GPU cloud data centers different than your traditional CPU cloud data centers that you had with DigitalOcean, and what's that GPU cloud data center footprint look like today? And where are you going to take it? Obviously, it's been a big issue in the industry lately.
Yancey L. Spruill
Well, GPU servers have heat -- different heat consumption, much higher heat consumption. So the physical footprint is less dense than you might with standard compute. I think that's one of the principal differences. We're still evaluating, as Matt mentioned, in the 2024 planning process around what the level of investment pacing and that may include how to think about evolving the data center infrastructure for both companies for the combined company and incorporate the fact that it's going to be a mix of GPU and standard compute CPU going forward.
I don't think any decisions have been made on that. We've been fortunate some of the Paperspace data center footprints are very close in proximity, including one particular case in the same building. And we'll see how we evolve that as the demand.
We're learning a lot. Obviously, the demand is incredibly strong. And we're seeing what our -- where we're playing in that market and how we're going to grow it and how we're going to need to add to the capacity footprint to support that growth, not just in terms of buying servers, but the physical input and are looking to make decisions on that as we get into next year.
Operator
Our next question comes from the line of Jim Fish from Piper Sandler.
Quinton Amedeo Gabrielli
This is Quinton on for Jim Fish. First, we've seen significant product launches in the kind of, call it, noncore compute side of DigitalOcean with storage and Managed Kafka kind of the most recent ones. Can you talk about how the team is thinking about bundling or potentially packaging those noncore compute with maybe compute and Paperspace coming on and how we kind of can adapt to some of the go-to-market motions with that?
Yancey L. Spruill
Well, we've incorporated some of the -- for combined customers, some of the feature functionality, security and other tools that make it easy and seamless across the board to use the AI and the standard products. I think bundling some of the items that I highlighted, scalable storage, the Premium Droplets reflect this concept of bundling, which we've been talking about for a year to 2 years with investors that is a real opportunity for us to -- as we learn more about customer use cases in industry verticals, the decoupling of compute from network bandwidth and storage is a key theme that we keep hearing and security and other attributes of the product set.
And so I think having a strong foundation now with standard and AI compute strong foundation in bandwidth and security and other capabilities in storage, it's giving us the building blocks to be able to do bundling and unbundling to better meet our customers' needs as we -- as we highlighted today with the recent launch of scalable storage. If the use case needs more storage, but doesn't need more compute, let's give them that option, and that's a more efficient growth path for customers.
So I think that's going to be a continuing theme. And as we get into next year, we'll have more to say about sort of more integrated product bundling as it relates to both AI and standard compute. But that's certainly the process of bundling in packaging is a pretty significant opportunity for us.
Quinton Amedeo Gabrielli
Makes sense. And then, obviously, you talked a little bit about wanting -- in the search for a new CEO, you're wanting a leader that has the kind of core cloud experience. But as you think about kind of the skills of the background, are you looking for someone with more technical background? Or are you looking for an operational or sales kind of leader. Maybe any sort of color you can give on the fact like underlying skills would be helpful.
Yancey L. Spruill
Yes. The Board is very engaged right now in finding a CEO. As we've described in the prepared remarks. It's an active process. It's a deliberate process, but it's moving with as fast as possible without sacrificing the screens needed to canvass the market. There's not a time line to announce anything and we sort of spoke to -- we're clearly looking for somebody with cloud experience, but I'm not going to speak any more to the process or to the candidate pool beyond that.
Operator
Our next question comes from the line of Kingsley Crane from Canaccord Genuity.
William Kingsley Crane
We noticed on Cloudways website that no longer offers infrastructure services from SMB cloud competitors like Vultr only including options for public clouds like AWS and GCP. So are you actively converting customers over DigitalOcean infrastructure? Is this primarily a new business dynamic?
Matthew Steinfort
It's mostly a new business dynamic. We still have such to support customers on each of those platforms. But from a go-forward basis, we still offer the options to go multi-cloud through the hyperscalers or through our platform. But we continue to support partners that drive opportunities through some of those other providers. And clearly, we have a healthy installed base of customers that are leveraging other providers as well.
Yancey L. Spruill
That's a change that was made earlier this year in Q2.
William Kingsley Crane
Okay. Very helpful. And the second one would be, how is the momentum behind the Hatch accelerator program? How important of the customer ramping mechanism has it become for you? And is there an opportunity to include Paperspace as part of the program?
Yancey L. Spruill
There's certainly an opportunity to include Paperspace as part of the program and the Hatch program is something that's been a part of the company for a long time, and -- and as we look for ways to accelerate the business and drive more adoption through the self-serve funnel and it's certainly one of the areas that we continue to invest in. But it's certainly an opportunity to bring in both Cloudways and Paperspace into that kind of a program.
Operator
Our next question comes from the line of Jason Ader from William Blair.
Jason Noah Ader
I just wanted to ask about the go-to-market strategy and how that...
Rob Bradley
Jason, we can't really hear you. Can you speak a little louder?
Jason Noah Ader
Is that any better?
Rob Bradley
Much.
Jason Noah Ader
Okay. Sorry. I was asking about the go-to-market strategy and just as you look out, especially over the next couple of years, how would you frame sort of where you've been, what you're doing now in terms of maybe some changes and then where you want to go just from an overall go-to-market standpoint? Because I know you've been very much sort of kind of bottom-up driven historically?
Yancey L. Spruill
That's a good question, Jason. The self-serve funnel has been the primary vehicle for adding new customers as you think about net new logos. Clearly, the NDR drives revenue growth from the cohort. But in terms of driving new go-to-market around acquiring new customers. It's been primarily through the self-serve funnel. And as we said, it delivers a very healthy, kind of, call it, in the $33 million to $35 million range of new revenue.
We've invested over the past, call it, several years to try to expand the go-to-market channels to include more of a direct sales motion where we would target customers that are on some of the larger clouds and maybe at the smaller end of their customer base that where the improved customer service and more of direct exposure to the company and ability to talk with customer service people would be attractive, also the economic it would be a cost savings and where the simplicity is valuable to them.
So that's something that we've invested in. I'd say that's still very early innings. And one of the things that we've learned over the course of this year and through the slowdown or some of the product enhancements and capabilities that you would need in order to attract the customer off of one of the larger hyperscalers. And so that's clearly feeding into our product road map and the capabilities that we're working to develop right now.
There's also the opportunity to expand the partner channel. We have a lot of partners on our platform today, in web agencies and developers building websites for other the companies. But historically, we've treated them very much just like a customer. So they come on and they have their own requirements and we meet those requirements, but we didn't treat them really differently as a partner, and we're investing to expand the capability there because that's a great source of potentially us reaching parts of the market. We might not otherwise reach, you wouldn't self identify and come to our platform.
So we're -- we think as you think about it over the next several years, that the ratio of new sales to self-serve should be -- it would be great to get it to be 50-50. But we've got a bit of ways to go to get there. We're still very nascent, I'd say, in the channels outside of the self-serve motion that's been the foundation of the company's growth.
Jason Noah Ader
Great. Just one quick follow-up on that. It would be helpful, I think, for investors as you go forward just to provide some proof points on some of those newer channels and maybe anecdotes of customers that you've taken away from the hyperscalers, et cetera, just because I think that would be helpful for everyone to kind of just understand how that motion is working.
Yancey L. Spruill
Great. It's a good suggestion, Jason. Thank you.
Operator
Our next question comes from the line of Jaiden Patel from JPMorgan.
Jaiden Rajiv Patel
I'm on for Pinjalim Bora. Any early feedback on the managed cost of service so maybe some of the use cases there. And do you think it could be a tailwind to growth next year?
Yancey L. Spruill
Well, as I mentioned, the feedback has been very good, very positive. We have lots of businesses on the platform that are big data streaming, whether it's e-commerce or gaming or other businesses where there's a lot of data and a lot of data streaming, which is the use case or a principal use case that's highly aligned and very productive or more productive on Kafka than other services. So it's a natural fit for our customer base.
And the feedback has been great because, again, others have Kafka, but it's not built with simplicity as we cited in the in the script, the feedback has been very positive as it's a productivity enabler for our customers because they don't have to spend as much time managing Kafka, which other solutions are more complex and that's giving them more productivity to focus on their customer-facing activity.
So we absolutely -- it won't be material or significant this quarter as the adoption ramps was launched late last quarter or just recently, but we will see that becoming a more meaningful contributor as we move through 2024.
Jaiden Rajiv Patel
Okay. And then on Paperspace, it seems like it's doing well. Are you still baking in less than $5 million in contribution for Paperspace for the second half of this year?
Yancey L. Spruill
I think what we've said is it's -- I mean, we disclosed the actual revenue in the -- for the third quarter. And we said that the run rate is 1 million in September. So just extrapolate that, it puts us about 6.
Operator
Our next question comes from the line of Wamsi Mohan from Bank of America.
Unidentified Analyst
This is (inaudible). We said taking the question on behalf of Wamsi. I was wondering if you could talk more about the underlying drivers in terms of the mix, in terms of the ARPU growth, especially for the upsell and the attach? And how are you thinking about like the trajectory for the ARPU growth?
Yancey L. Spruill
So the ARPU growth, again, year-over-year, we said was 6%, which was low relative to historical based on the difficult comp on a price changes. But if you look at the products that we announced during this year, you talked to a number of them, but we had launched a number of them earlier this year, and we have a road map that is designed to kind of appeal to the larger set of our customers, products and capabilities like integrated -- sorry, identity and access management IM or role-based access control virtual private clouds, certain capabilities that really resonate with some of the customers that have gotten to more scale on our platform. That's really where the focus is in our development in the near term.
That enables us not only to continue to kind of take a big share of our own customers' workloads as they grow, so they don't have to move off to the cloud -- other cloud providers. it also enables us, as I mentioned earlier, to accelerate the direct sales. If you're trying to poach customers, longer customers of the larger providers, these are some of the same capabilities you need. So that's really where the focus is on driving ARPU is on being able to provide the capabilities that enable our customers to continue to scale on our platform as we better meet and deliver on their requirements.
Unidentified Analyst
If I could follow up, could you guys talk more about like what's the health of the SMB customer at the moment? And if there's any improvement in the usage trend?
Yancey L. Spruill
Yes. We talked about this earlier. If you look at it from a macro standpoint, you think of what's been going on in the market, our customers have not been growing as fast their own businesses as they were historically. And we see that through the expansion that they get with us. A year ago, we were seeing expansion from our customers that were growing in the high 30s. And we've seen that be declining month-over-month for well over a year down into the 22%, 23% range, but that's stabilized. And so our customers -- the decline in their growth appears to be stabilizing in a similar vein because we're seeing that stabilization expansion.
What we also saw with the customer base over the beginning of this year was that they were optimizing their spend and that they were reducing the amount that they were investing in software, and that causes contraction. And while the contraction is still elevated relative to historical, it's around 15% right now. Historically, it was pretty consistently around 12%. That has been stable for the past 2 quarters and actually has started to improve.
So we think that just like we saw the downturn perhaps earlier than others because of the usage-based nature of our platform we don't have renewal cycles and when a customer needs to spend less with us, they just spend less. And so we saw the impact early and we'll likely see the return early because of the nature of the business model.
Unidentified Analyst
Congrats on the quarter.
Operator
And our final question comes from the line of Michael Turits from KeyBanc.
William Mandl
This is Billy on for Michael. Just a quick follow-up to that last question. Good to see some of the leading metrics kind of stabilizing in the business. Anything you'd call out that was notable in different verticals or customer segments?
Yancey L. Spruill
No. Actually, that's something that we spent a lot of time assessing and the nice thing, one of the nice things about DigitalOcean customer base is it's so diverse. 70% of the revenue comes from outside the U.S. There's no single vertical, there's no single use case, there's no single region that drives a disproportionate amount of our business. We really track kind of broad SMB and startup and developer trends. And so the recovery and the stabilization has been has been pretty consistent across all of them. We don't see kind of laggards and we don't see any that have kind of raced ahead. It's again -- part of the appeal is it's a very, very diverse distributed customer base.
Thank you all for joining us. We're excited to report on a very solid quarter with a lot of encouraging trends, and I appreciate your time this evening, and hope you have a good rest of the day.
Operator
That concludes today's call. Have a pleasant day.
