Elastic N.V. (ESTC) Earnings Call Transcript & Summary
February 26, 2026
Earnings Call Speaker Segments
Eric Prengel
executiveThanks. Good afternoon, everyone, and I appreciate you being with us on this conference call as we go over Elastic's financial results for the Third Quarter of Fiscal 2026. Joining us today are Ash Kulkarni, our Chief Executive Officer, and Navam Welihinda, our Chief Financial Officer. Once they finish their prepared remarks, we'll open things up for questions. We put out our press release earlier today after the market closed, and it's available on our website. You can also find slides that go along with this call on the Elastic Investor Relations website at ir.elastic.co. During our discussion, we'll be making forward-looking statements, and those might include predictions, estimates, what we expect for demand for our products and solutions, our future revenue, and other information. These forward-looking statements rest on factors we currently know about, they speak only as of today's date, and they carry risks and uncertainties that might cause actual results to come out materially different. We disclaim any obligation to update or revise these forward-looking statements unless the law requires it. Please look at the risks and uncertainties laid out in the press release we issued earlier today, in the slides posted on the Investor Relations website, and those described more fully in our filings with the Securities and Exchange Commission. We'll also be talking about certain non-GAAP financial measures. You can find disclosures about non-GAAP measures, including reconciliations with the most comparable GAAP measures, in the press release and slides. Unless we specifically say otherwise, all results and comparisons are on a fiscal year-over-year basis. A webcast replay of this call will be available on our company website under the Investor Relations link. Our quiet period for the fourth quarter of fiscal 2026 starts at the close of business on Thursday, April 16, 2026. We'll be at the Morgan Stanley Technology, Media and Telecom Conference on March 2; with that, I'll hand things over to Ash.
Ashutosh Kulkarni
executiveThanks, Eric, and hello everyone on the line this afternoon. We appreciate you being with us on today's call. Elastic has turned in one more exceptional quarter, surpassing the top end of our guidance on every key metric and demonstrating what the Elastic platform and our business model can do. Momentum in Q3 came from persistent demand for the platform, robust sales execution, and our unrelenting customer-first mindset. As LLMs advance their inference and reasoning abilities at a fast clip, it's growing clearer all the time that context matters most when it comes to making these models genuinely useful inside an enterprise. Against that backdrop, during Q3 we kept seeing enterprises pick Elastic to supply context for their most business-critical AI requirements. Putting that success into numbers, total revenue grew 18%, and non-GAAP operating margin came in at 18.6%. Sales-led subscription revenue climbed to 21% growth, in step with our expanding group of customers at $100,000 ACV, which has now passed 1,660. Q3 was our sixth quarter in a row of robust field execution, generating dependable customer commitments and underpinning healthy CRPO growth. That same execution is feeding a robust pipeline as we move toward Q4. Proprietary data that organizations generate, administer and examine daily to guide business decisions and operations is the lifeblood of those organizations; such data is enormous, frequently reaching many petabytes, and for reasons of cost and security it simply cannot be relocated beyond the organization's control. For businesses to adopt agentic AI, the data must be met by the LLM; that's precisely where Elastic fits, thanks to our capacity to help organizations store and administer all their data in highly cost-efficient manners and to supply precise, real-time context to AI by querying all that organizational data as it happens. On top of that, Elastic can accomplish this uniformly in cloud and self-managed settings. Thanks to this hybrid flexibility, sensitive data and workloads can stay in whichever environments are preferred, removing any need for expensive replatforming. It's this distinctive flexibility that keeps us winning against legacy vendors and niche cloud-native competitors alike; and the outcomes speak for themselves. Commitments exceeding $1 million in annual commitment value signed during the quarter rose more than 30% versus the equivalent period a year earlier, fueled by new logos and customer expansion. Consolidation and AI are potent tailwinds. As organizations cope with exploding data volumes, they look to Elastic to bring both innovation and efficiency to their search, observability and security requirements. To illustrate, we closed a 7-figure new logo deal with a Fortune 100 insurance institution for Elastic Security. Wanting to modernize its security operations, the company launched a competitive process to swap out a legacy SIEM solution that suffered from sluggish query speeds, inefficient data retention and inflexible SOC workflows. Using capabilities such as Logsdb and searchable snapshots, they are consolidating data into one cyber data lake with integrated AI-powered SIEM workflows, all enabled by Elastic and its features, including AI assistant, attack discovery and AI-driven orchestration. This shift lets their analysts reach notably quicker cybersecurity detection and remediation results while satisfying stringent regulatory requirements. Also in the quarter, we landed another big one: a worldwide leader in data resiliency software picked Elastic Observability to run the monitoring layer behind its brand-new cloud service. While shifting an enormous subscriber base over to the cloud, they're putting our entire Observability suite to work — AI assistant with Logsdb included — so they can move away from reactive firefighting and toward intelligent, semantics-aware analysis. With open telemetry integrated alongside our vector search, this customer can now spot anomalies ahead of time and fix problems through natural language queries, which cuts mean time to resolution dramatically. Our deep integration flexibility, stronger treatment of unstructured data, and capacity to deliver one source of truth across the whole organization are what won them over from incumbents. The key point: when businesses work through cloud migrations, the platform they need can't make them pick between the data centers they already run and the cloud. That asymmetric edge we hold in serving modern cloud plus hybrid environments produced a major win with a global financial group. This quarter, we signed a 7-figure expansion for Elasticsearch, the foundation of that group's online banking app serving tens of millions of users. What they required was a central data repository that handles cloud and self-managed architectures alike, so mission-critical workloads can run wherever they prefer without any performance trade-off. Elastic came out ahead where their current MongoDB deployment couldn't deliver the scalable retrieval and precision needed to go past simple search into production-grade context engineering. Going forward, they're layering in semantic search and advanced AI features to personalize the user experience further via retrieval that's quicker and more accurate. Underpinning these enterprise engagements is agentic AI's emergence. Buyers are shifting from passive Q&A toward active agents that push workflows along. Precise action calls for precise data. The discussion is no longer about picking a model — it's about feeding that model the most accurate context possible. To capture AI's value, enterprises understand they have to close the distance between their LLMs and their own unstructured and structured data. Elastic is what makes this AI work. We're the engine enabling enterprises to build production-grade AI systems genuinely deserving of their business. Others may sell plain vector databases, but we recognize vectors by themselves won't cut it. From hybrid search through advanced reranking, we provide the complete retrieval toolkit, so agents get the relevant context required for precise actions. Bridging enterprise data to the LLM through our platform is converting straight into broader AI adoption; during Q3, fresh customer commitments involving AI kept climbing. We've now got more than 2,700 customers on Elastic Cloud using us as a vector database, plus others using us for wider AI capabilities such as agent builder and attack discovery, which puts our overall AI customer count above 3,000. More than 470 customers now carry an ACV of $100,000 or greater while using us for AI. That figure includes over 410 using us as a vector database. Altogether, more than one-quarter of the customers in our $100,000 ACV group are now making use of AI use cases. Demand keeps coming steadily from the biggest corporations on earth, and we're also fielding interest from the fresh generation of AI native businesses. In the quarter, we signed several new logo and expansion agreements with AI-first trailblazers, which confirms our platform serves as the default choice for long-standing enterprises as well as challengers. One prominent AI recruiting platform, relied on by both big enterprises and young companies, picked Elastic's vector database to run its primary customer-facing application, since our search performance at scale beat what rivals offered. A company working on AI-enabled driver and fleet safety broadened its adoption of Elastic Search in Q3 as it grows into additional worldwide regions. Elastic delivers the real-time retrieval their platform depends on, so they're able to handle rising data volumes while keeping performance intact. Also, an AI-native cybersecurity leader specializing in AI automated penetration testing brought our SIEM solution into its offering. Elastic gathers all their logs in one place with no hassle, letting them scale effortlessly along their enormous growth path. Underpinning all of these victories is how well our Search AI platform performs.
Ashutosh Kulkarni
executiveIt's not merely about piling on new capabilities — we're pushing hard to make our engine as efficient as possible, directing our developers toward shipping relevance, speed, and efficiency that lead the market. Over the past 18 months, we've cut the RAM needed for vector search by 2 orders of magnitude, thanks to breakthroughs such as improved binary quantization, or BBQ, Disk BBQ, and our Acorn filtering algorithm, along with other work; because of this investment, Elasticsearch vector search runs as much as 8x faster than OpenSearch. That edge in performance brought us one 7-figure deal with a worldwide heavy equipment maker. This customer keeps moving mission-critical workloads from OpenSearch onto Elastic Cloud so they can gain better scalability and performance. Our platform is what powers their high-speed search across telemetry data gathered through the StarLink network. Using Logsdb, they've cut cloud costs substantially while handling 7 years of historical customer data. Performance remains our priority in our NVIDIA partnership too, where jointly we assist enterprises in deploying AI applications more quickly without overburdening IT infrastructure; we just announced the technical preview of our Elasticsearch GPU plug-in for a GPU-accelerated vector database, enabling indexing that's 12x faster. On top of that, the Dell AI data platform, now combined with NVIDIA and Elastic, provides a tightly integrated AI stack that makes building, deploying, and scaling AI much easier. Making Elasticsearch a central piece of the Dell and NVIDIA AI factories means we're addressing the essential need to build AI on infrastructure the customer controls. As we sharpen these technical strengths, we reinforce our technical moat and take friction out of scaling AI; during this quarter, we hit multiple product milestones meant to make the journey from data to action simpler for our customers; we're delivering an end-to-end framework for creating the next generation of intelligent applications. To start, we formally released the general availability of Agent Builder. With Agent Builder, developers can create secure, context-driven AI agents within minutes. Rather than consumer apps that browse the web, we concentrate on internal business applications that draw on company data. We trialed agent builder with a Global 100 financial group to investigate and troubleshoot its production infrastructure, showing an order of magnitude gain in performance on complex issues and bringing specialized expertise for fast troubleshooting to more people. An international entertainment and media company built a chat interface for customer interactions. They determined that Agent Builder's results were far more reliable and accurate than the other LLM-centric approaches they'd attempted. Creating an agent is only part of the challenge. The rest is making sure that agent has the most relevant information within reach; this quarter, we broadened our Elastic inference service to add Jina AI's multilingual reranking models. Jina AI provides a best-in-class model for search accuracy, with Jina V3 currently the #1 reranker in its model size category on the MTEB English retrieval benchmark, a gold standard for search and rag relevance. Jina AI's V5 Nano and V5 small models keep surpassing competitors too, earning high scores in retrieval, reranking, and other tasks. Offering these models in native form lets our customers fine-tune their AI apps to achieve the best possible precision and recall. Within a context engineering pipeline, reranking represents the essential following stage, guaranteeing that the data most relevant gets shown to the LLM. Across more than 80 languages, Jina's cutting-edge models turned in better results. Although reasoning comes from AI, businesses still need the dependability that rule-based automation provides for mission-critical operations. That's the reason we rolled out Elastic Workflows in technical preview. Through Workflows, automation gets built straight into our platform, which lets agents coordinate actions spanning internal and external systems, such as Slack or ServiceNow. It transforms Elastic from a search box into a full system of action. Last, we're fulfilling our commitment to hybrid flexibility through Cloud Connect for self-managed customers. We understand that plenty of our biggest customers, especially those in financial services and government, keep data on-premises owing to regulatory or sovereignty considerations. Even so, obtaining and running GPU hardware for AI presents an enormous obstacle for such teams. With Cloud Connect, customers can retain their data locally while securely bursting into Elastic Cloud to take advantage of NVIDIA GPUs for high-performance inference; bridging modern AI capabilities with stringent enterprise requirements in this way is precisely the reason we keep winning large-scale displacements against legacy providers. As organizations put priority on both innovation and operational efficiency, they are shifting away from disjointed legacy tools and toward Elastic's unified search AI platform. This quarter's results — accelerating growth, large deal momentum and major competitive displacements — verify that our strategy is connecting and that we are prevailing in the race to become the essential infrastructure for the next generation of AI-powered businesses. I'd like to thank our customers for their partnership, our shareholders for their trust and, most importantly, our employees for their tireless spirit of innovation. With that, I will hand the call over to Navam to review our financial results in more detail.
Navam Welihinda
executiveThanks, Ash, and good afternoon, everyone. We turned in another exceptional quarter, beating the top end of both our revenue and profitability guidance ranges, powered by yet another period of steady execution, robust consumption, and solid customer commitments spanning search, security, and observability. The momentum we've shown across this fiscal year speaks to how well our team innovates quickly and executes on sales consistently, quarter after quarter. The demand we keep seeing in the market is converting into growth in total revenue, in sales-led subscription revenue, and into healthy gains in pipeline generation that will fuel our future growth. Taken together, these elements highlight how our strategic value keeps rising as a critical data platform in the AI era. Total revenue for the third quarter came in at $450 million, up roughly 18% as reported and 16% in constant currency terms. Sales-led subscription revenue for the third quarter reached $376 million, an increase of 21% as reported and 19% on a constant currency basis. We benefited from commitment contribution out of both our self-managed and cloud offerings, and aggregate consumption trends stayed strong during the third quarter. Our current remaining performance obligations, or CRPO — the piece of RPO we expect to recognize as revenue over the next 12 months — surpassed the $1 billion mark for the first time in Q3. CRPO climbed to approximately $1.06 billion, up 19% as reported and 15% on a constant currency basis. Within our consumption business, we build customer contracts around their annual usage. That means our CRPO provides a very clear window into the revenue we'll recognize over the next 12 months, which gives us visibility and confidence in our business. As Ash noted, deal momentum carried on in Q3; strength this quarter was evenly distributed across all geographies, and we keep seeing customers sign multiyear commitments this quarter, a clear sign of how our customers regard the Elastic platform as a critical foundational element within their long-term data architectures. That positive momentum showed up in our RPO, which grew strongly at 22% for the quarter as reported and 18% on a constant currency basis. Our deal momentum also shows in the rising count of customers holding over $100,000 in annual contract value. We closed the third quarter with more than 1,660 customers at an ACV above $100,000, which is growth of 14%. Compared with the prior quarter, we brought in roughly 60 net new $100,000 ACV customers. We saw strong field execution and healthy growth across our solutions, with search continuing to benefit from ongoing AI momentum. This demand helps both cloud and self-managed, since both form factors matter for AI use cases; we keep seeing customers take a self-managed license and deploy Elastic into their own modern cloud and hybrid environments. This demand reflects customers' preference for Elastic, which uniquely delivers the control and cost efficiency AI initiatives require. AI also remains a powerful catalyst for customer expansion. Within our cohort of customers spending more than $100,000, 28% are now using Elastic for AI, which covers added AI features such as Attack Discovery and Agent Builder. We remain in the early phases of expansion today, and we anticipate substantial room for continued upside, for both new and existing customers, to speed up their AI adoption over the coming years, especially as they grow into and further within our $100,000 ACV cohort. Now shifting to margins and profitability for the third quarter. Every metric I cover will be on a non-GAAP basis. Our dedication to balancing growth against disciplined spending produced solid operating leverage and strong bottom line results; we stay focused on costs and efficiency across our business. Subscription gross margins came in at 82%, total gross margins were 78%, and operating margin was 18.6%. This Q3 operating margin outperformance stemmed from our strong revenue performance, the continued leverage in our model, and certain Q3 expenses shifting into Q4. Given that outperformance, we now anticipate full year margins coming in a bit better than we previously expected, with updated FY '26 operating margin guidance of 16.3%. On cash flow, adjusted free cash flow totaled roughly $54 million in Q3, a margin of roughly 12%. We expect our cash flows to vary quarter to quarter depending on the timing of bookings and collections tied to enterprise booking seasonality. Therefore, we keep managing cash flow on a full year basis. For fiscal 2026, our full year outlook is unchanged, and we still expect to maintain the adjusted free cash flow margin levels we reached in fiscal 2025. We have made considerable progress on the $500 million share repurchase program announced in October. In the third quarter, we returned about $186 million to shareholders, reflecting purchases of roughly 2.4 million shares. In total, we have bought back 3.8 million shares. At our Financial Analyst Day in October, I noted that we expect to deploy more than 50% of the $500 million authorized amount in fiscal 2026, and we have already surpassed that goal. By the close of Q3, we had finished 60% of our repurchase program, and we are carrying on with the repurchase program here in Q4. Let's turn to our outlook for the fourth quarter and the rest of fiscal 2026. For the fourth quarter of fiscal 2026, we anticipate total revenue between $445 million and $447 million, which is 15% growth at the midpoint, or 13% constant currency growth at the midpoint. We anticipate sales-led subscription revenue between $371 million and $373 million, which is 18% growth at the midpoint, or 15% constant currency growth at the midpoint; we anticipate non-GAAP operating margins of approximately 14.5%; we anticipate non-GAAP diluted earnings per share between $0.55 and $0.57, using between 105.5 million and 106.5 million diluted weighted average ordinary shares outstanding. Taking into account what we've guided for the fourth quarter, we're also lifting our targets for the full year on both total revenue and sales-led subscription revenue. For total revenue, our expectation sits between $1.734 billion and $1.736 billion, which works out to roughly 17% growth at the midpoint, or 15% constant currency growth at the midpoint. On sales-led subscription revenue, we're looking at a range of $1.434 billion to $1.436 billion, which is 20% growth at the midpoint, or 18% in constant currency growth at the midpoint. For full fiscal 2026, we anticipate non-GAAP operating margin of about 16.3%. Non-GAAP diluted earnings per share should land between $2.50 and $2.54, assuming diluted weighted average ordinary shares outstanding of 107 million to 108 million. There are a few additional financial modeling items worth noting. The diluted weighted average shares outstanding figure takes into account only buybacks that were finished as of January 31, 2026. When you look at comparing sequential quarters, remember that Q4 contains 3 fewer days than each of the first 3 quarters of the year had, and that produces a sequential headwind to revenue, something we've built into our guidance. Beyond that, as tends to happen in prior Q4 periods, we anticipate seasonally higher expenses tied to the timing of employee benefit costs. Those costs were already included in the guidance we originally set out for the year. Just like in years past, we wrapped up our planning for the next fiscal year during the fourth quarter, and we'll share our initial FY '27 guide on our earnings call in May. To sum up, Q3 was yet another very strong quarter here at Elastic. Steady sales execution across FY '26 keeps pushing our expectations for sales-led subscription revenue growth higher for the year, which confirms how durable this business motion is. As I mentioned last quarter, although quarterly revenue can naturally fluctuate in a consumption model, our strong customer commitments are what drive strong annual growth. Powered by a highly differentiated platform and the growing value we bring to our customers, we're still on track to hit our medium-term targets for both sales-led subscription revenue growth and adjusted free cash flow. Looking ahead, we have confidence in our ability to keep driving profitable growth. We are the critical technology that accelerates data discovery, secures infrastructure and maximizes application performance. With that, I'll open it up for Q&A.
Sanjit Singh
analystCongratulations on how steady things have been throughout the business. Navam, I'd like to revisit a few of the points from the Investor Day that took place a couple of months back. You gave us a data point regarding AI native customers, or AI customers, making up a fairly small portion of customers in fiscal year '24, yet generating a disproportionately large amount of expansion — that year 1 to year 2 expansion. So the essence of what I'm asking is this: once we reach roughly 25% penetration of your $100,000 customer cohort, could growth not merely stay steady but actually speed up on a more lasting basis as we arrive at those critical tipping points, so to speak?
Navam Welihinda
executiveYes. Thanks for asking, Sanjit; and the trends we presented during the financial Analyst Day for the generative AI cohort, those are unchanged; so we keep performing well, and we're seeing stronger growth on those generative AI cohorts today than was the case when we disclosed it to you during Financial Analyst Day. So those tailwinds are with us right now, and more of our customers are reaching the $100,000 mark. Now keep in mind that every one of these customers at the $100,000 mark is also early in their journey. So there's this additional dimension of further penetration and maturation in their own AI journey, which will drive faster growth as well; so those tailwinds are with us right now; we've seen tailwinds that average to 5%, but there's obviously more that — some customers have higher growth than that. And to answer what you asked, yes, absolutely, a possibility exists. The art of the possible is there for us to actually accelerate beyond that 5% we laid out during Financial Analyst Day, and the trends remain positive.
Ashutosh Kulkarni
executiveAnd Sanjit, to build on what was just said, this is precisely the reason our attention is so fixed on how deeply AI has been adopted across the customers we serve. And when it comes to those customers, at present, on a quarterly basis, you can observe that adoption depth going up. At the outset, that depth begins when they engage with us in a limited fashion; and as that engagement expands, exactly as you noted, it will contribute to consumption, it will contribute to revenue overall, and that will become evident in the business continuing to strengthen and accelerate.
Sanjit Singh
analystGot it. And Ash, this one might be for you. In your script, you noted that vector databases and vector search alone aren't sufficient when it comes to creating an AI application that's both powerful and resilient. Plenty of folks, I imagine, would share that view. So if we position the company as a context engine, which essential components must be in place to earn the standing of top provider of context for AI applications?
Ashutosh Kulkarni
executiveAppreciate that question. What matters most, I believe, is remembering that context shifts depending on the task; therefore the data platform—the context engineering platform—that you offer has to handle many different things simultaneously in a highly consistent manner. First comes the capability to ingest any type of data whatsoever; and as you're aware, we possess distinctive strengths in pulling in not only structured information but also unstructured, extremely messy information. Second comes taking that data and turning it into vectors to enable vector search, an incredibly potent method, particularly within AI for semantic search, yet beyond that, being able to combine it with hybrid search methods. That covers textual search plus the capacity to rerank using various methods so the context you get is as accurate as possible; thus the Jina AI embedding models along with the Jina AI reranker models form an essential component of the platform infrastructure as a whole; beyond that, you require something enabling you to put together agents drawing on each of these capabilities, which is precisely what Agent Builder was intended to do. As you're aware, it's a fairly recent feature and capability from us, yet we're witnessing strong traction and adoption among our customers; beyond that, workflows are necessary since agents today aren't solely about chat. They aren't solely about conversations; they involve executing precise actions. That's precisely why the workflow functionality we shipped matters so much. Finally, there's the capacity to observe everything, which is precisely where our LLM Observability functionality proves essential. Thus we think it's each of these capabilities, Sanjit, that collectively render the platform extremely compelling for context engineering. Beyond all that, we've likewise introduced our Elastic inference service. That means you don't have to supply your own LLM. We're able to help you proxy to whichever LLM you'd like to use. We integrate with essentially every one of them.
Robbie Owens
analystI want to say sorry right away for piling several questions into one, though I'll limit myself to a single question that may have 3 parts. My main interest is the better-than-expected results in other subscription. I realize you plan to serve customers wherever they prefer to make purchases, so to begin with, could part of that upside have come from deals that slipped from the earlier quarter? Also, your Q4 forecast for sales-led subscription is lower from one quarter to the next, something that hasn't happened before, since it normally rises slightly. Should we attribute that entirely to how strong the January quarter was, or is there another explanation? Finally, regarding how you monetize self-managed customers compared with cloud customers and your capacity to grow those accounts in the years ahead, could you explain what separates the 2, assuming a meaningful gap exists? Once more, I'm sorry for the 3 questions, though I hope the answers can be short.
Ashutosh Kulkarni
executiveSure, Rob, I'll begin. This is Ash speaking. I'll open and afterward hand things over to Navam. Regarding the momentum we have in self-managed, this isn't merely a matter of delayed deals or something similar; we keep witnessing considerable momentum in that part of our business. Ultimately, what we observe today, particularly because of AI, is that many customers are running AI against data they view as highly mission-critical and highly confidential. That applies not only to government accounts but also to other regulated sectors. Consequently, they decide, or they lean toward, retaining that data in a place they control, inside their own environment. That isn't necessarily their own data centers. It could instead be their own cloud VPCs, and we provide the flexibility for them to do so. So these are modern workloads that keep expanding as their AI usage expands, and we will keep gaining from that, which explains why we feel it's truly essential to consider not only cloud but the entire picture and factor in the robust growth we see even in self-managed. I'll have Navam take the remaining questions.
Navam Welihinda
executiveAbsolutely, Rob — I'll take your question about the sequential quarterly performance. Let me begin by talking about the overall state of the business. Our sales-led motion continues to be executed really well. This marks yet another quarter of solid execution on the sales side. We also saw that reflected in our CRPO and RPO figures, which accelerated. When it comes to commitments, the volume looks good, and we aren't seeing any slowdown there. Beyond that, our pipeline remains quite healthy and keeps expanding quarter after quarter. So all in all, from a business standpoint, we're quite pleased with how the quarter finished and feel very optimistic about the quarters ahead. That brings me to the guide. When you consider the guide, we've always set it with a suitable degree of prudence regarding what we can deliver, and we beat it each quarter. So when you're looking at — examining past figures alongside actuals and guidance, what you're doing is putting an actual figure up against a guidance figure, and that guidance figure already bakes risk into those forward-looking projections; that's the first thing I'd highlight. The second thing I'd mention is that Q4 has 3 fewer days, and that works out to a 3% headwind, or roughly a $14 million — or $14 million to $15 million headwind on the revenue line, simply because there are fewer days in which to recognize revenue. All of that is factored into the guide. And if you review last year's Q4 guidance, or Q4 guidance from prior periods, there have been times when we guided below the then-current quarter. So just bear that in mind. We remain firmly on track toward hitting our midterm targets, and we feel really good about how strong the business is on its own.
Matt Hedberg
analystAsh, I'd like to ask about AI. Clearly, everyone has witnessed the strain on the software sector. I appreciate what you said at the beginning of this call — it genuinely helped to hear how you see AI. From a customer standpoint, there seems to be a great deal of positive momentum. My question, I suppose, is this: looking at these frontier models, are they future rivals in your eyes, or is this more about partnership potential?
Ashutosh Kulkarni
executiveHonestly, our view is that AI does not replace us. Whether it does comes down to what we make of it, since these frontier models are, after all, remarkable reasoning engines. My mental model for them is that they will serve as tomorrow's operating systems. Yet in the same way that operating systems right now still need data systems to supply suitable data and context so that applications can actually be built on top of them, the very same requirement will hold going forward. Within this entire ecosystem, our job is to ensure that we can, very quickly and in real time, across every petabyte of data that any organization holds, deliver the correct context to these LLMs so that they can carry out their work. That is precisely why I hold the belief that in tomorrow's world, agents will converse with one another. Agents that you create using Elastic agent builders will converse with Claude Cowork, which will converse with whatever you create using OpenAI Frontier; and the MCP protocols and the A2A protocols that enable that sort of communication are things we already support. So this is a landscape in which we believe that the tremendous position we occupy — our vector database capabilities, our capabilities across the whole context engineering platform — makes us a critical piece of the infrastructure going forward; and today we are already partnering with hyperscalers, and we already integrate with every one of these frontier class models.
Matt Hedberg
analystThat's a really useful way to look at it; and then perhaps one brief follow-up regarding Elastic's use of AI on the inside. What tangible advantages are you observing? And how could that affect headcount down the road?
Ashutosh Kulkarni
executiveYes. Look, AI is something we're fully committed to, not only regarding what we do on the outside by delivering the platform we're developing, but also regarding the way we apply AI within our own operations. To give you a bit of background here, 2 years back, we created our initial agent, our first support agent inside the company. And it has now been running in production for quite a while. It's the first thing customers encounter when they bring support questions, and the volume of queries it can answer and the count of support tickets it can deflect has not only boosted the general performance, the general experience for our customers when they reach out to us for support, but it has also meaningfully lowered the headcount demand on our end. So across the last 2 years, even while our business has expanded, and as you'd expect, support workloads usually scale alongside the business, we've been able to handle that workload increase with no headcount additions to that support team; in other areas of the business, be it HR, be it finance, in legal, we make heavy use of AI tools. A number of these are built on our stack. A number of them could be outside products that we're using. And even within engineering, we're discovering enormous value from employing several distinct code generation tools that we use inside the company. So on the whole, we think this will certainly help us not only speed up the rate of innovation, which we're already witnessing today, but also lift the productivity and lift the general efficiency of the business; and that's the exciting part about this. We can assist our customers with this, yet we can also gain from it ourselves.
Brian Essex
analystThanks for addressing what Matt asked about your vector database features and your context engineering platform. I suppose, when you consider how the landscape is shifting and you examine various strategies, various ways of approaching these things, is there anything — how should we think about the platform and how well it can align with some of those strategies? Take, for instance, the Page Index method for RAG. If they manage to fix the cost and latency problems tied to that method, are you in a strong position to gain from something like that and shift your own approach accordingly?
Ashutosh Kulkarni
executiveYes. Look, RAG, retrieval augmented generation itself has advanced considerably over the past several years since it first came about as a concept. But at its core, this boils down to locating the most suitable context that's relevant so the LLM can perform its task. At times that means you need to grasp particular data relationships that could be present. At times it means you simply need to search across all your data. At times it means you need to grasp particular things, such as preferences and similar details that you may have recorded in other data systems. And it's a combination of all these elements. And as RAG keeps developing, as these methods grow increasingly advanced, we're genuinely at the forefront of incorporating more than just one single method into our platform. We were among the earliest to embrace hybrid search, and we were the first to discuss it. And from then on, we've maintained that sort of momentum; so without a doubt, I feel extremely, extremely confident that we'll be at the bleeding edge. Ultimately, this is what Elastic was created to do. We've always been focused on relevance. Without relevance, you don't get good search. Without relevance, you don't get good accurate AI.
Brian Essex
analystGreat, that's really useful. Perhaps one brief follow-up. Is there any momentum stemming from that CISA deal you recently secured? Are any federal organizations using it as a SIEM reference point? And has activity picked up because of that victory?
Ashutosh Kulkarni
executiveYes. Thank you. For us, it has already been a tremendous success. I believe we also noted it in our press release. That SIEM as a service with CISA keeps expanding, and we saw more agencies joining even during Q3. So my expectation is that the CISA win will merely mark the start of numerous agencies adopting that service across the coming several quarters; and at its core, CISA is regarded as the main agency in charge of cybersecurity for the civilian government within the United States. And that — an endorsement of that sort carries significant weight. So it's a really thrilling win. As I said, we will reap the benefits from it across many quarters and many years ahead.
Brent Thill
analystAsh, regarding CRPO — 15% in constant currency, and 15% the prior quarter. I suppose that's solid mid-teen growth, yet I believe the question on everyone's mind is why a quicker inflection isn't showing up. I understand your true north sits at 20%. The figures appear to back the idea that reaching 20% is possible. Still, I'm curious how you get from here to 20%, and maybe why a somewhat bigger AI tailwind isn't appearing in the near term?
Navam Welihinda
executiveYes. Appreciate that question, Brent. Let me begin. CRPOs have now passed $1 billion. At present we're growing 19%. RPO growth stands at 22%. That marks the strongest result we've witnessed in 2 years, and the progress being made makes us quite pleased. If you examine just the absolute dollar additions during the quarter, things are moving along extremely well; all of this signals the core drivers behind that CRPO growth, namely robust customer commitments, something we've discussed for a couple of quarters now, and this was one more quarter of solid — really strong sales execution producing robust customer commitments. As for the AI tailwinds we mentioned at Financial Analyst Day, those are visible to us right now, and they keep expanding as an increasing share of the $100,000 adopt AI workloads with us — or have adopted them. So we believe a solid, strong trajectory lies ahead from this point as AI penetration deepens across our $100,000 customer base.
Ashutosh Kulkarni
executiveOne more point on that, Brent: when you take a look at our full-year guidance for sales-led subscription revenue, it's clear the momentum in the business is still there. And honestly, for us, the midterm target we put out isn't our endpoint. It's a level we think we can move past. As you'll recall, we discussed 20-plus; that's genuinely how we view it. So with an increasing number of customers picking up our AI functionality, and given that those cohorts typically grow and expand more quickly, we feel really, really confident about how we're pacing toward that midterm, and we feel really confident that, as that traction keeps building, we're confident about even surpassing what we've described previously.
Howard Ma
analystI'd like to ask about cloud. I think this question is directed at Navam. First, let me add one caveat — I do value that you're deployment agnostic and that Q4 has fewer days. If I compare cloud revenue in Q4 against Q3 in FY '22 and before, the sequential increase was larger than what we saw in FY '23 through FY '25, and those years clearly were affected by industry-wide cloud optimizations plus Elastic's own go-to-market problems. That said, go-to-market execution has now gotten much better, and you have visibility into how big customers ramp consumption against their commits, including some of those $10 million-plus TCV contracts signed last quarter. So my question is: is there any reason the sequential cloud growth in Q4 wouldn't look more like those earlier years?
Navam Welihinda
executiveLet me begin with the same thing I always begin with — sales-led subscription revenue growth is the metric you should be watching to gauge us, both as a measure of how the company is doing and as the barometer of the company's success; I mentioned this in our prepared remarks too. There are several examples, this quarter included, of AI workloads sold as self-managed and deployed in the customer's cloud or in their hybrid environments. Sales-led subscription therefore grew a healthy 21% this year. If you isolate cloud and look at that figure, again, what is the sales-led cloud number — it rose 27% year-over-year this quarter; so traction is very good on the metric we care about, the metric that matters to us, namely sales-led subscription revenue. And the annual cloud number this quarter was also very good, at 27%. As for the quarters ahead, first, you have 3 fewer days, so that's 3 fewer days to focus on. The forward quarter is a risk-adjusted number. So comparing an actual against a guidance figure isn't really valid. But what I want to emphasize is that commitments are very strong and performance on sales-led is very strong.
De Shaun Fontenot
analystThis is DeShaun, filling in for Ryan MacWilliams. What I'd like to ask — it genuinely appears, judging by certain efforts we've been undertaking, that the count of agents plus AI services plus production has genuinely climbed across the last couple of months. So from your perspective, I'd like to hear what you're observing among your customer base. For instance, are the kinds of AI use cases you're witnessing expanding relative to what you saw perhaps 2 quarters back, and in what way is that affecting usage and spend among those customers?
Ashutosh Kulkarni
executiveYes, usage is broadening in the sense that we're witnessing an ever-greater variety of use cases involving AI. 8 quarters ago, most of what we saw centered solely on vector databases, vector search, hybrid search, semantic search; it was largely centered on chat-style interface type work. Today we're witnessing agentic workflows being assembled not only around what you'd ordinarily consider search-related workflows, but additionally around security workflows, around observability workflows. That's why we provided the stat regarding our overall customer count using us for a range of AI use cases past just vector database; and that encompasses things such as Agent Builder; that encompasses things such as Attack Discovery. And in those sorts of scenarios, people are attempting to automate their SOC workflows, their cybersecurity workflows for detection, for remediation. They're attempting to accomplish the same for SRE workflows around Observability. So the variety of use cases keeps growing. And as it grows, we perceive an opportunity, not only within our core search business, but also within the work we're doing in security and observability.
William Miller Jump
analystI want to say congratulations on the momentum you're seeing from sales-led efforts. Ash, during your prepared remarks, you brought up a competitive win against MongoDB; until somewhat recently, we hadn't been hearing all that much regarding this direct matchup between the two of you; so is it the case that you're running into MongoDB more often in bake-offs as customers try to develop AI applications? Or would you call that more of an isolated case?
Ashutosh Kulkarni
executiveNo, what happened here was that the customer had begun using that technology for a simple search application. They ran into some problems when it came to scaling it. And while attempting to put together a solution that scaled better, particularly for hybrid search, they came to understand that they required something capable of performing, and that was the customer win I referenced. Ultimately, the space where we usually tend to compete is unstructured data; we don't tend to encounter them as often. But every so often, you do come across these types of situations.
William Miller Jump
analystAnd could I quickly ask one more thing for Navam? Given that bigger contracts are playing a larger role in your go-to-market approach as you move up market, could you remind us of the way you deal with those sizable deals within your guidance process, plus any thoughts regarding seasonality in that area?
Navam Welihinda
executiveYes. On the seasonality side, my sense is that it simply tracks the usual enterprise pattern, meaning they turn out to be weighted more toward the tail end during Q3 and Q4. But we discussed big deals during the prior quarter. They show up each quarter. It's only that the amount of bookings grows larger near the close of the year. Regarding the way we manage it, my view is that this is a natural outgrowth of simply succeeding with our customers, especially the bigger ones inside the G2K. So we're glad to see it. Whenever we review our guidance and what we anticipate for the full year, we automatically apply a haircut to particular deals that might shift between quarters; that's the way we build it into our guidance, a risk-adjusted figure that doesn't actually assume everything will break in our favor.
Ashutosh Kulkarni
executiveGeorge McGuen here, filling in for Koji. Thanks for letting us ask questions today. What I'd like to know is this: when you talk with customers about how they plan to bring AI into their operations, how does the tone of those discussions compare with what you heard 12 months ago? And where would you place them right now, in terms of which inning they're in, relative to where they sat perhaps a year back on their adoption path with Elastic? The overall mood, without question, reflects more excitement about AI. There seem to be sufficient proof points at this stage showing AI's value across every sort of use case — code development, customer support, legal e-discovery, an enormous range of use cases spanning all functions. As a result, what we hear is less preaching and more a matter of assisting them in assembling these sorts of advanced agentic applications. So maturity has clearly arrived. As for how many of these agents an organization typically has in place, that figure remains at an early stage. Consider how many business processes and workflows AI could potentially automate — you have to stay realistic that we're still at an early stage, since AI is simply a rather potent and transformative capability. And what these LLMs can accomplish when it comes to reasoning can be turned toward a great many functions and a great many work processes. So our view is that the opportunity remains quite substantial and still lies in front of us.
Matt Calitri
analystIt's Matt Calitri here, filling in for Mike Cikos at Needham. Given all the progress you're making on search, such as the Jina reranking models, can you get customers to pay more? Or does better speed and accuracy serve mainly as a way to bring in new customers?
Ashutosh Kulkarni
executiveWe bill based on consumption, correct? As you're aware, we operate on a consumption model. Basically, anything you do on our platform gets metered, and it's essentially tied to compute, tied to storage, and so forth. For anything tied to LLMs or models, it's tied to tokens, and all our pricing is kind of public on our pricing pages. That said, yes, with these newer models, we monetize everything. And as usage keeps growing, as customers do more and more on our platform, that's what generates revenue for us.
Matt Calitri
analystUnderstood, that's quite useful. Now perhaps let's look at the guidance issue from another angle. You exceeded the 3Q guidance on a constant currency basis, and afterward you lifted the constant currency outlook for sales-led subscription revenue, yet for the full year guide you kept constant currency where it was; and I recognize the 3 fewer days plus the risk adjustment, but those factors would already have been reflected in the earlier guide. Could you just explain how that works, why it wouldn't have gone up?
Navam Welihinda
executiveSure. Honestly, it's very straightforward. Sales-led subscription revenue is the figure we focus on. This quarter we comfortably exceeded that figure, and the amount we raised exceeds the amount by which we beat. That reflects our view of what's going on in the business and the positive momentum we're observing on the sales line. So on the whole, we aren't overthinking it beyond the fact that we feel good about the forward momentum of sales-led subscription revenue, we exceeded the figure, and we're raising by more than we beat.
Ashutosh Kulkarni
executiveWe appreciate everyone taking the time to be with us on this call. At Elastic, we take real pride in what our business has achieved, and we're thrilled about what lies in front of us. Thanks.
Operator
operatorThat brings us to the end of this call. We appreciate everyone joining the presentation today. You're welcome to log off at this point.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Elastic N.V. transcript — plus 132,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.
Get the API View API docs →This call discussed
For developers and AI pipelines
Programmatic access to Elastic N.V. earnings transcripts and 132,000+ others is available through the
EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments,
full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.
More tools for stock research
OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in Elastic N.V. or any ticker on your watchlist.
StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.