Datadog (DDOG.US) Q3 Conference Call: Traditional Customers Grow Five Times in a Row, AI Smart Reshaping Software Operation and Maintenance, AI Applications, Major Customers Renew Their Contracts Early
The Zhitong Finance App learned that the primary incremental information conveyed by the cloud monitoring and analysis software company Datadog (DDOG.US)'s third quarter 2026 results conference call can be described as the growth momentum is spreading from a small number of large AI customers to a wider range of enterprise customers. The revenue growth rate of non-AI customers increased for the fifth consecutive quarter, indicating that the demand brought about by AI came not only from the expansion of model companies themselves, but also from traditional enterprises using AI to write more software and accelerate application modernization, thereby increasing the demand for continuous integration, testing, deployment, and observability tools.
At the same time, the integration of the Datadog software ecosystem is increasing the initial contract scale for new customers: customers such as US federal defense agencies and major Japanese banks have used various products since the beginning of cooperation, and management clearly stated that these newly signed customers have not fully increased usage, and their revenue contributions have yet to be released.
The analyst Q&A session can be described as further clarifying the three factors behind the fourth quarter guidance: large customers receive a more favorable unit price after early contract renewal, the largest customers are only included in the guidelines according to the minimum spending commitment, and the seasonal impact of year-end holidays and code freezes. Early renewal was accompanied by higher committed spending and wider product adoption, and short-term unit price adjustments coincided with long-term customer expansion. Management also made it clear that these early renewals are mainly related to the growth of the customer's own business, and are not directly related to the company cutting expenses to control the word budget. The actual consumption of the largest customers in the third quarter was significantly higher than the minimum commitment, but the company did not speculate on their future spending based on this; other customers made conservative cuts in line with recent growth trends, and not all of them were included in the minimum commitment. The official 2027 guidelines will be announced during the year-end conference call.
In terms of products and business models, Bits AI has become the fastest product in the company's history. The billing method has changed from charging separately for investigation tasks to AI credits covering various usage scenarios. Management emphasized that the relevant pricing will remain close to the cost of inference and pass on the savings from improved model efficiency to customers to support more investigation and wider use of the platform. Self-developed site reliability engineering model Scout and enhanced learning and post-training capabilities are driving the company from providing monitoring tools to autonomous operation and maintenance. Its commercial value includes both direct revenue from Bits itself and driving further adoption of products such as logs, metrics, link tracking, security, and cost management.
Another key addition is that GPU monitoring and training observability are still in the early stages of commercialization. Management clearly stated that the current growth of infrastructure monitoring services is not directly driven by GPU monetization, and related products have yet to generate extensive revenue; as more companies train, customize, and post-train models themselves, this may create new demand space. At the same time, the actual demand for intelligent observability is spreading from AI native companies to traditional enterprises, and AI cost management benefits from the rapidly growing demand for tracking word expenses, usage effects, and return on investment. As a result, the company is simultaneously participating in the two growth paths of “using AI management software” and “managing AI itself”.
The following is the full text of the Datadog performance conference call (artificial intelligence tool assisted translation):
Conference host
Hello everyone, and thank you all for your patience. Welcome to Datadog's Q3 2026 results conference call. Currently, all participants are in listen-only mode. Once the speakers have made their presentations, we will be open for questions. To ask a question during the Q&A session, press the star key and 11 on the phone. You'll then hear an automated voice prompt letting you know you've raised your hand. To withdraw the question, press the star key and 11 again. Please note that today's conference call is being recorded.
I'll now leave the conference to the first speaker — AJ Lubitsch, Senior Vice President of Strategic Finance and Investor Relations. Please get started.
AJ Lubitsch (Senior Vice President, Strategic Finance and Investor Relations)
Thank you, Victor. Good morning everyone, and thank you for attending the conference to review Datadog's financial results for the third quarter of 2026 with us. We announced these results in a press release issued this morning. I was on the conference call today with Olivier Pomel, co-founder and CEO of Datadog, and David Obstler, chief financial officer of Datadog. First, I'd like to thank Yuka Broderick for her outstanding performance over the past five and a half years leading investor relations work, and she is about to take on a new role within Datadog.
I'm excited to be back in Investor Relations. I worked with many of those present during the company's 2019 initial public offering. In this conference call, we will be making forward-looking statements, including statements relating to our future financial performance, outlook for the fourth quarter and full year of 2026 and related explanations and assumptions, product capabilities, and our ability to seize market opportunities. “Expect,” “believe,” “continue,” “estimate,” “anticipate,” “intend,” “will,” and similar statements to identify forward-looking statements or other similar future expectations.
These statements reflect our views as of today and are subject to various risks and uncertainties that may cause material differences in actual results. For a discussion of significant risks and other important factors that may affect actual results, please refer to our 10-Q table for the quarter ended June 30, 2026. More information will be included in our upcoming Form 10-Q for the fiscal quarter ending September 30, 2026, as well as other documents filed with the U.S. Securities and Exchange Commission.
This information, as well as a replay of this conference call, can also be viewed in the Investor Relations section of our website. We'll be discussing non-GAAP financial measures. A reconciliation table between these indicators and the most comparable US GAAP financial indicators is included in the results press release and can be found at investors.datadoghq.com. Next, I'll leave the meeting to Olivier.
Olivier Pomel (Co-founder and CEO)
Thank you, AJ, and thank you all for attending the conference today to review our third quarter results with us. Let me start by talking about the business drivers for this quarter. I'm happy to tell you that we had strong business performance in the third quarter. All aspects of the business performed strongly, including various product lines and a diverse customer base. Revenue growth from non-AI customers accelerated for the fifth consecutive quarter. The year-on-year increase was in the high 20% range, about 1 percentage point higher than the previous quarter, and also higher than the level of about 20% in the same period last year.
Furthermore, the AI customer base remains our strong growth driver. Whether it's the number of customers we serve or the size of those customers, this customer base continues to grow and become more diverse. Overall, we continue to see healthy needs. From the most technology-forward-looking startups to the world's largest enterprises, customers of all kinds are using AI and modern technology to develop. We believe this is driving the further adoption of the cloud and Datadog platforms.
Regarding financial performance and key metrics for the third quarter, our revenue was $1.2 billion, up 35% year over year, above the upper limit of the performance guidance range. At the end of the third quarter, we had around 33,500 customers, up from around 32,000 a year ago. There are approximately 4,860 customers with annual recurring revenue of $100,000 or more, up from about 4,060 a year ago. These customers contribute approximately 91% of our annual recurring revenue. We generated $251 million in free cash flow, with a free cash flow rate of 21%.
Let's take a look at product adoption. Our platform strategy continues to gain market recognition. Currently, 61% of customers use four or more products, up from 54% a year ago; 41% use six or more products, up from 31% a year ago; 16% use ten or more products, up from 8% a year ago. As a result, we are increasing the number of customers and creating value for customers on more products through a unified platform. We're very proud of one milestone: the infrastructure monitoring product suite's annual recurring revenue now exceeds $2 billion, and year-over-year growth is in the middle of the 30% range. Infrastructure monitoring was our first product to market over a decade ago. Today, customers rely on it to fully understand their environments, including cloud infrastructure, local servers, containers, serverless applications, and GPU clusters. While we're excited to reach this milestone, we still feel like we're just getting started.
According to data from industry analysts, we rank first in the global observability market share, but our market share is still at about 10%. Next, let's talk about R&D. As AI changes the way customers develop and run software, we have been investing in and advancing work in two directions: empowering Datadog with AI, and using Datadog to serve AI. The first is empowering Datadog with AI. These AI products and features make the Datadog platform more practical for customers, and the degree of automation is constantly increasing. First, our flagship smart device, Bits AI, had a very strong start. Since launching Bits, it has become the fastest customer adoption product in our history. Bits Testing officially launched in October this year. With just one click, it can discover the app's key user processes and generate tests that cover those processes.
At the same time, our AI research team is making every effort to expand the capabilities boundaries of bits and platforms. Last month, we launched Scout, an inference model that we developed in-house and optimized specifically for site reliability engineering tasks. Scout is built using our in-house distributed reinforcement learning infrastructure and currently supports the Bits investigation function. This is exciting for two reasons. First, in terms of alarm investigation, it performs as well as the basic model, or even better, while the cost of inference is only a fraction of that. This is valuable because it allows Bits to investigate more alarms, intervene earlier, and conduct more in-depth investigations into incidents, becoming the customer's always-on first responder. Second, it also shows what results can be achieved by using our own production data, key user workflows, and internal tools to train the model.
Scout is just the beginning. By joining our Adaptive ML team in June, we're fully accelerating our work on reinforcement learning and post-model training. Over time, we expect that these studies will drive a leap forward in the capabilities of the entire platform, bringing us closer to the goal of autonomous operation and maintenance. Next, use Datadog to serve AI. These products provide end-to-end observability and security capabilities for the new AI technology stack. As AI moves from testing to production environments, customers now need to answer three key questions: How is it performing? Is it safe? What's the return on investment, or is it worth the investment? And since AI doesn't operate in isolation, they want these answers on the same platform that manages the rest of the tech stack.
In terms of performance, we cover the entire AI technology stack, from computing resources to agents. GPU monitoring now goes beyond hardware reliability. New features help customers maximize cluster utilization, increase training throughput, and reduce the cost of inference services per request. Since this year, the number of GPU devices we monitor has increased eight times that of the beginning of the year, and some cutting-edge AI lab customers are developing these features with us as design partners.
Looking further at the upper layers of the technology stack, agent observability products help teams monitor and evaluate agents in production environments, including response quality, latency, and token usage. Since this year, the number of link tracking clips sent to intelligent observability products has now increased nine times that of the beginning of the year. We are also already providing intelligent observability products on the Anthropic application market, and we are also the first partner for OpenAI's new enterprise application market. As a result, customers can now buy our products directly through the cutting-edge AI labs we have partnered with.
In terms of security, our AI Guard product can discover custom agents running within an organization and rely on the same set of real-time data collected to monitor these agents to protect them from alert injections and data leaks at run time. In terms of costs, businesses everywhere are coping with rapid growth in terms of vocabulary and AI spending. In addition to the three major cloud service providers, AI providers have now become the largest category of expenses that customers monitor in Datadog cost management products, CCM for short. Now, through the AI cost feature in CCM, customers can understand which teams, models, and agents are generating AI expenses and manage those expenses with budgets, forecasts, and anomaly alerts. Currently, more than one-third of CCM customers are tracking AI costs.
The number of customers doing this has increased more than 11 times that of the same period last year. Looking at the upper level of the technology stack again, based on CCM data, our smart device console can accurately show which agents are actually being used by whom, what tasks are performed, how much they cost, and most importantly, related results. As we continue to innovate around empowering Datadog with AI and using Datadog to serve AI, independent research institutes have begun to pay attention. Forrester named Datadog as a leader in its 2026 intelligent operation and maintenance platform evaluation report; in the August 2026 AI vendor competition report, Gartner called Datadog an enterprise that other companies in the observability platform field need to catch up with. We believe this reflects our leadership in observability and our ability to help customers manage the complexity of the AI era.
Let's talk about sales and marketing and take a look at a few deals our market development team signed during this very strong quarter. First, we signed an annualized seven-digit expansion contract with a leading software company. The customer has more than doubled its annual commitment amount, is putting AI agents into production environments, and will replace a standalone commercial tool with Datadog's agent observability products to monitor cost and quality. By using Flex Logs, they were also able to collect and correlate larger logs at a lower cost. In total, they will integrate three tools into Datadog and increase the number of Datadog products they use to 20, including on-call duty management products and Bits AI.
Next, we signed a five-year contract with a US federal defense agency for a seven-digit annualized amount of dollars. This customer runs thousands of cloud accounts across hundreds of task-holders. As they modernize, their reliance on human tools has left them facing fragmented visibility and security gaps. They chose Datadog as their enterprise-grade observability and security platform, and will first adopt nine products, including infrastructure monitoring and cloud security.
I want to congratulate the public sector team on this landmark customer breakthrough. This is directly due to our investment in the government sector and the federal Risk and Authorization Management Program high-level certification received earlier this year. We are also actively expanding our government-facing platform to support workloads up to impact level 5, or IL5. This is the highest level of certification for non-confidential data and is part of our long-term commitment to the US public sector. Next, we signed a seven-digit annualized contract with a leading home-based healthcare software company. As this customer continues to grow, fragmented tools slow incident response and hurt the customer experience. They will integrate four tools into Datadog, replace the original log and security vendors with Flex Logs and cloud security information and event management products, and rely on our sensitive data scanners to help them maintain compliance with the Health Insurance Portability and Accountability Act. The customer will first use nine Datadog products, including incident management and Bits AI, to resolve issues faster.
Next, we signed an expansion contract with a leading online entertainment company for an annualized seven-digit dollar amount. The customer experienced a surge in traffic during the peak season and scaled up effectively with Datadog. Their own AI agents are already using a model context protocol server and will now join Bits AI. The customer's use of Datadog products will increase to 23, including agent observability and cloud cost management. Next, we signed a seven-digit annualized contract with two of Japan's most important banks, which are modernizing their core financial systems. Both banks chose Datadog as a uniform standard to unify their understanding of cloud and local environments. Both customers are adopting a full-stack platform, covering all three pillars, as well as digital experience, security, software delivery, service management, and Bits AI, and moving towards AI-driven operations.
Next, we signed a six-digit annualized contract with a leading European AI lab. This customer's developers spent valuable time due to slow builds and unstable tests, which delayed launch in a competitive market. They chose Datadog's continuous integration visibility and test optimization products as an alternative to a commercial tool so their engineers could focus on delivering the product. Their engineers also use Bits AI to reliably build monitoring items and custom analysis. Next, we signed an extended contract with a leading AI programming company with an annualized incremental amount of seven digits, bringing the total annualized contract to eight digit dollars. The customer grew rapidly and completed the renewal ahead of schedule to keep up with the growing scale of the business. In this expansion, they will be adding Flex Logs and our new unlimited cardinality metrics feature to effectively control costs while data volumes grow. They are also one of the biggest users of our Model Context Protocol server. The number of Datadog products used by this customer will increase to 13.
Finally, we signed an expansion contract with the AI division of one of the world's largest technology companies for an annualized amount of eight digits. In just two quarters since they first signed up for this client, their annual commitment has nearly tripled. They are expanding the use of Datadog in large AI training workloads. They are also using application performance monitoring and Bits AI to further reduce friction in engineering work. While making progress with these customers, at the other end of the customer spectrum, we are also increasing our investment in the large number of AI-native companies that are emerging every day. By investing in the Datadog Startup Program, we're expanding our coverage to a large number of earlier-stage startups and developing deeper engineering partnerships with companies that have achieved explosive growth. This is truly an exciting time for working with startups.
Now, before I leave the conference to David to review the financial situation, I would like to say a few words about the long-term outlook. Our general view hasn't changed: digital transformation and cloud migration are the long-term growth drivers of our business. Today, AI is another major growth driver for us, as AI workloads enter production environments, which will drive more cloud resource consumption and greater use of data platforms. Today, we're already seeing this across our business: the top 15 AI leaders are all our customers.
From a broader perspective, we currently serve more than 850 AI customers, which was around 750 last quarter. Within the broader customer base, we're also seeing a significant increase in AI activity. Currently, about one-third of our customers use one or more of our AI products, including three-quarters of customers with annual recurring revenue of $100,000 or more. Many of these customers are just starting to use these products, so there are still huge opportunities for future expansion. We continue to see a rapid increase in related activity, and the number of calls to the Model Context Protocol tool has increased 50 times since the beginning of the year since the beginning of the year.
We are extremely excited about this direction of development. As AI writes and runs more customer software, customers are delivering more code faster on top of an ever-expanding portfolio of models, agents, and infrastructure. Of course, the complexity it brings is rapidly exceeding the ability of even the best engineering teams to track and master on their own. This is exactly the problem we're investing in solving. Using AI to empower Datadog to take on more of this kind of work; using Datadog to serve AI to help them run AI with confidence; and our AI research transforms data into models to achieve a higher degree of autonomous operation over time.
Because of this, we believe more than ever that we are ideally placed to help customers of all sizes and industries, and all types of users, whether humans or AI agents, transform, innovate, and create value through the adoption of AI and the cloud. Next, I'll leave the meeting to our Chief Financial Officer David.
David Obstler (Chief Financial Officer)
Thank you Olivier and good morning everyone. Our third-quarter revenue was $1,197 million, up 35% year over year. We continue to see strong growth in usage from existing customers, and a rapid increase in usage from new customers. Among the broad non-AI customer base, revenue growth accelerated to a high 20% range compared to the same period last year, an increase of about 1 percentage point over the previous quarter, and also higher than the level of about 20% in the same period last year. We've seen strong growth, with strong performance across different customer groups, spending levels, and industries.
After eliminating the largest customers, revenue growth accelerated again. As we communicated in August, our largest customer reduced usage in the third quarter but was still above their minimum commitment level. Meanwhile, AI customers continue to grow significantly faster than other businesses. This customer base of over 850 continues to grow and diversify, including 39 that pay us more than $1 million a year. We continue to acquire larger new customers and achieve faster growth, and the contribution of new customers to revenue growth is increasing. In the third quarter, new customers accounted for about 30% of our year-over-year increase in revenue, up from 25% in the same period last year.
Now let's talk about retention metrics. The net revenue retention rate for the past 12 months was slightly higher than 120%, which is similar to the previous quarter; the customer churn rate is still very low, and the gross revenue retention rate has stabilized in the middle to high 90% range. We believe this metric highlights the importance of our platform to our clients' critical businesses. The financial results are presented next. Invoicing for the quarter was US$1.16 billion, an increase of 30% over the previous year. Remaining performance obligations were US$3.85 billion, up 38% year on year; the year-on-year growth rate of remaining performance obligations in the current period was in the middle of the 30% range. The remaining performance obligations have been slightly extended compared to the same period last year. As a reminder, we still believe that revenue better reflects business trends compared to invoiced amounts or remaining performance obligations.
Let's review some of the key results in the income statement. Unless otherwise specified, all metrics are in a non-US GAAP caliber. We have provided a reconciliation statement between US GAAP and non-GAAP financial data in the results press release. Gross profit for the third quarter was US$972 million and gross margin was 81.2%; in comparison, gross margin for the previous quarter was 79.6%, compared to 81.2% in the same period last year. Operating expenses increased 25% year over year in the third quarter, compared to 26% in the previous quarter and 32% in the same period last year. Third-quarter operating profit was US$334 million, and operating margin was 28%; compared to 23% in the previous quarter and the same period last year.
Now let's look at the balance sheet and cash flow statement. At the end of the quarter, we held $5.2 billion in cash, cash equivalents, and marketable securities. Cash flow from operating activities for the quarter was $290 million. After accounting for capital expenses and capitalization software, free cash flow was $251 million, and the quarterly free cash flow rate was 21%. The timing of receipt and payment will vary; based on the past 12 months, the free cash flow rate is 26%.
Next up is our outlook for the fourth quarter and fiscal year 2026. First, our overall performance guidance philosophy remains the same. As a reminder, we are developing guidelines based on trends observed in recent months and making careful assumptions about these growth trends. For the largest customers, our fourth quarter guidance accrues revenue based on their minimum commitment amount. For the fourth quarter, we expect revenue of between $1.19 billion and $1.2 billion, up 25% to 26% year over year. After excluding the largest customers, the fourth-quarter revenue growth rate implied by our guidelines was in the low to medium range of 30%. Non-GAAP operating profit is estimated to be between US$280 million and US$290 million, with a corresponding operating margin of 24%. The fourth quarter included significant marketing investments, including the AWS re:Invent conference.
Based on a weighted average diluted share count of approximately 379 million shares, we expect non-GAAP net profit per share between $0.69 and $0.71. For fiscal year 2026, we expect revenue to be between $4.515 billion and $4.525 billion, up about 32% year over year; non-GAAP operating profit is expected to be between $1,095 billion and $1,105 billion, with a corresponding operating margin of 24%. Finally, based on the weighted average number of diluted shares of approximately 375 million shares, non-GAAP net profit per share is expected to be between $2.73 and $2.75.
Regarding the guidance, there are a few additional notes: We expect net interest and other income to be around $200 million for the 2026 fiscal year. Cash taxes are expected to be around $30 million to $40 million, and we will continue to apply a 21% non-GAAP rate for 2026 and beyond. Finally, we expect capital expenditure and capitalized software combined to account for 3% to 4% of revenue for the 2026 fiscal year. Overall, we were satisfied with the third quarter's performance, which combined strong growth with strong profitability. We continue to invest to seize long-term opportunities in the cloud and AI sector. I would like to thank my Datadog colleagues around the world for their efforts. Below, we're open to questions. Moderator, please begin the Q&A session.
The following is the performance conference Q&A session (AI-assisted summary and translation):
Conference host
Thank you. To ask a question, press the star key and 11 on your phone and wait for us to state your name. To withdraw the question, press the star key and 11 again. Please wait while we put together a list of questions and answers. Please wait while you get ready to access the first question. The first question came from Morgan Stanley's Sanjeet Singh. Your line is connected.
Sanjit Singh (Morgan Stanley Analyst)
OK, thank you for taking the questions, and I'm also happy to see strong results in the third quarter. David, I'd like to talk about the fourth quarter guidance first. If you go back 90 days, have your basic assumptions about different customer groups, including core business and largest customer aspects, changed in any way compared to that time? Or does this outlook just reflect the minimum commitment amount approach you adopted in the fourth quarter? I just wanted to know if your perception of the future had changed.
David Obstler (Chief Financial Officer)
Yes, there are a few points. Thank you, Sanjit. There are a few things I can talk about. First, I think we said that the largest customers have renewed their contracts and are spending more than the minimum commitment amount. That was true in the third quarter. We said that although their spending in the third quarter was higher than the minimum commitment amount, we only included the minimum commitment amount in our guidelines. As we mentioned, after excluding our largest customers, our growth accelerated in the third quarter.
So I would say, the only change is that we completed contract renewals and extensions with some large customers and large AI customers in the third quarter. As a rule, we basically extend the term, increase the commitment amount, and offer a better price. This has been the case for some large customers. This is our normal practice. This is the only thing that has really changed since we last communicated. Ali, do you have anything to add?
Olivier Pomel (Co-founder and CEO)
Yes, we talked about a few of them during the conference call. We mentioned some large customers who renewed their contracts much earlier than we had anticipated. Another renewal occurred in October, and we haven't discussed it here. So this is all good news. But they do put a damper on short-term earnings, which is a change from the previous quarter. In addition to this, the rest of the situation has basically developed according to our expectations, and there are also some encouraging surprises. In particular, the growth of non-AI customers continues to accelerate. The overall situation is very good.
Sanjit Singh (Morgan Stanley Analyst)
Understood. If we consider growth after excluding native AI customers, this growth rate has been accelerating for several consecutive quarters. For this core non-AI-native business, your AI product portfolio clearly has many beneficial factors. It appears that the customer base with annual recurring revenue of $100,000 is adopting more AI products. Do you think this will continue to drive continued growth? Or is the growth rate of the customer base stabilizing after many consecutive quarters of acceleration after excluding native AI customers?
Thank you.
Olivier Pomel (Co-founder and CEO)
I mean, you see, these customers are using AI; they're consuming more cloud resources, developing more software, and changing the way they work. As a result, we expect these customers to experience a long-term phase of growth. As for the overall growth rate — will it continue to increase significantly, or will it remain stable? The current growth rate is already very high, and we feel pretty good about it. It's hard to say where it will go in the future, but we are very confident about the overall trends that will support this growth.
Conference host
OK, thank you. Please wait while we move on to the next question. The next question comes from Goldman Sachs's Gabriela Borges. Your line is connected.
Gabriela Borges (Goldman Sachs Analyst)
Hello, good morning. Thank you for taking the question. David, I want to continue talking about AI's native customer base because I think you've always been very transparent when discussing situations between your biggest customers and other customers in that customer base. What I want to ask is that your current month-on-month performance is a bit irregular, because in the past 12 months, this customer base has achieved a very impressive growth rate, and now the biggest customer has experienced changes in terms of contract renewals. You guys also talked about the early renewal trend.
What I want to ask is, what ideas can you provide on how to model the month-on-month performance of this business? Will we be back to a stable state in about two more quarters? Or are we likely to see another step-like decline in month-on-month performance every year due to early contract renewals? Of course, this in itself is a good situation because you signed these very large contracts, and the prerequisite for the customer to get a better price is an increase in the contract amount.
So, maybe we can talk about what positive and negative factors we should consider when judging the month-on-month performance of AI's native customer base, and when it will return to a state that seems “normal.” I also know that the word “normal” itself is difficult to define.
David Obstler (Chief Financial Officer)
Yes, that's a good question. Regarding the largest customer, I think as we said, this one has already renewed its contract. We decided to only count the promised amount, which is the minimum commitment amount. As a result, this customer spent significantly more than the minimum commitment amount in the third quarter. Therefore, a very important factor affecting the month-on-month performance is what will happen in this area. We have no control — customers can't spend less than the minimum promised amount, but they can increase it by varying degrees. So, that's one of the factors.
Basically, if this amount stabilizes and is no longer above the minimum commitment amount, it will reduce the month-on-month increase in guidance or actual performance. So, that's the first point.
Olivier Pomel (Co-founder and CEO)
I'd like to broaden my perspective a bit. I mean, some of these situations are very common: we keep getting these customers, and they keep growing and renewing their contracts; when they renew, they get more favorable terms and adjust their traffic over time, while always building on growth. This is something we've been seeing since the company was founded. What's new is that some of these companies are expanding at an astonishing rate to a level we've never seen before in this industry — in fact, in any industry — and this has brought about a new dynamic.
That's why we're also a little surprised when we see customers renew their contracts early, just a few quarters after starting to adopt the product. Overall, the number of such customers is growing. The AI native customer base is expanding. There are more customers in this group, and there are also many more large customers. We mentioned several customer developments by name — for example, we signed an AI lab this quarter, which we mentioned earlier. So in the long run, this group will continue to expand and account for a much larger share of our customer base, so stability will increase over time. However, I still don't think this is a very unstable business.
I mean, we're just looking at it season by season, and the final results may be a few percentage points different from what we expected. But you know, at the end of the day, we think this is a usage-based model, and our visibility over the medium to long term is higher than our visibility in the very short term.
David Obstler (Chief Financial Officer)
I'd like to remind everyone that, other things being equal, the month-on-month growth from the third quarter to the fourth quarter was generally lower because the end of December holidays and partial code freeze will affect our usage-based products. Depending on other factors, this effect may be apparent or masked. But that impact did exist in the fourth quarter and not in the other quarters.
Gabriela Borges (Goldman Sachs Analyst)
These instructions are very helpful. Thank you.
David Obstler (Chief Financial Officer)
OK, thank you.
Conference host
Thank you. The next question comes from Barclays Raimo Lenschow. Your line is connected.
Raimo Lenschow (Barclays Analyst)
Thank you, thank you. Can I just ask one last question about this? My understanding is that customers renew their contracts early to get a better price, but their usage is clearly increasing dramatically, which will cause short-term resistance because the same usage now corresponds to a lower price; but then usage continued to grow, and after a few more quarters, we should be able to see a normal growth trend. Is that correct to understand? I remember this happening in the past, just wanted to confirm it again.
Then I have one more question.
Olivier Pomel (Co-founder and CEO)
Yes, that's really part of it. Usually in these situations, there are other things that happen, and that is that the customer will try to figure out what they are going to need. So they look at the services they're using and think, OK, we probably don't need that much for this part, and we might need more for this part. Or in some cases, we'll tell them that, given the way you use this kind of data, maybe they should use a different data storage location.
Therefore, when we work with clients, we also make some optimizations to their account usage configurations. But all of this is happening against the backdrop of growth. These customers are growing, deploying much larger workloads, and are hoping to double, triple, or even tenfold their usage on our platform over time. That's basically what we're working with them to advance.
Raimo Lenschow (Barclays Analyst)
Ok, very clear. Then, the main question I want to ask is the business after excluding AI customers. In the past, we would have thought that part of this growth came from cloud migration, customer modernization, and more business. You mentioned another aspect, which is that because of AI, we can write more code, modernize apps more, and all kinds of things are happening. So is this changing the way you grow your business? In other words, in the past, customers usually needed to use Datadog when modernizing, but now the way companies work has actually changed, and this is also starting to drive the growth of non-AI businesses?
Could you please talk about this?
Olivier Pomel (Co-founder and CEO)
Yes, I wanted to say that there's one thing I didn't mention on the call, but it's always been there. That means we're also doing more for our customers. We've integrated many of the tools our customers used to use. We have more products, and we can cover a wider range. As customers use AI to build more things, some of our other products also become more important. As an example, customers are developing a lot more software, so the pressure on their continuous integration, deployment, and testing has increased dramatically, so they are also increasingly adopting our continuous integration products.
I think we've also mentioned a few customer contract results in this area. As a result, this opens the door for us to promote more products beyond core observability.
Raimo Lenschow (Barclays Analyst)
Yes. Ok, very clear. Thank you. Very well done.
Conference host
Thank you. The next question comes from J.P. Morgan's Samik Chatterjee. Your line is connected.
Samik Chatterjee (J.P. Morgan Analyst)
Hello. Thanks for answering my questions, I have two questions. First, I know you mentioned that the biggest customer actually spent more than what they promised in the third quarter. Can you tell us how much of the excess came from this major customer spending more than the promised amount compared to the third quarter results guidance, and how much came from the rest of the business? Of course, that depends on how you forecast this customer in the third quarter. I have one more brief question.
David Obstler (Chief Financial Officer)
Yes, we won't provide specific numbers, but it's true, if your guidelines are mostly lower than theirs — we've said we'll be conservative in the third quarter, and we've also said that our predictions are conservative compared to their spending levels at the time. For the fourth quarter, we've made it clear that it will be based on a minimum commitment amount. Indeed, if their spending is above this level, then that naturally means — mathematically speaking, yes.
Olivier Pomel (Co-founder and CEO)
We haven't changed our assumptions in this regard over time. We've always kept our original assumptions the same. I think one of the differences we mentioned earlier, and one difference compared to the previous quarter, is that overall customer growth showed some positive trends, while some larger customers also renewed their contracts early, including some other AI customers. This is the change you should keep in mind.
Samik Chatterjee (J.P. Morgan Analyst)
OK, I understand. Another question is related. Revenue contributions from new customers continue to rise steadily. Can you talk about what is the average initial sign-up size for new customers now, how it has changed, and what behaviors have you observed in this regard? Looking ahead to next year, how do you see new customers contributing to overall growth? Thank you.
David Obstler (Chief Financial Officer)
We haven't provided guidance on this yet. We've said that over time, we've seen an increase in the average initial contract size, as the platform continues to grow in size, and as shown by our disclosed product adoption metrics, customers are also adopting more parts of the platform. Ali, do you want to...
Olivier Pomel (Co-founder and CEO)
Yes, I think it's a phenomenon we're observing. What we've seen is that among newly established companies and AI-native companies, as discussed earlier, there are companies that are starting on a larger scale at an earlier stage. However, we're also seeing businesses scale faster than in the past, so they can earn more revenue faster. But we are also developing new market segments and signing major clients from the first day of cooperation. We have listed a few examples in the statement we are preparing for today. I remember we mentioned a defense agency of the US government.
We also mentioned Japanese banks, such as signing two of Japan's top three banks in the same quarter. The scale of these initial contracts is quite impressive. If we compare our current business to five years ago, it's clearly a new, sizable piece of business that we didn't necessarily own in the past. Incidentally, these sign-ups I just mentioned haven't even been reflected in revenue because they've only just become customers and usage hasn't increased. Thank you.
Conference host
Thank you. The next question comes from Royal Bank of Canada's Matt Hedberg. Your line is connected.
Matt Hedberg (Capital Markets Analyst, Royal Bank of Canada)
OK. Thank you all for answering my questions, thank you very much. First, I'd like to ask about the training situation. You mentioned in your prepared statement that an additional purchase deal was reached with the AI department of a major software company, and I remember that the annualized amount reached eight digits of dollars. It sounds like they're using your services in an AI training workload. This seems like a very interesting growth opportunity for us. Can you talk about the opportunities in the field of training from a broader perspective, and what it could mean for your growth?
Olivier Pomel (Co-founder and CEO)
I would say that this field has been moving in a positive direction. If you look back a year or two, almost no one was doing training; only a few cutting-edge AI labs did their own training; no one else did it. Now, we're seeing an order of magnitude more large AI labs or new labs that are training on a large scale. In addition to these laboratories, a whole group of relatively small companies, including us, have also begun their own training. We have already begun training on a considerable scale, and we also showed some of the results on this conference call.
So, I think there may be some notable developments here: training is part of every company's business, and every company is customizing models and building their own models. So we're investing. I think it's too early to call it mass market. Currently, we're seeing some signs worth watching. We've also seen that we can play a huge role for some very large customers. But I think we still need to invest, we still need to develop, and we still need to watch how this market develops.
Matt Hedberg (Capital Markets Analyst, Royal Bank of Canada)
Understood. David, I'd like to ask one more question if I could. Regarding the biggest customer, I think it's interesting that you used the phrase “significantly higher than their minimum commitment amount.” I know you only used the minimum commitment amount in the fourth quarter assumption, but judging from the overall judgment, considering this customer and their recent usage pattern, could this change the way you think about how they use Datadog for the long term? For example, in addition to the current usage method, new usage scenarios may emerge, making this customer's growth opportunity really worth looking forward to?
David Obstler (Chief Financial Officer)
I think this applies to any customer, and Datadog has many different use cases. Specifically, as we've been emphasizing for many quarters, we have no control — we have contracts, we have renewals — but we don't anticipate these situations. However, for our entire customer base, this is true: there are many use cases within an enterprise, whether it be different departments, end users, or other aspects. Ali, you add it.
Conference host
OK, thank you. The next question comes from UBS's Carl Kirstead. Your line is connected.
Carl Kirstead (UBS analyst)
OK, thank you. Olivier, maybe you can answer this question. You mentioned in your statement that many customers are dealing with higher term costs and AI spending and are trying to control these costs. I'd like to know how the company's spending behavior is changing based on the various types of communication you have with customers, given their efforts to control the cost of terms? Also, if some of the customers who renew their contracts early in order to get price discounts aren't native AI companies, but traditional companies, do you think this could be because they are dealing with budget pressure?
Thank you very much.
Olivier Pomel (Co-founder and CEO)
Yes, I think this is a very common question. We have seen an explosion in the adoption of AI, particularly the adoption of programmed agents. And that wasn't in the plan. Basically, a year ago, no one was in their budget to spend as much on AI as it is today, so they really had to figure out how it works, what parts of it are what they really need, and what they can get out of it. As a result, there is a strong demand for relevant tools in the market, and this is exactly the service we provide. This is a great opportunity for us, and our cloud cost management products have excelled in both business growth and adoption.
As for the early renewals we've seen, they're not directly related to this. I think these renewals are from big customers who are either AI labs or focused — in summary, they are tech companies. And this actually has nothing to do with their own vocabulary consumption; more to do with the growth of their own business.
Carl Kirstead (UBS analyst)
OK, thank you.
Conference host
The next question comes from Citi's Fatima Bulani. Your line is connected.
Joel (Citigroup Analyst)
Good morning everyone. I'm Joel, asking questions for Fatima. I'd like to ask two quick questions. First, let's ask David to answer about profit margins and operating leverage for operating profits. The quarter's results were strong, at least in line with our expectations, but we also know that you are investing quite heavily in some areas. Therefore, I would like to know the specific drivers of this quarter's performance and what your updated or latest investment policies are.
The second question is also very brief, so I'm going to ask it together: Is there any data we can share about the penetration rate of Bits AI in the customer base? Thank you.
David Obstler (Chief Financial Officer)
Do you want it—OK, I'll talk about profit margins first. OK, OK. This is very similar to what we said before. We are actively investing in market expansion. Whether in terms of geography or end market, we still have gaps we haven't covered, such as the federal government market. We have also continued to successfully expand the size of the sales team that undertakes sales targets, and the corresponding results have already been reflected. We continue to invest in research and development. In this regard, we are investing in both AI and talent. We saw plenty of opportunities. I'd like Ali to talk about the product next.
Shall we talk about Bits AI?
Olivier Pomel (Co-founder and CEO)
OK. Regarding this, we mentioned on the call that its adoption has been excellent. If you look at the number of unique customers or users using it, this is our strongest performing product ever since launch. So we're very happy with that. Over the past quarter, we've also redesigned the way it packages its products. We used to charge fees specifically for investigation tasks. Now, we use the AI credit model to cover a range of different usage scenarios, with Bits AI acting as our agent, and we have seen very good results as a result.
So it's very exciting. I'd say it's still in the early stages because we're still developing a ton of features for Bits AI: there's so much more that customers want us to do for them, and so many opportunities to automate their workflows. However, we are very, very satisfied with the current state of affairs.
David Obstler (Chief Financial Officer)
I'll go back and add one more point. I think I just missed the gross profit margin you mentioned. I think this is very similar to what we said before, and is a common result of platform penetration and engineering work. Our gross margin has always been around 80%. The quarter ended up being around 81%. So this is normal practice for us: we invest, optimize, and carry out engineering projects. The gross margin we disclosed reflects the stability of this indicator, which was also reflected in previous quarters.
Conference host
Thank you. Please wait while we answer the next question. The next question comes from Kirk Materne of Evercore ISI. Your line is connected.
Kirk Materne (Evercore ISI Analyst)
OK, thank you so much for taking the questions. I'd like to ask David or Ali about these AI native customers who renewed their contracts slightly early. Obviously, they came back to renew their contracts because of the high expenses at your place and they wanted to get a better price. I'm curious what is the additional purchase rate for other products in these recent renewals? Will you still have time to cross-sell them some other products when they come back for early renewals? What are you guys seeing?
I know this will have some negative impact on revenue in the short term, but I think it will also present an opportunity for you to discuss expanding their use of multiple products. Thank you.
Olivier Pomel (Co-founder and CEO)
Yes, I mean if you look at these customer groups, especially if you compare them to the customer groups we might have had five or six years ago, they use far more products. As a result, we see that these customers are adopting products and the entire platform much more widely than they were five or six years ago. This is indeed happening. Part of this happens during contract renewals; sometimes, this also happens naturally as new requirements emerge and usage continues to expand to more teams within these companies.
David Obstler (Chief Financial Officer)
But I think the important point that needs to be clarified is that these are indeed renewals, but the promised spending amount for the renewal is higher than before, and it is for this reason that they received volume discounts. When we renewed our contracts with these customers, their promised spending amounts increased.
Kirk Materne (Evercore ISI Analyst)
OK. David, I'd like to ask you one more time. These customers are entering the renewal phase earlier. I think this is a good thing in the long run, but from a month-on-month perspective, predictability will be slightly weaker. Is that what Ali just meant? I think that's a good thing in the long run, but this will cause you a little more fluctuation as you set guidelines for the next quarter. Is that an accurate understanding?
David Obstler (Chief Financial Officer)
I'd say this will cause a small change in the time distribution of these clients' revenue growth paths. The month-on-month situation here actually involved three factors. First, this is the factor. Second, I think we mentioned the biggest customer in the guide, and you know, we don't predict what they'll do next. Third, we know there are still a few seasonal factors. So, this is one of the factors, and we'll try to figure it out as accurately as possible. Of course, as our remaining performance obligations and contract terms are extended, there will indeed be marginal fluctuations in month-on-month performance.
Olivier Pomel (Co-founder and CEO)
Once again, this is all great news. We use a pay-per-use model, and when we work with high-growth customers, those customers themselves don't fully understand how the next few months will grow, what parts of the business will grow, and how usage will change. So we're happy to work with them to ensure we support their growth. That's it. The only time we found this difficult was when preparing for these conference calls.
We have to figure out what parts we can communicate to the outside world, what parts we know, and what parts we don't. But overall, it's all good news: great customers, strong growth, excellent adoption, and strong trends underpinning the industry. So we feel really good about it.
David Obstler (Chief Financial Officer)
Also, it's similar to what you just mentioned that we've been doing since the company was founded. Our approach is to increase the customer's committed spending amount, expand the scope of use of the platform, and extend the contract period. That's what we've been doing. It's going to cause some fluctuations, but it's very similar to what the company has been doing throughout its development. Also, that's a good thing.
Kirk Materne (Evercore ISI Analyst)
Very good, thank you very much.
Conference moderators
Thank you. The next question comes from Bernstein Research's Peter Weed. Your line is connected.
Peter Weed (Bernstein Research Analyst)
Thank you very much. We just talked about that, but it's really exciting to hear about the Bits AI Scout launch and strong early adoption. Can you provide some more specific data and remind us how this product is priced? Is it directly generating revenue or is it mainly driving the adoption or use of Bits AI? Overall, it would be exciting to understand how widely Bits AI is currently being adopted and how it might drive overall growth and even become another major pillar of the business.
Olivier Pomel (Co-founder and CEO)
Yes, it actually has both. It's largely based on all the data we get, how we process that data, and the variety of products we have. This is the core. However, based on this, we also have a model that charges AI points specifically. The model we chose is to make the fee quite close to the deduced cost, so as the efficiency of the model increases, we can pass on the cost savings to the customer, and we can also conduct more investigations.
We're doing this because the field is changing too fast, and we're improving the efficiency of these models very fast. Therefore, setting a fixed price for a strategic activity you carry out when you expect the number of tasks to be performed will increase by several orders of magnitude, and the price will also drop by several orders of magnitude, is not a very effective practice. So we've used this model for this, and it works really well. Therefore, in the long run, we hope not only to directly generate revenue through Bits as it is now, but also to promote wider adoption of the platform and provide more value to customers.
As for which one will become the main method of monetization, it is currently difficult to determine, and we are not particularly concerned. Again, we only charge for what you use, so as long as usage increases and the value provided increases, we have no problem.
Conference moderators
Thank you. Thank you. The next question comes from Bank of America's Koji Ikeda. Your line is connected.
Koji Ikeda (Bank of America analyst)
OK, thank you guys. I'd like to ask a question about infrastructure monitoring. You disclosed that the business's annual recurring revenue now exceeds $2 billion, and the growth rate is in the middle of the 30% range. Can you quantify how much the growth rate has changed compared to when you last disclosed that the business was $1.5 billion earlier this year? How much of this growth comes from AI infrastructure? How sustainable is this rate of growth as GPU efficiency increases? Thank you.
Olivier Pomel (Co-founder and CEO)
We're seeing the overall business grow at an accelerated pace, not only from AI-native customers, but also from other customers. We just mentioned this, and so is the infrastructure monitoring business. Currently, this growth is actually not directly related to GPUs. From what we've seen with our customers, GPUs are still mainly used for training. We have products in this area, but have yet to generate extensive revenue from these products. So for us, this is an opportunity for the future.
This opportunity has yet to be reflected in current growth and current performance figures.
Conference moderators
Thank you. The next question comes from Scotiabank's Patrick Covo. Your line is connected.
Patrick Covo (Scotiabank Analyst)
Thank you so much for letting me participate in the question. For me, the most interesting thing about the prep talk is that Datadog is training its own cutting-edge model. Olivier, you later stated that this practice has not yet been popularized in the mass market, but you are watching how the market develops. I'd like to ask you how this affects Datadog's choice to use a proprietary model versus an open weighting model internally or for customers since you're at the cutting edge of these changes?
Then I wanted to ask David what this shift as Olivier described if it were to accelerate, what would that mean for Datadog's profitability? Your profitability has improved quite a bit this quarter. Is this change also likely to help increase net profit? Thank you.
Olivier Pomel (Co-founder and CEO)
What we're seeing internally and with our customers is a mix of different models. So, I think we can use the generic term “we” here to include both customers and ourselves, because the practices of both parties are quite similar. We use a lot of top models from top labs, and we also use open weighting models, and the use of open weighting models is growing. We also train our own models and post-train them. I think this probably represents the direction of the industry as a whole.
We're doing this because some abilities can only be obtained from top models trained by other institutions. At the same time, there are many things that use a ready-made open weighting model that's good enough and cost effective, and you can run it directly. Finally, there are some things you can do to greatly improve your efficiency through your own data, your own workflow, and your own post-training. Some of the models we've published use exactly this approach.
So I expect this combination to stay largely the same. As for the impact on profit margins, at the end of the day, you have some control over where your money is spent. If a competency is very specific to itself, and you can achieve better cost effectiveness, then you should definitely do it. If not, for example, it's less differentiated, and applies to some scenarios in your business, as well as to millions of other businesses, then it's probably fine to use the ready-made model, and you can keep using it.
Conference moderators
Thank you. The next question comes from Alex Zukin of Wolfe Research. Your line is connected.
Alex Zukin (Analyst at Wolfe Research)
Hello everyone, thank you for giving me the opportunity to ask questions. Ali, I'll ask you a quick question before asking Dave. Have you seen to what extent is the demand for intelligent observability from native AI customers and core customers driving the demand for products such as link tracking products? Given the increased level of adoption, how has this changed in the longer term? Also, can you talk about cybersecurity and your plans in this area?
Olivier Pomel (Co-founder and CEO)
Yes, we see that demand for intelligent observability is growing, and there have always been notable demands from AI and other fields in the past. But the most notable thing here is that non-AI-native customers are also advancing more and more, and as they further apply AI, we see real demand, actual usage, and very exciting sales opportunities for these products. So we think this is arguably crossing the divide and exciting for this product, even though we haven't disclosed any specific numbers today.
When it comes to cybersecurity, AI brings a host of new opportunities. We are very happy to see that we are able to enable our larger customers to adopt the entire platform. We mentioned a defense agency, and also mentioned a few other winning cases. I remember in those cases, we not only had our customers use an observability platform, but also our security platform. Therefore, we are confident in how well the product fits the market, and we are also satisfied that AI-driven comprehensive tool renewal and reconstruction is creating favorable conditions for us.
Alex Zukin (Analyst at Wolfe Research)
Very good. David, maybe you can clarify next. First, it's great to hear that you've renewed your contract with your biggest customer. I think we fully understand the fourth-quarter approach of giving guidelines based on minimum commitments. What are your suggestions on how to build a model for 2027? Because it's probably a one-year contract. Should we consider keeping it at a certain level? Or are you guys going to take it off the model?
Also, if you give guidelines for some other AI-native customers who renew their contracts early, are these customers also calculated based on the minimum committed spending amount? Or is it slightly above this level? I think any help with these two points would be useful.
David Obstler (Chief Financial Officer)
Yes, just a reminder - yes, yes, sorry. We will provide guidance for 2027 during the year-end conference call. I think we've said in the past few quarters that we are guided by the minimum commitment amount stipulated in the contract, and this quarter has been repeated again. More instructions will be provided when we give guidance for 2027. As for other businesses, our approach is, as we've been saying for years, we base the growth rate of these customers and then apply a certain discount. This is the approach we use across our entire customer base: based on recent organic growth rates and then cut back. We will of course use this approach for customers who renew their contracts early and for the wider customer base.
Olivier Pomel (Co-founder and CEO)
Thank you all. OK, I think that's all for today's Q&A. Thank you all for listening. I would also like to thank every colleague at Datadog for working together to create a fantastic third quarter. I know we're now in the middle of an even better fourth quarter. Everyone is very busy. We'll meet — in a few weeks, I'll be meeting many customers at the largest AWS conference, and I hope to share more exciting product launches with you soon. Thank you so much and see you next quarter.
Conference moderators
Thank you for joining today's conference call. This concludes this meeting. You can hang up now. I wish everyone a happy day.