MDB, 2026 2Q 컨퍼런스콜 요약

소개

MDB는 NoSQL 계의 강자로써 데이터베이스 관리 시스템으로 돈을 버는 기업이다. AI가 학습하기 위해서는 데이터가 필수적이고, 그 데이터를 가공하기 쉬운 형태로 보관하는 것을 데이터베이스라고 하는데, 데이터베이스를 관리하는 시스템을 구독형으로 판매하며 돈을 버는 기업인 셈.

아직까지 영업이익 흑자를 내보지 못한 기업이기에 투기적이다고 할 수 있는 종목이며, 성장세에 따라 주가가 크게 급등락하는 종목이다. 이번 실적발표로 주가가 무려 37% 넘게 급등하며 시장이 다시 한 번 MDB에 주목하는 계기가 되고 있어, 이 기업의 컨콜을 읽어 보기로 했다.

요약

MDB가 좋은 투자기회라는 생각이 들었다. 이유는 다음과 같다.

1. MDB가 노력했건, 시대의 흐름이건, 대형 고객들이 MDB로 이동하고 있다. 즉, SQL을 버리고 NoSQL인 MongoDB를 채용하고 있는 것.
2. 소규모 AI 스타트업들도, 초기에는 Postgres를 사용하다가 스케일링을 위해 MongoDB로 이동하고 있다.
3. AI 테마는 아직 초기 상태다.

이에 따라 주가도 그만큼 오르고 있다. 그러나 단기적으로 너무 빠르게 오른 주가는 결국 일시적으로 조정을 받을 것이기 때문에, 거래량이 크게 동반된 장대 양봉 근처로 주가가 조정 받으면 진입해볼 기회가 생길 것 같다.

하이라이트

MongoDB, Inc. (NASDAQ:MDB) Q2 2026 Earnings Conference Call August 26, 2025 5:00 PM ET

Company Participants

a – Corporate Participant
Dev C. Ittycheria – President, CEO & Director
Michael J. Berry – Chief Financial Officer

Conference Call Participants

Aleksandr J. Zukin – Wolfe Research, LLC
Brad Robert Reback – Stifel, Nicolaus & Company, Incorporated, Research Division
Bradley Hartwell Sills – BofA Securities, Research Division
Ittai Kidron – Oppenheimer & Co. Inc., Research Division
Jason Noah Ader – William Blair & Company L.L.C., Research Division
Kasthuri Gopalan Rangan – Goldman Sachs Group, Inc., Research Division
Michael Joseph Cikos – Needham & Company, LLC, Research Division
Raimo Lenschow – Barclays Bank PLC, Research Division
Rishi Nitya Jaluria – RBC Capital Markets, Research Division
Sanjit Kumar Singh – Morgan Stanley, Research Division
Sitikantha Panigrahi – Mizuho Securities USA LLC, Research Division
Tyler Maverick Radke – Citigroup Inc., Research Division
Brian Denyeau – ICR Inc.

Dev C. Ittycheria

Thank you, Brian, and thank you to everyone for joining us today. Before discussing our strong quarter, I want to remind everyone about our upcoming Investor Day, which will take place on September 17 at the Javits Center in New York City during our .local conference. We’ll spend a day discussing the investments we’re making to drive durable growth and margin expansion and our view of the future. I look forward to seeing you there.

Now on to Q2. I’m pleased to report another strong quarter as we continue to execute against a large market opportunity. Let me start with our results before giving you a broader company update. We generated a revenue of $591 million, up 24% year-over-year and above the high end of our guidance. Atlas revenue grew 29% year-over-year, representing 74% of total revenue. We delivered non-GAAP operating income of $87 million for a 15% non-GAAP operating margin, and we ended the quarter with over 59,900 customers.

Atlas performance was strong, accelerating to 29% year-over-year growth, up from 26% in Q1. Our customer additions were also robust. We have added over 5,000 customers over the last 2 quarters. These results reflect the strength of MongoDB’s platform, our flexible document model, expanded capabilities like search and vector search, enterprise readiness and the ability to run anywhere. Many of our recently added customers are building AI applications, underscoring how our value proposition is resonating for AI and why MongoDB is emerging as a key component of the AI infrastructure stack.

At the same time, we significantly outperformed on operating margin demonstrating that we can drive durable revenue growth while expanding profitably. In short, our results show that customers are choosing MongoDB. Let me tell you why. First, MongoDB is an enterprise-ready database capable of meeting the most stringent enterprise requirements. Over 70% of the Fortune 500 as well as 7 of the 10 largest banks, 14 of the largest 15 health care companies, 9 of the 10 largest manufacturers globally are MongoDB customers.

MongoDB is a battle-tested enterprise platform relied on by some of the most sophisticated and demanding organizations of the world in part because of our strong enterprise posture across security, durability, availability and performance. Atlas enable one of the world’s largest automakers to overcome Postgres scalability and flexibility limits while reducing complexity. The company’s management console tracks over 8.5 million vehicles requiring a modern schema to handle both structured and unstructured data, something Postgres could not handle.

Ultimately, Atlas consolidated infrastructure, accelerated innovation and support the scale of millions of connected vehicles. Second, MongoDB is suitable for a broad range of use cases, including the most mission-critical and transaction-intensive applications. MongoDB has also supported full asset transactions for more than 6 years, ensuring strong consistency and data integrity at scale. This is why some of the world’s most demanding transactional workloads run on MongoDB today. For example, Deutsche Telekom selected MongoDB Atlas as the foundation for its internal developer platform, which includes mission-critical workloads like contract management, device purchases and billing for 30 million customers. With 90 Atlas clusters managing over 60 million customer records, Deutsche Telekom’s customer data platform now handles 15x the concurrent logins of legacy systems.

By consolidating these high-volume transaction-intensive applications on MongoDB, Deutsche Telekom has improved resiliency, accelerated innovation and delivered a step change in customer engagement. Third, MongoDB has redefined what’s core for the database by natively including capabilities like search, vector search, embeddings and stream processing. Comparing MongoDB to another database like Postgres is not an apples-to-apples comparison. Take a global e-commerce application that manages inventory and order data while enabling product discovery through sophisticated search across millions of SKUs.

The choice for this application is not between MongoDB or Postgres, is between MongoDB or Postgres plus other offerings like Pinecone, Elastic and Cohere for embeddings. MongoDB’s complete solution allows developers to spend less time stitching together and maintaining a patchwork of disparate systems and more time building differentiated functionality that drives the business forward. For example, Agibank, Brazilian neobank with 2.7 million active customers migrated their content management system, storing customer records from Postgres to Atlas. As data volumes grew, Postgres’ inflexibility and task execution latency drove performance issues and the database like sophisticated secondary indexes and full-text search, hurting sales of core offerings such as loans, insurance and card approvals. Agibank was constantly updating the database and manually scaling infrastructure, which is both time-consuming and error prone. With Atlas, Agibank gained a resilient flexible system that handle rising demand and support new services, delivering nearly 5x better performance and 90% lower cost, all with no outages.

Fourth, MongoDB is emerging as a standard for AI applications. Over the last few quarters, we’ve seen a strength in our self-serve channel, driven in part by AI native startups choosing Atlas as the foundation for their applications. In the enterprise segment, adoption is real but early. Much of the activity today centers on employee productivity tools and packaged ISV solutions. Enterprises are still in the very early stages of building their own custom AI applications that will transform their business. We consistently hear from customers that when teams try to scale from vibe-coded prototypes built on relational back ends to enterprise-grade deployments, these platform quickly hit limits in flexibility, scalability and performance. Across startups and increasingly enterprises, our unified platform is resonating strongly.

In the enterprise segment, a leading electric vehicle company chose Atlas and vector search to power its autonomous driving platform. After testing vector search against Postgres pgvector for their in-vehicle voice assistant, they selected MongoDB for superior performance at scale and stronger ROI. They now rely on Atlas to handle over 1 billion vectors and expect 10x growth in data usage by next year.

DevRev, a well-funded AI-native platform with proven founders disrupting the help desk market built AgentOS, it’s a complete agentic platform that autonomously handles billions of monthly requests on Atlas. DevRev accelerated development velocity, lower cost and scale globally with low latency by using Atlas. AgentOS also leverages Atlas Vector Search for semantic search enriching its knowledge graph and LLMs with domain-specific content.

Companies in nearly every industry and across every geography are choosing MongoDB because we deliver the features, performance, cost effectiveness and AI readiness they need, all-in-one platform. As we look ahead, we remain confident in MongoDB’s position to lead both the current wave of digital transformation and the next wave powered by AI.

With that, here’s Mike.

Michael J. Berry

Thanks, Dev. I’ll begin with a detailed review of our second quarter results and then finish with our outlook for the third quarter and fiscal year ’26. I will be discussing our results on a non-GAAP basis, unless otherwise noted. As Dev mentioned, we had a great quarter as we exceeded all of our guidance ranges and are increasing our full year guidance across the board.

Now on to the results. In the second quarter, total revenue was $591 million, up 24% year-over-year and above the high end of our guidance. Shifting to our product mix, Atlas revenue outperformed our expectations and year-over-year growth accelerated to 29% in the quarter and now represents 74% of total revenue. This compares to 71% in the second quarter of fiscal ’25 and 72% last quarter. We had an impressive Atlas growth quarter, which benefited in part from the strong start to consumption in May that we referenced on our last call as well as broad-based strength, especially in larger customers in the U.S.

Let me provide some context on Atlas consumption in the quarter. In Q2, Atlas consumption growth was strong and relatively consistent with last year’s growth rates. This drove the acceleration in revenue as well as the growth in absolute revenue dollars year- to-date for the first half of fiscal ’26. Turning to non-Atlas. Revenue came in ahead of our expectations in the quarter as we continue to have success selling incremental workloads into our existing EA customer base.

Non-Atlas ARR which reflects the underlying revenue growth of this product line without the impact of changes in duration grew 7% year-over-year. In addition to the good underlying trends in non-Atlas, in Q2, we also benefited from more multiyear deals than expected, reflecting our customers’ desire to commit to building with MongoDB long term. Approximately half of the non-Atlas revenue outperformance versus guidance was attributable to multiyear outperformance.

We had another strong quarter for customer adds in the second quarter as we grew our customer base by approximately 2,800 sequentially, bringing the total customer count to 59,900, which is up from over 50,700 in the year ago period. This quarter, we incorporated new customers added from the Voyage AI acquisition to our customer count representing 300 of the 2,800 added. The growth in our total customer count is being driven primarily by Atlas, which had over 58,300 customers at the end of the quarter compared to over 49,200 in the year ago period. It is important to keep in mind the growth in our Atlas customer count reflects new customers to MongoDB in addition to existing EA customers deploying workloads on Atlas for the first time.

Of our total customer count, over 7,300 are direct sales customers, a decline of 200 customers sequentially and flat year-over- year. These metrics are largely due to our decision to reallocate a portion of our go-to-market resources from the mid-market to the enterprise channel, starting in the second half of last year. This does not impact our total customer count, but as an output of fewer self-serve originated customers being elevated to our direct sales team as we move upmarket. In Q2, our total company net ARR expansion rate was approximately 119%, which is consistent with recent quarters. We ended the quarter with 2,564 customers with at least $100,000 in ARR, representing 17% growth versus the year ago period.(중소 고객보다는 대형 고객에 집중)

Moving down the income statement. Gross profit in the second quarter was $436 million, representing a gross margin of 74%, which is down from 75% in the year ago period. Our year-over-year gross margin decline is primarily driven by Atlas growing as a percent of the overall business. Our income from operations was $87 million for a 15% operating margin compared to 11% in the year ago period. We are very pleased with our stronger-than-expected margin result — operating margin result, which benefited mainly from our revenue outperformance.

Additionally, I’d like to provide a little context on the modest restructuring we undertook in the quarter. It impacted less than 2% of employees and resulted in approximately $5 million of onetime charges, which we have excluded from our non-GAAP financials. This action is consistent with the key priorities I outlined for you last quarter, to identify ways to both reallocate existing spend to higher ROI opportunities and be more disciplined about incremental spending. We are focused on running an efficient, scalable business that supports growth in revenue and profitability to drive long-term shareholder value. Net income in the second quarter was $87 million or $1 per share based on 87 million diluted shares outstanding. This compares to a net income of $59 million or $0.70 per share on 84 million diluted shares outstanding in the year ago period.

Turning to the balance sheet and cash flow. We ended the second quarter with $2.3 billion in cash, cash equivalents, short-term investments and restricted cash. During the quarter, we spent $200 million to repurchase approximately 930,000 shares which was under our previously announced $1 billion total share repurchase authorization. Operating cash flow was well above our expectations at $72 million and free cash flow was $70 million, which compares to negative $1 million and negative $4 million, respectively, in the year ago period. Our strong cash flow results were driven primarily by strong operating profit and higher cash collections.

Before turning to our outlook in greater detail, I’d like to share the key points driving how we are looking at the rest of fiscal year ’26. Number one, we are raising our expectations for revenue based on our confidence in Atlas as well as a strong performance in the first half of the year, providing a higher starting point for Atlas heading into the second half. Number two, we are increasing our operating margin guidance by 150 basis points at the high end, reflecting our strong Q2 performance and continued focus on margin

improvement. And number three, we are raising our operating margin guidance while still continuing to make incremental investments for growth with a focus on R&D and developer awareness.

Now moving on to our full year guidance. I’d like to provide some incremental comments on our expectations. First, as we discussed, we had a strong start to the year and are confident in our ability to drive continued revenue and profitability growth. We are raising our full year revenue guidance by $70 million, including the $38 million outperformance in Q2. This reflects the strong Q2 consumption benefiting revenue in the second half and our continued confidence in Atlas growth. All in, this implies mid-20s percentage growth for Atlas in the second half of the year.

Second, incorporating our strong performance in the first half, we expect non-Atlas subscription revenue will now be down in the mid- single digits for the year compared to our prior expectation of high single-digit decline. We also expect a headwind from multiyear license revenue for fiscal ’26 to now be $40 million due to the Q2 outperformance compared to our prior expectation of approximately $50 million. Please note, we expect non-Atlas ARR will continue to grow year-over-year.

Finally, we are raising our expectations for operating margin to 14% at the high end, up from 12.5% in our prior quarter guidance.(Non-GAAP인 듯) This reflects the better-than-expected revenue performance, the impact of our more disciplined approach to investing for growth and our increased focus on efficiency. For fiscal year ’26, we now expect revenue to be in the range of $2.34 billion to $2.36 billion, an increase of $70 million from our prior guide. We are raising our non-GAAP income from operations expectations by $44 million and are now targeting a range of $321 million to $331 million and non-GAAP net income per share to be in the range of $3.64 to $3.73 based on 87.4 million diluted shares outstanding. Note that the non-GAAP net income per share guidance for the third quarter and fiscal year ’26 assumes a non-GAAP tax provision of 20%.

Moving on to our Q3 guidance, a few things to keep in mind. First, we expect to see a low 20% year-over-year percentage decline in the non-Atlas business after the strong multiyear outperformance we experienced in Q3 of fiscal year ’25. As a reminder, Q3 of last year was our strongest multiyear revenue quarter and is the largest portion of the multiyear headwind. Second, we expect operating margin will be lower than in Q2, primarily due to the expected sequential decline in non-Atlas revenue, which is very high-margin revenue. In addition, it is also impacted by the timing of operating expenses, specifically R&D hiring and seasonality of our marketing investments.

With that context, I will now turn to our outlook for the third quarter. For the third quarter, we expect revenue to be in the range of $587 million to $592 million. We expect non-GAAP income from operations to be in the range of $66 million to $70 million and non-GAAP net income per share to be in the range of $0.76 to $0.79 based on 87.7 million diluted shares outstanding. To summarize, we had a very strong quarter. We are pleased with our ability to drive revenue growth across the business and increase our operating profit expectations. We remain incredibly excited about the opportunity ahead, and we’ll continue to invest responsibly to drive long- term shareholder value.

I would also like to take a moment to extend a warm welcome to Jess Lubert, our new Vice President of Investor Relations, who started with us yesterday. Jess joins us from Juniper Networks, where he led their Investor Relations effort, including the most — including most recently helping the company navigate the acquisition by Hewlett Packard Enterprise. We’re excited to have him on board and eager to see the impact of this work. Last but not least, we look forward to seeing many of you in a few weeks at our Investor Day. Please reach out to our Investor Relations team at ir@mongodb.com with any questions. With that, we’d like to open it up for questions. Carmen, take it away.

Question-and-Answer Session

Operator

[Operator Instructions] Our first question is from Sanjit Singh with Morgan Stanley.

Sanjit Kumar Singh

Congrats on a heck of a quarter in Q2. I wanted to dive into some of the drivers into Q2. When I look at the acceleration of Atlas, which has now accelerated for 2 quarters in a row, and I kind of just look at the sequential dollar adds, I add that up, more than $40 million in Q2, which is kind of the strongest sequential dollar adds we’ve seen in quite some time in what’s been a pretty sober sort of cloud spending environment. So I was wondering if you could give us some sense of the drivers of the strong sequential adds this quarter, I know you pointed to May. But if anything you can give us from like a workload perspective or any other new factors, maybe the workloads from last year starting to ramp. I’d just love to understand that trajectory a little bit better.

Dev C. Ittycheria

Yes. Sanjit, thanks for the question. So clearly, we’re really pleased by the quarter and really pleased by the accelerating growth in Atlas. I would say a lot of it was due to the workloads that we acquired over the past year, especially with our move up market that are growing faster and becoming bigger than previous workloads we’ve seen. So I think the move-up market is really paying off. And what we’re also seeing is that there’s a great uptick of some of the other capabilities we offer like search and vector search that are also adding to that growth of those workloads. And then as we mentioned, we also acquired a ton of new customers, obviously, the self- serve customers that tend to spend less on a per customer basis, but we obviously have added lots of customers over the last 6 months. And I think that’s also helping drive some of the growth.(대형 고객들 위주로 목표를 바꾼 것이 주효했다.)

Sanjit Kumar Singh

Yes, that’s great color. I wanted to follow up on the go-to-market side. Over the last couple of years, you’ve been sort of tinkering and optimizing the go-to-market organization across sort of territory investment, but also sort of quotas and moving to incremental consumption. Could you give us an update on the state of operations for the sales force today? And in some sense, if I look at the customer adds, it seems like things are humming quite well. But just want to understand how — like what’s the state of the organization there, that would be really helpful.

Dev C. Ittycheria

Yes. Sure. So nothing really has changed. We’re just doubling down on what we said previously. We are moving up markets. We’re focusing our high-end sales force focus on the most sophisticated and demanding customers. These are typically enterprise customers all around the world. And then we’re using our self-serve channel to better serve the SMB market. I know there are a lot of questions about were we kind of abandoning the self-serve — the early-stage market by this move. And I think the results over the last couple of quarters have shown that we are not. I think we’re just becoming much more effective in serving that market while also being very effective in growing our wallet share in these larger accounts. So we’re really just continuing with the strategy that we articulated before. And obviously, we’re pleased with the results.

Operator

Our next question is from Raimo Lenschow with Barclays.

Raimo Lenschow

First of all, congrats to Jess, all the best. Two quick questions from me. Staying on that theme of self-service, that acceleration, Dev, obviously, you changed things around, but it kind of — it’s accelerated despite kind of you actually moving up market. Like can you help us understand that and what’s driving that a little bit? And then I had one follow-up for Mike.

Dev C. Ittycheria

Yes. I mean, clearly, the output metrics look really good. But I would say the work around self-serve began — has been going on for a while. The team is really good at running experiments using a data-driven approach to figure out what’s working, to figure out what’s not working. A new motion that we’re also doing that’s showing good results is going after SQL developers who don’t really know MongoDB and attracting them to our platform, really helping them understand the value proposition of MongoDB, even running like things like office hours where we spend time with SQL developers to explain the benefits of modeling data on a document database. And all these experiments and tactics that we’re doing, which are very data-driven are really paying off. And May Petry used to run that group is now our CMO, and she had a strong team under her, and we feel really good about what that self-serve team has been doing. But again, we don’t want to declare victory too early, but obviously, we’re very pleased with the results.(SQL 개발자에게 집중하여 MongoDB의 장점을 알린 것이 중요했던 것 같다.)

Raimo Lenschow

Yes. No, that’s really nice to see. And then Mike, thanks, first of all, for all the excellent disclosure, the ARR for the non-Atlas or EA part is kind of really helpful. If you think about the — I get the logic around the renewal cohorts, especially Q3, but am I doing the math correctly that actually next year, that part of the business looks more interesting because the cohort looks better. Like just trying to get your idea or — and maybe you might not even give it to us because you just do ARR.

Michael J. Berry

Sure. So thanks for the question. So I’m going to hold that answer until we get to Q3 of next year because it kind of depends on what happens in Q3 of this year. So the one thing is, as we talked about, the big impact in Q3 of this year is the multiyear. We’ll see how it comes back next year, but it really depends Raimo on how we do in Q3 this year.

Operator

And our next question comes from Tyler Radke with Citi.

Tyler Maverick Radke

And nice job on the Atlas growth. I wanted to dig into the AI commentary that you had, Dev. Obviously, last quarter, you talked about cursor, which obviously is ramping up significantly in terms of their ARR. And I think you called out many examples this quarter, including an autonomous vehicle company. It sounds like expecting pretty significant growth there. But how much of that is playing into the Atlas strength that you’re seeing here in the quarter? Any way to quantify that cohort or use cases, whether it’s vector search or maybe even if you throw in Voyage, just help us understand if that’s starting to move the needle because it sounds like there’s some pretty high profile wins in there?

Dev C. Ittycheria

Yes. So thanks for the question, Tyler. While we’re adding thousands of AI native customers, I will tell you that the growth that we delivered this quarter was not material to that growth. Growth is really driven by our core business and our core customer base. And so — and while we’re very happy with the AI customers increasingly choosing MongoDB, it was not a material mover of the needle for our growth.(자율주행 관련 기업으로부터의 수요가 성장의 주요 동력은 아니었다.)

Tyler Maverick Radke

Great. And then a follow-up on the migration opportunity. I know you’ve been investing in Relational Migrator. You’re working with companies like Cognition to accelerate the code migration opportunity. And you’ve seen professional services ramp up a little bit. But where have you started to see sort of the time to migration or replatform improve a bit? Just anything you could share in terms of that migration opportunity if that’s started to improve in terms of velocity or size of workload migration would be helpful.

Dev C. Ittycheria

Yes, sure. So yes, we’re super excited about what we call app modernization or legacy app modernization. You’ll hear a lot more about this at Investor Day in September, Tyler. But what I will say you is that the value proposition is very clear. Customers are very, very motivated to try and modernize these legacy systems for a wide variety of reasons. We are seeing a lot of progress. We’ve actually

brought in a new leader — new product leader, who brings a lot of depth and scale, especially around AI to help us build the tooling to leverage AI to really drive more automation in terms of how we analyze and refactor the code.

We brought in a new leader last quarter to really help drive the delivery and the go-to-market efforts around app mod. So we’re definitely beefing up resources. And I would say that we’re investing a lot in product and there’s a lot more to do. And I would say, this is something that we’re very excited about, but it will drive more of our longer-term growth, less — it won’t be as pronounced in terms of this year, but we’re very, very excited about the opportunity, and we’re definitely — we’ll spend more time discussing this and what we’re actually doing on the product side in September.(데이터베이스를 기존 sql 과 같은 시스템에서 mongoDB로 마이그레이션 하려는 수요가 매우 강력하다.)

Operator

One moment for our next question. It comes from Jason Ader with William Blair.

Jason Noah Ader

Dev, I was hoping you could talk about some of the kind of latest industry developments just on the technology side, in particular, I’m thinking about Lakebase from Databricks and then DocumentDB in the Linux Foundation. Can you just comment on both those things? And how they might impact MongoDB and how you differentiate?

Dev C. Ittycheria

Yes. So let me tackle them one by one. Clearly, what we are seeing is that the strategic high ground for AI, especially when it comes to inference is OLTP. So we talked about this on the last call where some companies that acquired early-stage OLTP start-ups. And what it really spoke to and those companies had spoken about their organic efforts to build an OLTP platform. And I think what it spoke to was the fact that they building an OLTP platform that’s ready and mission-critical and enterprise can serve the most demanding requirements of enterprises is not trivial.

And I think they basically threw in the towel and decided to do these acquisitions. And what it just reinforces that OLTP is the strategic high ground for AI, and we believe that if now customers are going to be choosing what OLTP platform that they want for AI, just given our architecture, just given the fact that we have a durable architectural advantage in terms of JSON support, which addresses messy, complicated and highly interdependent and constantly changing data structures. The fact that we integrated search and vector search, I think, really helps us position going after AI.

With regards to your second question around the Linux Foundation, I think what this really also suggests — shows is that real JSON is much more important now with AI than ever before and the clones and bolt-ons that have traded off features and performance and developer experience have just not met customer expectations. And candidly, what I see this is that the hyperscalers are investing less and really handing off to the open source community to kind of really take on the bulk of the work in terms of product development. Our hyperscaler partnerships remain strong. And I think we have the right open source model where we can balance the access to free software while preserving the ability to both generate and capture value.(답변을 요약하면, 우리가 최고다.)

Jason Noah Ader

Great. And then just one quick follow-up. Why don’t we hear so much about Postgres adoption for AI start-ups. You talked about the success you guys are having. But if Postgres has the disadvantages that you’ve talked about multiple times, scalability, JSON support, how come we hear so much about that kind of at least in the early stages of AI?

Dev C. Ittycheria

Yes, that’s a really good question. And I think it’s important to understand, and we spend a lot of time — we have now invested in the team in the Bay Area that spends a lot of time with the startup community. What’s become clear is a lot of these startup founders don’t think that hard about their database choice, they kind of go with what they know. And what we are seeing is that as some of these startups are scaling, they’re running to real scaling challenges with Postgres. And what — and we’ve talked about this in the past, like when you add a JSON — when you use JSONB on Postgres, a 2 kilobyte document or bigger starts really creating performance problems because Postgres has to do something called off-row storage, which creates enormous performance overheads. And so the developers need a platform that can handle structured, semi-structured and unstructured data, they need obviously a platform that performs well, and they need a platform that can scale as they grow.

And what we’re hearing clearly from the startup community that Postgres, in many cases, is not scaling for them, and they’re now coming to us. And so we feel really good about our position. But the reality is that a lot of these AI founders kind of start with what they know or what they’ve used in the past and only when the business starts scaling, do they start recognizing the challenges. And we realize we need to do more developer education and do more work. And so we’re investing a lot in the startup community. We’re running a big event in October in San Francisco with a big hackathon, and we’re inviting lots of customers to participate. But that’s just the start of a meaningful investment we’re making in the Bay Area and the AI startup community to rethink their decisions around just going with what they know.(Scale이 커지면 Postgres는 비효율적이기 때문에, mongoDB가 우월하다.)

Operator

One moment for our next question that comes from Mike Cikos, with Needham.

Michael Joseph Cikos

I just wanted to come back to Atlas specifically. And Mike, I appreciate last quarter, you gave us some very granular color around Atlas trends. I was hoping we could get an update on how Atlas trends played out this quarter or just at the very least why we did see such broad-based strength from large customers this quarter?

Michael J. Berry

Sure. Thanks for the question, Mike. So when we talk about consumption in the second quarter for Atlas, as we talked about, it perform well, grew 29% year-over-year. As we talked about, Mike, the consumption growth were relatively consistent with last year. And as we talked about on the last call, we started out with a strong May, and we saw a broad-based strength across most of the geos and segments. So nothing to call out there, but we did see notable strength in the larger customers in the U.S.

And if we dive deeper on that one, as Dev talked about, we are seeing some workloads from our larger customers grow for longer and expand more than we have seen in the past, so that’s good. While there’s many moving parts in the consumption business, we also expect that there is benefit from our go-to-market changes and given the preponderance of our strategic accounts being in the U.S., no surprise that we saw that growth mostly in the U.S. And then lastly, Mike, there is some benefit from comparing it to a little slower growth in Q1. So that would be the detail on Q2 as it relates to consumption growth.

Michael Joseph Cikos

And if I could just squeeze maybe one more in on the outperformance that we saw this quarter from the multiyear deals. And maybe I’m just misunderstanding here, but my assumption was the reason we were facing this outperformance was really tied to the fact that in prior years, we’ve had some pretty big deals on the multiyear front. And so to see some of these deals come in this year, is that a function of customers renewing earlier, which is helping fill that larger divot that we previously expected? Is that a fair assumption? Or can you help me think through that a little bit more?

Michael J. Berry

So thanks for the golf analogy. No, it did not fill the divot. So in Q2, it was really — it was good underlying strength in ARR growth and then greater-than-expected multiyear. There were really no pull forwards, Mike. And this is a hard business to forecast because sometimes even customers don’t know whether they’re going to opt for an annual renewal or a multiyear. So it was — there was no pull forwards and there was nothing out of the ordinary. Very importantly, we left the non-Atlas assumptions consistent with our last guidance. Hence, pulling down the multiyear headwind from $50 million to $40 million. And again, nothing to call out on Q2, no pull forwards, and there were really no large multiyears in there. It was just across a good subset of customers.(매출을 당겨와 실적이 잘 나온게 아니다.)

Operator

Our next question comes from the line of Alex Zukin with Wolfe Research.

Aleksandr J. Zukin

And I’ll echo the congrats on truly amazing quarter. I guess, Dev, when you think about the AI comments that you’ve talked about both in the press release and in the call, maybe just a little bit more nuance on the use cases, not necessarily that you’re seeing kind of

contribute materially today, but the differentiation of the platform that you’re able to incrementally take market share as it becomes available, both in net new kind of AI native companies, but also in some of your larger existing companies or customers that are starting to modernize for this kind of conversational or AI-native era. Where are you seeing the most momentum in terms of workload construction and scale? And when do you think we should expect to kind of actually start seeing that contribute more materially to the growth in consumption?

Dev C. Ittycheria

Yes. So thanks for the question, Alex. A couple of points. Again, we’re very pleased with the results of this quarter, but I would say the AI cohort was not a material driver of the growth. That being said, what we are seeing is a lot of customers very, very interested in our architecture. Let me again walk through why. One, we’re a JSON database. JSON is the best way to express and model the complicated and messy and highly interdependent and constantly evolving data structures that you have to deal with in the real world. So that’s point number one. So it’s much easier to do that in MongoDB than to do that on some kludge kind of set up on top of a relational database.

Second is that we integrate search and vector search. So you can do very sophisticated things to what people call hybrid search and retrieval, you can do very sophisticated things in finding information quickly, which is a very unique differentiator for us. So what this means that rather than stitching together multiple systems, you can do this all on MongoDB, so it becomes less complexity and lower cost. The third thing is that we’ve now embedded Voyage models on our platform, right? So the — if you control the embedding layer, you sit at the gateway of needing of AI, right? What the embedding models do is really a bridge between a company’s private data and the LLM. So that becomes really important because the better the quality of the embedding model, the better the quality of the signal of your own data.

So that reduces things like hallucinations or just bad outputs. And so customers are now — as people start caring more and more about like higher state use cases, they really want to ensure those outputs are high. And the fact that it’s part of our platform, we can enable you to do auto embeddings. It becomes an incredibly compelling feature. In terms of the market, what I would say is that the enterprise uptake of AI is still early. I’ve said this for a couple of years now, and I think a lot of people were a little skeptical of what I said, but it’s proving to be true. As we predicted, like the lack of skills and the lack of trust with AI systems is kind of slowing — people are being have very cautious about deploying AI.

Where it is being deployed is really on end user productivity, whether it’s developers with cogen tools or business users using tools to summarize documents, extract data or things like deflecting tickets from people to systems with like conversational AI. I think you are starting to see the first steps in people deploying agent-based systems, and I can talk a little bit about that, but that’s still very, very early. We’re seeing small ISVs, some of them are taking off, who are really driving most of the impact. But the real enduring value will come. When you talk to a customer today, most of them when you ask them is AI really transforming a business, they will say no.

Yes, we’re seeing some productivity gains here and there, but it’s not really transforming my business. I think the real enduring value will come when they build custom AI solutions that can truly transform the business, whether it’s to drive new revenue opportunities or dramatically reduce their existing cost structure. But we’re really pleased. I mentioned this electric car company that’s very tech savvy that’s using MongoDB, I should mention one of the fastest-growing startups in the Bay Area has bet big on MongoDB. DevRev, the company going after the help desk space has built their own agentic platform on MongoDB. So we feel really good about what this all portends for the future. But as I said, it was a small part of our growth this quarter.(아직 AI의 인프라 구축 정도의 수요만 있을 뿐이고, 진정한 AI 수요는 아직 시작도 안됨.)

Aleksandr J. Zukin

Very helpful. And then maybe if I could just sneak one in for Mike. You’ve been kind of saying from, I think, the first day you started about how the margin profile of this business, it’s not an or, it’s an and, and it’s clearly coming through in both the growth acceleration, but also the meaningful margin outperformance. As you think about sustaining this kind of accelerating pace and investing in things like the Bay Area start-up community, how are you finding that balance, that and versus or balance that quite frankly is elusive to a lot of companies that are doing what you guys are doing?

Michael J. Berry

Well, I think it’s the funniest part of my job, quite frankly. So I would give kudos to not only the management team, but everybody at MongoDB to really jump in this. I think that this has been a company-wide effort. And as we look forward and as we talked about, Alex, the #1 driver of margin expansion for Mongo is the revenue growth. So those 2 are directly connected. It’s a great business

model, where when we can grow Atlas in the 20%-plus range and then keep that ARR of EA in that single digit, it generates a ton of gross profit that funds a lot. And the team has done a really — has done a great job of making sure that we are investing in growth, that we go back and look at what we’re doing, making sure that it’s driving growth, if it’s not, then we have an open discussion about whether we should reallocate. So I felt good about it when I started. Candidly, I feel better about it 90 days later.

Operator

Our next question comes from Kash Rangan with Goldman Sachs.

Kasthuri Gopalan Rangan

It’s always tough to go after, Alex, because he asked such good question, but that’s not going to stop me. So Dev and Mike, congratulations on the quarter. It’s super interesting, you were talking about how some Silicon Valley AI startup founders don’t have time to think about databases, but our good friend, Dheeraj, at DevRev seems to have made a wise choice here. So as you set encampment up in the Bay Area and start to evangelize the need for an Atlas consumption AI savvy database, how do you reconcile that with the fact at the same time enterprise is where we really saw the bread-and-butter value proposition of Mongo resonate. So could what is happening with DevRev be a leading indication of what’s going to happen in the enterprise? Because we’ve all much to your observation, not seen much of a productivity impact in the enterprise because of AI at the business level. And so what could be that unlock is, what are folks like Dheeraj doing correctly that is a — could be a precursor, if it is, for what is to come in the enterprise?

Dev C. Ittycheria

Yes. So Kash, thanks for the question. Obviously, I have so much respect for Dheeraj, he built Nutanix into a real great business, and he’s going to do the same at DevRev. I will tell you that the AI cohort, as I said earlier, is — was not really material to our growth. So I think these are all customers kind of earlier in their journey. So what we are seeing, what’s driving the growth right now is these large enterprises with workloads that we acquired both last year and this year that are really driving the growth, especially the Atlas growth that we saw this quarter. And what that really confirms is that our move up market made sense. The quality of those workloads, the durability of the growth, they become — grow for longer and become bigger than what we’ve seen in the past is really making us feel good about that decision.

And to juxtapose that, we also obviously decided to double down on self-serve to better serve the small- and medium-sized business market, and that’s also become — obviously becoming more and more effective and just given the number of customers that we’ve added over the last 6 months. So we feel like those motions are working well in concert together. And we feel like this allows us to be much more efficient about how we go to market. And there’s also going to be continued more work to continue to drive efficiency even better, but we also are investing for the long term. And so we’re just constantly debating those decisions internally, but we feel good about what’s working. And we feel good that like someone like Dheeraj is betting early on MongoDB because that’s a good signal for other founders who are thinking about doing the same.(열심히 하고 있으니 두고 보자는 이야기)

Kasthuri Gopalan Rangan

Awesome. We’ll drill into this more in a couple of weeks when we see you in San Francisco.

Dev C. Ittycheria

Absolutely.

Operator

One moment for our next question, is Brad Reback with Stifel.

Brad Robert Reback

The 7% EA ARR growth seems fine. I’m assuming you’re not satisfied with single-digit growth there. Dev, any sense of where we should think about that longer term?

Dev C. Ittycheria

Clearly, EA is a large enterprise motion. And what we’ve seen is that it’s typically less new customers choose EA and it’s more of our existing customer base who have a mix of EA and sometimes they then also start deploying Atlas. I think one thing that’s becoming more and more clear is that customers are becoming much more thoughtful about like how they think about using deployments on- premise versus using the cloud. I think 4, 5 years ago, there was a belief that everything was going to move to the cloud.

I think large enterprises have become much more sophisticated and nuanced in their thinking, and they believe that some workloads make sense to run on-prem and some workloads make sense to run in the cloud. And I think that’s where the MongoDB story becomes really attractive because the same code base can be used. And so it also gives them optionality for the future where they can move from on-prem to the cloud, and a lot of our EA customers have done that, either with new workloads and some existing workloads, and then they can also move from cloud to cloud. And they can also move back to on-prem if they choose to do so. So that optionality becomes a very powerful value proposition for our customers.

Operator

Our next question is from the line of Ittai Kidron with Oppenheimer.

Ittai Kidron

Great numbers and congrats to Jess and good luck in the new role. Dev, I wanted to dig into the AI opportunity again, but take it from the perspective of a go-to-market motion. Clearly, you can power a lot of AI use cases that are embedded with bigger platforms through a self-serve motion, but it sounds like to really capture the big workload opportunities, it’s going to have to be more of an enterprise push. So I’m kind of wondering how do you think about targeting the AI opportunity from go-to-market motion that doesn’t just fall into — if you’re a big enterprise, I’m going to send you to an enterprise salesperson and all the rest call our self-serve and do it yourself. Is there something a little bit more you think targeted perhaps that you need to take care in order to capitalize on this opportunity?

Dev C. Ittycheria

Yes. What I would say, Ittai, is that we’ve seen this movie before with the cloud, where some early-stage customers are growing very, very quickly, and then we then put dedicated sales focus on those accounts and they grew then even faster. So we’re clearly watching the market and when self-serve customers are to a point where they really need a higher touch kind of engagement model then we’re more than happy to do that. And we have a team that kind of helps transition customers from self-serve to more of a direct sales approach. And that has worked for us.

I think what we’ve learned is that, that line by which we actually engage high-touch model can move higher because we’ve become so sophisticated with self-serve that we can really serve customers for — early-stage customers for a long period of time. In terms of the enterprise, what I would say is what I’ve said earlier is that the enterprise is still quite early in their journey to AI. Most of the investments right now are more on end user productivity like developers using cogen tools, and what I call low stakes use cases.

In fact, I had 2 meetings today with 2 different leaders of 2 different financial institutions here in New York, and they both talked about what they’re doing in AI. They both admitted that they’ve kind of started with low stakes use cases, but their appetite to start doing more is increasing as they get more and more comfortable with the technology, and they’re quite excited to leverage MongoDB as part of that journey. But again, I think that’s kind of a microcosm into the enterprise market where I think they’re still quite early in their AI journey. And if you remember, this is something I’ve been saying for a while that most customers — most people overestimate the impact of a new technology like AI in the short term but underestimated in the long term. And I think we’re just in that classic journey right now.

Ittai Kidron

I appreciate that. And maybe as a follow-up, Mike, I just want to make sure I dig in a little bit into the non-Atlas business, the EA — predominantly EA business. Can you tell us roughly what percent of your customers here are multiyear deals versus just annual deals? I’m just kind of curious how — where we are now? And what was it, say, a year or 2 ago? And where do you think that mix is going to be a year or 2 from now?

Michael J. Berry

Yes. Thanks for the question. We don’t break out the percentage of customers on multiyear versus 1 year. What I would say is, in fiscal ’25, obviously, we saw a lot of larger multiyear deals, and you see that in the numbers. This year, we will always see multiyear deals. They haven’t been, I would call it, as large, so it’s more widespread. So we — that’s really the change that we’ve seen. We haven’t broken that out. I don’t think that it has changed much, especially over the year, as Dev talked about, it’s going to be a mix of Atlas and on-prem, and that mix has stayed relatively consistent.(25년과 비슷할 것.)

Ittai Kidron

When you look at the customers that are choosing multiyear deals, has anything changed in the way they think about the reasoning behind doing that versus not?

Michael J. Berry

No. Reasons are the same. It’s typically the — if it aligns with their long-term strategy, they want to be able to lock in the pricing and as everybody knows, hey, data has gravity, moving data around is not fun for everybody so they want to be able to lock in and guarantee their prices for that period of time.

Operator

Our next question comes from the line of Siti Panigrahi with Mizuho.

Sitikantha Panigrahi

Dev, I think some of the comments you were talking about the AI slowdown, and you heard about recent MIT report about 95% AI implementation not getting any kind of return. How do you see — what’s kind of do you think the inflection point? When we think we’ll start seeing some of the adoption of this AI? Like you said, they are testing, but what can trigger — I know you have been talking about a year ago, probably we are a few years out. But it’s good to see some of the traction. So how do you — first of all, what would be your view on that report? And how should we think about in terms of revenue contribution — material contribution from AI?

Dev C. Ittycheria

Yes. So I think it just comes down to the fundamental principles. I think customers need to feel, one, that the quality of the output of these AI systems is high. Obviously, AI systems are probabilistic in nature, not deterministic in nature. So you can’t always guarantee the output. You can hope that you’ve trained the models well. You’ve hoped that you’ve given it the right information, but you can’t always guarantee the output. So as I mentioned, I had meetings with 2 financial services customers earlier today, and both of them are still hesitant to roll out an end user-facing AI applications for those specific reasons. So it’s going to take a little bit of time for people to really get comfortable that they can really deal with the last mile issues and make sure that they don’t have any errors that potentially could be impacting their brand or really cause a lot of customer problems. So that’s point number one.

Then there’s issues around obviously the security of these systems, the stability and reliability of these systems, the scalability of these systems. As I mentioned, some of these early-stage companies are running into scaling issues with existing architecture, which is why they’re coming to us. So I think we’re just in that learning journey. I mean I don’t know if there’s going to be some massive tipping point. I think what we are seeing with the frontier models is that all these frontier models are kind of clustering around the same ballpark in terms of performance and the efficacy of their models.

So I think what’s going to start happening is how people start leveraging these insights to build what I call scaffolding around these frontier miles to address the needs of their business. Obviously, everyone’s talking about agents and people are very, very focused on essentially using agents to drive a lot of work. Agents require — if you think about — if you’re using agents, agents will use your systems much more intensely than humans will because they can do things much more quickly.

So you need platforms that can massively scale up and down, which is, again, a good sign and support indicator for MongoDB. So I think it’s going to take a little bit of time. It‘s going to take time to be comfortable with technology. It’s going to take time where people start with low stakes use cases and start gravitating to higher state use cases. So I don’t think there’s going to be some seminal inflection point. I think it’s just going to take time. But I think that time is coming.(아직 멀었다. 시장은 계속 커지는 중)

Operator

Our next question is from Brad Sills with Bank of America.

Bradley Hartwell Sills

I wanted to ask about some of the investments that you alluded to earlier that you’re making in R&D. How are you thinking about that? Is it incremental investments in some of these newer offerings like vector and streaming? Are there other new workloads that you’re thinking of addressing here? Would love to get some color on just where you’re investing in the stack?

Dev C. Ittycheria

Yes, sure. So we talked about the fact that R&D is a big part of our investment focus for this year. One, we came out with 8.0, which is the most performant release ever. So we’re already starting to see dividends of our investments in our platform, 8.1 is even better. And then we’re also making investments in the expansion parts of our platform. What I will say is we’re going to go into a lot more detail around this Investor Day. So if you can hold until September 17, we’ll go into a lot of things that we’re doing on the R&D side as well as what we’re doing on application modernization and the tooling that we’re building there that will really speak to those investments that we’re making, and we’ll give you a lot more color.

Bradley Hartwell Sills

Got it. Great. And one more if I may, please. I know there’s been an effort to focus on driving higher quality workloads in that larger account base. I mean, to what extent would you attribute some of this upside to that effort? And maybe just an update on that effort as you may…

Dev C. Ittycheria

I would attribute a lot to that effort. I would say a big part of this growth is the fact that we’re acquiring higher-quality workloads that are growing faster and for longer than the workloads required, say, in earlier years. And I think that’s a big part of why you’re seeing this growth happen now.

Michael J. Berry

Carmen, I think we have time for one more question.

Operator

Alright, one moment please. And we have the line of Rishi Jaluria with RBC.

Rishi Nitya Jaluria

I’ll keep myself to one question. Dev, really nice to see the early traction with AI native companies. It’s always made sense to us, especially given your scalability and your ability to work with unstructured data. If we were to fast forward 5, 10 years and we start to see a real paradigm shift where instead of agents built on kind of the traditional GUI mobile interface that we’ve been in for the past 30 years, we actually entered kind of a multi-agentic world where maybe the interaction vector may move away from what we’ve been used to into more natural language. Can you talk about why MongoDB still has a strong role and some of the investments that you might be making to position yourself well for the world, understanding that’s at the very least several years away?

Dev C. Ittycheria

Yes, sure. So again, just to make sure we’re all talking the same language. We believe that agents essentially do 3 things. One, they perceive or understand the state of things. So you need essentially a way to understand the state of what’s happening in your business, then you need to decide what to do or plan. So basically, you have to come up with the plan saying, “I want to take this action or these sets of actions.” And then you have to act. You actually have to go execute those actions, right?

So why is MongoDB good for agents. One, as I said before, the JSON document database is the best of being able to model the real world, the messiness, the complicated nature. The real world does not fit easily in rows and columns. And that’s why our document

database, I think, is the best way to do that. Two, we obviously support search and vector search. So you can do very sophisticated hybrid search. So that becomes super important. And then with memory, if agents didn’t have memory, they would act like goldfish. They could only react to the last thing — last piece of information that they saw.

So memory lets agents connect the dots across time and situations. So you have different kinds of memory, things like short-term context, past experiences, knowledge, skills, et cetera, they need to be able to share quickly. You need to be able to orchestrate those agents because you may have multiple agents doing a certain task. You need to register and have governance policies around those agents. We think that the underlying platform needs to be able to support those things while there’s a lot more work needs to be done, the underlying architecture that we have in MongoDB is well suited to address those needs. And we think that we’ll be positioned to be a winner as people deploy more and more agents in their enterprise.

Operator

And with that, we conclude the Q&A session, and I will pass it back to Dev Ittycheria for his final comments.

Dev C. Ittycheria

Sure. Thank you again for joining us today. In summary, I think it’s clear that we delivered another strong quarter, highlighted by the accelerating Atlas growth, the continued adoption of — for AI applications and our expanding profitability. We are raising our revenue and operating margin guidance for the full year fiscal year 2026. And these results really reinforce that MongoDB is well positioned to capture the next wave of AI application development while driving durable and efficient growth. So with that, thank you, and we’ll talk to you soon. Take care.