This conference call transcript was computer generated and almost certianly contains errors. This transcript is provided for information purposes only.EarningsCall, LLC makes no representation about the accuracy of the aforementioned transcript, and you are cautioned not to place undue reliance on the information provided by the transcript.
Operator
Good day, and thank you for standing by. Welcome to the MongoDB's second quarter fiscal year 2025 conference call. At this time, all participants are in a listen-only mode. After the speaker's presentation, there will be a question and answer session. To ask a question during the session, you will need to press star 11 on your telephone. You will then hear an automated message advising your hand is raised. To withdraw your question, please press star 11 again. Please be advised that today's conference is being recorded. I would now like to turn the call over to your speaker for today. Brian, then you. Please go ahead.
Brian
Thank you, Lisa. Good afternoon, and thank you for joining us today to review MongoDB's second quarter fiscal 2025 financial results, which we announced in our press release issued after the close of market today. Joining me on the call today are Dave Itacheria, President and CEO of MongoDB, and Michael Gordon, MongoDB's COO and CFO. During this call, we will make forward-looking statements, including statements related to our market and future growth opportunities, our expectations for the macroeconomic environment in fiscal 2025, and the impact of AI, the benefits of our product platform, our competitive landscape, customer behaviors, our financial guidance, and our planned investments and growth opportunities in AI. These statements are subject to a variety of risks and uncertainties, including the results of operations and financial conditions that could cause actual results to differ materially from our expectations. For discussion of the material risks and uncertainties that could affect our actual results, please refer to the risks described in our quarterly report on Form 10-Q for the quarter ended April 30th, 2024, filed with the SEC on May 31st, 2024. Any quarterly statements made in this call reflect our views only as of today, and we undertake no obligation to update them except as required by law. Additionally, we will discuss non-GAAP financial measures in this conference call. please refer to the tables in our earnings release on the investor relations portion of our website for reconciliation of these measures to their most directly comparable GAAP financial measure. With that, I'd like to turn the call over to Dave.
Dave
Thanks, Brian, and thank you to everyone for joining us today. I'm pleased to report that we had a good quarter and executed well against our large market opportunity. Let's begin by reviewing our second quarter results before giving you a broader company update. We generated revenue of $478 million, a 13% year-over-year increase against a very difficult year-over-year compare and above the high end of our guidance. Atlas revenue grew 27% year-over-year, representing 71% of revenue. We generated non-GAAP operating income of $52.5 million for 11% non-GAAP operating margin, and we ended the quarter with over 50,700 customers. Overall, we were pleased with our performance in the second quarter. We had a strong new business quarter, and we saw improving sales productivity year over year. We saw strength across the board with both Atlas and Enterprise Advance exceeding our expectations, demonstrating the enduring appeal of our run anywhere strategy. Our Q2 performance reinforced our belief the slow start to the new business Q1 was purely operational, and we feel good about our new business outlook in the second half of the year. Moving on to Atlas consumption, the quarter played out modestly better than our expectations. Michael will discuss consumption trends in more detail. Finally, retention rates remain strong in Q2, demonstrating the quality of our product and the mission criticality of our platform. Our performance this quarter reinforced our confidence in our ability to execute on a long-term opportunity. As we said before, companies today rely on software to express their business strategy. This trend has driven our success for the past decade, and we anticipate it will continue to do so for the foreseeable future. Even with our success to date, We only have a low single digit share in one of the largest and fastest growing markets in all of software. When you combine this foundational tailwind with the opportunities for our customers to incorporate generative AI into their businesses and modernize their legacy application estate, it is clear that MongoDB has multiple long-term growth opportunities. Turning to AI, AI continues to be an additional long-term opportunity for our business. At the start of the fiscal year, we told you that we didn't expect AI to be a meaningful tailwind for our business in fiscal year 2025, which has proven accurate. Based on recent peer commentary, it seems that the industry now mostly agrees with this view. Companies are currently focusing their spending on the infrastructure layer of AI and are still largely experimenting with AI applications. Inference workloads will come and should benefit MongoDB greatly in the long run, but we are still very early when the monetization of AI apps will take time. AI demand is a question of when, not if, and our discussions with customers and partners give us increasing conviction that we are the ideal data layer for AI apps for a number of key reasons. First, more than any other type of modern workload, AI-driven workloads require the underlying database to be capable of processing queries against rich and complex data structures quickly and efficiently. Our flexible document model is uniquely positioned to help customers build sophisticated AI applications because it is designed to handle different data types, your source data, vector data, metadata, and generated data right alongside your live operational data, updating the need for multiple database systems and complex backend architectures. Second, MongoDB offers a high performance and scalable architecture. As the latency of LLMs improve, the value of using real-time operational data for AI apps will become even more important. Third, We are seamlessly integrated with leading app development frameworks and AI platforms, enabling developers to incorporate MongoDB into their existing workflows while having the flexibility to choose the LLM and other specific tools that best suit their needs. Fourth, we meet or exceed the security and compliance requirements expected from an enterprise database, including enterprise-grade encryption, authorization, and auditability. Lastly, customers can run MongoDB Anywhere on-premise or as a fully managed service in one of the 118 global cloud regions across three hyperscalers, giving them the flexibility to run workloads to best meet their application use cases and business needs. We see three main opportunities where we believe AI will accelerate our business over time. The first is that the cost of building applications in the world of AI will come down, as we've seen with every previous platform shift. creating more applications and more data, requiring more databases. The second opportunity is for us to be the database of choice for customers building greenfield AI applications. While we see that there's tremendous amount of interest in and planning for new AI-powered applications, the complexity and fast-moving nature of the AI ecosystem slows customers down. That's why we launched the MongoDB AI Applications Program, or MAP, which became generally available to customers last month. MAP brings together a unique ecosystem, including the three major cloud providers, AWS, Azure, and GCP, as well as Accenture and AI pioneers like Anthropic and Cohere. MAP offers customers reference architectures, an end-to-end technology stack that includes pre-built integrations, professional services, and a unified support system to help customers quickly build and deploy AI applications. The third opportunity is to help customers modernize their legacy application estate. As you know, this segment of the market is a massive opportunity for us, as most of the existing 80 billion plus database industry is built on dated relational architecture. Modernizing legacy applications has always been part of our business, and we have taken steps over the years to simplify and demystify this complex process through partnerships, education, and most recently, our relational migrator product. AI offers a potential step function improvement, lowering the cost and reducing their time and risk to modernize legacy applications. For that reason, earlier this year, we launched several pilots with our customers where we worked with them to modernize mission critical applications, leveraging both AI tooling and services. The early results from these pilots are very exciting as our customers are experiencing significant reductions in time and cost of modernization. In particular, we have seen dramatic improvements in time and cost to rewrite application code and generate test suites. We see increasing interest from customers that want to modernize their legacy application state, including large enterprise customers. As a CIO of one of the world's largest insurance companies said about our pilot, this is the first tangible return he's seen on his AI investments. While it's still early days and generating meaningful revenue from this program will take time, we're excited about the results of our pilots and the growing pipeline of customers eager to modernize the legacy estate. Finally, I understand that there are a lot of questions about the current business conditions and the macro environment more broadly. So let me give you a sense of what we're seeing across the business. As a reminder, when thinking of the macro influence on our business, it's important to distinguish between consumption of existing workloads and new business. Starting with consumption of existing applications on our platform, this is where we have historically seen a macro impact as usage of applications is impacted by the underlying business conditions of our customers. As we discussed on our last earnings call, in Q1 we did see broad-based consumption growth slowdown, suggesting some macro softening. Our usage trends suggest a similar macro environment in Q2 as in Q1, even though Q2 Atlas consumption growth was modestly ahead of our expectations. Moving on to new business, we generally have not seen the macro environment impact our ability to win new business, and that was true in Q2 as well. We realize that this is different from what you hear from some other software vendors. Ultimately, software application development continues even in uncertain environments as customers know they need to continue investing in internally developed software to run the business as well as to drive competitive differentiation. In addition, we still have relatively low market share in a large market, which means we have an opportunity to gain share in any environment. Now, I'd like to spend a few minutes reviewing the adoption trends of MongoDB across our customer base. Customers across industries and around the world are running mission-critical projects on MongoDB Atlas, leveraging the full power of our developer data platform, including Fanatics, Occidental Petroleum, and Indeed. Fanatics Betting and Gambling, a division of the sports ecosystem company Fanatics, leverages MongoDB to significantly enhance the user experience of their mobile app. Initially, the team launched a platform on Postgres, but faced challenges with scalability, flexibility, and excessive complexity. After migrating to MongoDB Atlas, the team also integrates Atlas Search to provide users with a better experience to find all available betting options. With Atlas handling scaling, partitioning, and operations, developers can focus on writing code and improving the user experience. Looking ahead, Fanatics plans to continue to expand on MongoDB Atlas as they ensure they can operate at scale as they prepare for the start of the NFL season. L'Oreal, McKesson, and Nationwide Building Society are turning to MongoDB to modernize applications. L'Oreal's Tech Accelerator, a department dedicated to catalyzing digital innovation at L'Oreal, is utilizing MongoDB for an application designed to bring products and solutions to market while quickly improving employee efficiency. The team's previous database solution had limited out-of-the-box functionality and was unable to handle the complex calculations needed to retrieve and restructure large amounts of data from their data warehouse. L'Oreal migrated to MongoDB Atlas to streamline the application architecture and simplify a previously highly complex and time-consuming data access layer. With this migration, L'Oreal achieved a 40-fold performance improvement. On Atlas, the existing code is easier to maintain, more scalable, and more efficient, making life easier for developers. Mature companies and startups alike are using MongoDB to help deliver the next wave of AI-powered applications to their customers, including Delivery Hero, Generali, and Questflow. Delivery Hero, a longtime MongoDB Atlas customer, is the world's leading local delivery platform, operating in 70-plus countries across four continents. Their quick commerce service enables customers to select fresh produce for delivery from local grocery stores. Approximately 10% of the inventory is fast-moving, perishable produce that can go quickly out of stock. The company risks losing revenue and increasing customer churn if the customer didn't have viable alternatives to their first choice. To address these risks, they are now using state-of-the-art AI models and MongoDB Atlas Vector Search to give hyper-personalized alternatives to customers in real-time if items they want to order are out of stock. With the introduction of MongoDB Atlas Vector Search, the data science team recognized that they could build a highly performant real-time solution more quickly and for less cost than alternative technologies. In summary, we had a healthy Q2 with both Atlas and EA exceeding expectations. We saw a strong new business quarter and improved sales productivity, and we're confident in our ability to keep winning new business in the second half of the year. Looking forward, we see great opportunity to help our customers modernize legacy applications and build the next generation of AI-powered applications. With that, here's Michael.
Michael
Thanks, Dave. I'll begin with a detailed review of our second quarter results and then finish with our outlook for the third quarter and full fiscal year 2025. First, I'll start with our second quarter results. Total revenue in the quarter was $478.1 million, up 13% year-over-year and above the high end of our guidance. As a reminder, while we have difficult compares throughout fiscal 25, we're facing a particularly difficult year-over-year comparison in Q2, given we had a number of large multi-year partnership licensing deals in Q2 last year. Under 606 accounting rules, we recognize the license component upfront which creates a difficult compare a year later. Shifting to our product mix, let's start with Atlas. Atlas grew 27 percent in the quarter compared to the previous year and now represents 71 percent of the total revenue compared to 63 percent in the second quarter of fiscal 2024 and 70 percent last quarter. We recognize Atlas revenue primarily based on customer consumption of our platform, and that consumption is closely related to end-user activity of their applications. Let me provide some context on Atlas consumption in the quarter. First, as a reminder, in Q1, Atlas consumption growth was below our expectations, and we updated our growth assumptions for the remainder of the year. In Q2, Atlas consumption growth was modestly ahead of those updated expectations across the board. While this is encouraging, consumption growth is still below our original forecast from the beginning of the year. Turning to non-Atlas revenue, non-Atlas came in modestly ahead of our expectations in the quarter, as we continue to have success selling incremental workloads into our existing customer base. On a year-over-year basis, non-Atlas revenue is down 13% due to the especially difficult compare I referenced earlier. Turning to customer growth, during the second quarter, we grew our customer base by approximately 1,500 customers sequentially, bringing our total customer count to over 50,700, which is up from over 45,000 in the year-ago period. Of our total customer count, over 7,300 are direct sales customers, which compares to over 6,800 in the year-ago period. The growth in our total customer count is being driven primarily by Atlas, which had over 49,200 customers at the end of the quarter compared to over 43,500 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 adding incremental Atlas workloads. Continuing on, in Q2, Our net AR expansion rate was approximately 119%. The decline versus historical periods is attributable to a smaller contribution from expanding customers. We ended the quarter with 2,189 customers with at least $100,000 in ARR and annualized MRR, up from 1,855 in the year-ago period. Moving down the income statement, I'll be discussing our results on a non-GAAP basis unless otherwise noted. Gross profit in the second quarter was $360.8 million, representing a gross margin of 75%, which is down from 78% in the year-ago period. Our year-over-year margin decline is primarily driven by a lower mix of high-margin upfront license revenue compared to last Q2, as well as Atlas growing as a percentage of the overall business. Our income from operations was $52.5 million, or an 11% operating margin for the second quarter, compared to a 19% margin in the year-ago period, The primary reason for our more favorable operating income results versus guidance is our revenue outperformance. In addition, Q2 operating income benefited from the timing of certain marketing and other spend, which we now expect to incur in the second half of the year. Net income in the second quarter was $59 million, or 70 cents per share, based on 83.8 million diluted weighted average shares outstanding. This compares to a net income of $76.7 million, or 93 cents per share, on 82.5 million diluted weighted average 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. Operating cash flow in the second quarter was negative $1.4 million. After taking into consideration approximately $2.6 million in capital expenditures and principal repayments of finance lease liabilities, free cash flow was negative $4 million in the quarter. This compares to free cash flow of negative $27.3 million in the year-ago period. In addition this quarter, we also received $170.6 million in cash from the settlement of the capped calls associated with our 2024 convertible notes. I'd now like to turn to our outlook for the third quarter and full fiscal year 2025. For the third quarter, we expect revenue to be in the range of $493 million to $497 million. We expect non-GAAP income from operations to be in the range of $57 million to $60 million, and non-GAAP net income per share to be in the range of 65 cents to 68 cents, based on 84.6 million estimated diluted weighted average shares outstanding. For the full fiscal year 2025, we expect revenue to be in the range of $1.92 billion to $1.93 billion, non-GAAP income from operations being the range of $187 million to $195 million, and non-GAAP net income per share being the range of $2.33 to $2.47, based on 84.3 million estimated diluted weighted average shares outstanding. Note that the non-GAAP net income per share guidance for the third quarter and full fiscal year 2025 includes a non-GAAP tax provision of approximately 20%. And I'll provide some context around our updated guidance. First, Because of the stronger than expected Atlas consumption in Q2, our starting Atlas ARR for the back half of the year is higher than we anticipated in May. As a result, we're raising our Atlas revenue forecast to reflect that higher starting point. We are not changing our underlying Atlas Q3 and Q4 consumption growth rate assumptions. Second, we are slightly increasing our EA assumptions for the rest of the year to reflect the strength of the second half EA pipeline. Finally, thanks to the performance in Q2 and the increased revenue outlook, we now expect a 10% operating margin at the midpoint of our fiscal 25 guidance. We will continue investing to capture our long-term opportunity with a focus on our strategic priorities. Separately, although we don't provide guidance on cash flow, we wanted to call out two items that will impact fiscal 25 cash flow but benefit our long-term financial profile. In Q2, we started paying some of our cloud provider commitments up front in exchange for better economics. We see this as a very low-risk, high ROI use of our cash and one that will benefit our gross margin going forward. These prepayments will represent a negative impact to our operating cash flow in the back half of the year of roughly $20 million per quarter. In addition, in Q3, we will incur approximately $20 to $25 million of CapEx to acquire IPv4 addresses, which will allow us to further reduce our cloud infrastructure costs in the future. To summarize, we're pleased with our second quarter results and especially our ability to win new business. We have a small share, one of the largest and fastest growing markets in all of software. The secular tailwinds at our back are getting stronger in the age of AI, and we're excited about the future. We'll continue investing judiciously and focusing on our execution to capture this long-term opportunity. With that, we'd like to open it up to questions.
Operator
Operator? Thank you. As a reminder, if you would like to ask a question, please press star 11 on your telephone. We also ask that you limit yourself to one question and one follow-up. As well, please wait for your name and company to be announced before you proceed with your question. One moment while we take the first question. And the first question today is going to come from Sanjit Singh of Morgan Stanley. Your line is open.
spk08
Yes, hi. Thank you for taking the question. Congrats on Q2. Nice to see Mongo beating and raising again. David, I wanted to follow up on some of the changes, the operational changes you're making to the business that you announced last quarter, just in terms of the update, in terms of realigning some of the sales incentives to drive higher workload quality while maintaining some of that record workload growth that you saw last year. Could you sort of walk us through what the changes were you put in place and how much confidence you have that you will drive on better workload growth as we get into early next year?
Dave
Yeah, sure. So, first of all, thanks, Sanjit. Just a reminder for everyone, we made some slight incentive comp changes to really get our reps to focus, to have a little bit more balance on size versus volume of workloads acquired. These changes were well received by the field. We had a good workload quarter, as you just described. Workload starts small, so it's too early. While we're pleased with our results, obviously, it's too early to declare a victory and to really see if this is going to have a material impact and changes, but clearly we're happy with what we've seen so far.
spk08
Right. And then as a follow-up for Michael, I remember at the Q4 last year, you gave us the update on sort of the unused credit dynamics. As we start to think about the back half of this year, is there anything else that we should think about that? I mean, you just mentioned some of the cash flow impacts just now. But is there anything on sort of like the revenue side of the equation that we should keep in mind? I kind of noticed at the end of this year, there's this more holiday days in the quarter in Q4. Does that have an impact? Anything that we should be, that you would like to call out in terms of our modeling when we think about expectations for Atlas more broadly going into the end of the year?
Michael
Yeah, I think there are a few things to take into account that have been throughout the year, but of course are reflected in the updated guide. Specifically on Atlas throughout the year, we have tough compares given the headwinds from the lack of the unused commitments. That's obviously the hardest. As the year goes on, that sort of number builds, so that makes for a tough year-over-year comparison. I think the other thing to think about in the context of Atlas is, you know, the ending year dynamic is really a compounding over the course of the year. As we talked about at the beginning of the year, we got off to a slower start and some of the prior year workloads, those cohorts were growing at slower rates. And so that does affect and kind of compound over the course of the year. And so I think that's important to keep in mind and is factored into our guidance. And then looking at the revenue picture more broadly, you know, moving away from Atlas and thinking about, you know, EA, we've had throughout the year a headwind on EA and we'll have that as a multi-year headwind in the back half as well. The only other things I think I would call out is, you know, we saw a slower seasonal rebound in Q1 and so we're expecting that slower seasonal rebound to, you know, occur in Q3 as well. And then, as you mentioned, Q4 is a seasonally weaker quarter. And so all those are some of the considerations to think about when you're thinking about the rest of the year.
Atlas
Appreciate that, Michael.
Operator
Thank you. One moment for the next question. And our next question today will be coming from the line of Remo Linschild of Barclays. Your line is open.
metadata
Hey, thank you. Two quick questions from me. Michael, on the EA, you kind of talked about like a slightly better pipeline in the second half. Can you talk a little bit, is that kind of still the continuation of what we saw in Q2, Q3 last year, where people wanted to modernize their kind of self-service footprint ahead of a move to the cloud, or what's the dynamic there? And then one for Dave, like, There's a big debate about reference architecture on AI, and I think it's a little bit too early, but it comes up with investors a lot. What are you seeing there? I know AI infrastructure, we're still in the early innings, but what are you seeing there in terms of people engaging with you? Thank you, and good to see you back on track. Thank you.
Michael
Thanks, Raimo. On the EA question, I would say there are lots of reasons why we're seeing strengths of EA today. I wouldn't uniquely or exclusively tie it to, you know, AI. I think it's broadly, you know, in support and shows the strength of the run anywhere strategy. The other thing that I'd call out is, you know, given the 606 dynamics as it relates to EA, we're always sensitive to sort of the multi-year aspect, and that's sort of what's, you know, provided, you know, given the strength of multi-year in fiscal 24, some of the tough compares here in fiscal 25. I think what we've seen in terms of the pipeline is, is strength in EA broadly. So not just a multi-year phenomenon, but really sort of a volume phenomenon as well that we're seeing when we kind of look out on the horizon, and that is reflected in the updated guide.
Dave
Thanks, and Raimo, on the question about AI, I think in terms of reference architectures, I think it's important to understand, and I said this in the prepared remarks, is that Unlike most of the workloads, AI-driven workloads really require the underlying database to be capable of processing queries against very rich and complex data structures, both quickly and efficiently. And MongoDB is well positioned to do that. We can basically unify and handle source data, vector data, metadata, generated data from your LLM right alongside your live operational data. And then as the performance of these LLMs and latency of these LLMs increase, accessing real-time data becomes really important. Like, say you're calling and talking to a customer support chatbot, you want that chatbot to have up-to-date information about that customer so that they can provide the most relevant and accurate information possible. You know, there are some questions about LLMs, whether a general-purpose LLM or a fine-tuned LLM, what the tradeoffs are. Our belief is that given the performance of LLMs, you're going to see The general purpose LLMs probably win and will use RAG as the predominant approach to marry generally available data with proprietary data. And then you are starting to see things like advanced RAG use cases where you get much more sophisticated ways to ask complex questions, provide more accurate and detailed answers, and better adapt to different types of information and queries. That's what we're seeing. I think it's a quickly evolving space, but we feel very good about our positioning for AI, even though it's still very early days.
metadata
Okay, perfect.
Operator
Thank you. Thank you. One moment for the next question. And our next question will be coming from Cash Reagan of Goldman Sachs. Your line is open.
Cash Reagan
That's perfect. I was about to call Raimo and say, can you please explain reference architectures? I'll buy you a nice glass of wine. So hopefully he'll take me up on that offer because that was a very technical question. Thank you. But anyway, coming back to the call here. Dave, a question for you. Why is EA doing so well? So you talked about a pipeline in the second half of the year. Last year you had some significant wins for EA. We were supposed to be on this cloud journey. So Definitely Atlas has reached parity in many senses. It can support big scale applications, whatnot. So curious to get your thought on why is EA still an important piece of the business? And I suppose when you look at AI, could something surprise us? I know that you've had this view that we're building infrastructure first and then the platform, then applications. Could it be by any chance a different way to approach AI? in this cycle and that we don't really need applications, but somehow these LLMs are going to be a perfect replacement for the way we think about old world applications. I'm just curious to see what the devil's advocate view might be if someone were skeptical of the whole AI applications built out on top of the infrastructure and platform. Thank you so much.
Dave
Yeah, so thanks, Kash. What I would just say is, you know, we did have a better than expected both Atlas and EA quarter. There's no question about that. I don't want to suddenly say there's some inflection point on EA and our business. I think I would really attribute it to, one, better execution, and two, customers do really appreciate our run anywhere strategy. There's still lots of customers, either for regulatory reasons or other reasons, who want to run workloads on-premise, and they're not going away. I mean, we've been in this cloud journey for 10 years, and if Some workloads are just very hard to move to the cloud, or some workloads for many customers don't make sense to move to the cloud, at least not anytime soon. So the fact that they can build it on MongoDB and have the optionality to move it to the cloud later very easily is something that's very compelling for customers. With regards to AI, I mean, we predominantly see most of the AI workloads in the cloud, but there are definitely lots of customers who are looking at using open source LLMs in particular things like LAMA, and running those workloads locally. Obviously, it means that they also have to have access to NVIDIA hardware like GPUs, but we do see some customers do that. Again, I wouldn't suggest that that's also an inflection point or cause for the EA outperformance. I think it's very, very early days, and most of those are experiments that customers are running. But clearly, we feel really good about running anybody's strategy, and as we've said in the past, We are investing in basically introducing search and vector search to our community product, and that will then show up in EA. So EA is definitely an area where we're also investing from a product point of view.
Atlas
Thank you. One moment for the next question.
Operator
And our next question. will be coming from the line of Tyler Radke of Citi. Your line is open.
Tyler Radke of Citi
Hi, this is Tyler Radke. I think the question was for me.
Tyler Radke
I couldn't hear the name, but thanks for taking the question. Dave, you talked about a post-gress displacement in the quarter, and I think that's the first time you've talked about that, at least in recent quarters. So I was wondering if you could just sort of frame for us, what use cases do you come across them? I know it is popular in terms of a SQL displacement for migrating legacy applications, but if you could just frame for us sort of the competitive environment in Postgres, both the open source stuff and some of the new, you know, venture-backed startups in the space. Thank you.
Dave
Yeah, sure. So thanks for the question. Yeah, it's important to understand that Postgres has been around for almost 40 years. I mean, Postgres, the name is termed from post-ingress. So that technology has been around a long time. As you said, they're really the beneficiary of lift and shift from Oracle, SQL Server, and MySQL. So they're kind of consolidating the relational market. In terms of why do we compete or why do we win, I would say it's a few things. One, our schema flexibility is MongoDB has a very flexible scheme allowing you to store documents in a JSON-like format, so this is beneficial for application structures that evolve over time. We can horizontally scale, so we're making it very easy to distribute data across multiple servers or virtual servers, so it's for applications that require massive amounts of data. Performance of large data sets, again, we can handle that better than Postgres. The built-in sharding allows for automatic data distribution. And then we also, I think, developer productivity, the JSON-like format and flexible schema can lead to faster development cycles, especially for customers who really work in agile environments. So we feel like we compete. Our win rates against Postgres are high. But again, there's lots of decisions being made where we're not party to, where people are just doing a lift and shift off of. you know, a legacy platform, or they just want to stay on relational because that's what they know. And obviously it's our job to educate them on the benefits of MongoDB, but we feel good about our competitive position against Postgres.
Tyler Radke of Citi
Thank you. And follow-up question for Michael.
Tyler Radke
You talked about how consumption in Atlas in the quarter tracked a bit better than planned. Sounds like it was not a macro-related improvement. What do you think the driver that was and are you seeing any improvement in some of the recently acquired? workloads those starting to ramp up better or maybe they have different seasonal patterns than any thought any any color on sort of that driver of Outperformance would be helpful.
Michael
Thank you Yeah, sure, I would say that yes q2 consumption growth was better than our expectations that was great to see and I would describe it as within a reasonable or typical range of outcomes, and so there are no signs that we've seen that would specifically point to any material changes in the underlying macroeconomic environment, better or worse. We've certainly seen, to your question on the workloads and some of the cohorting, we did see a little better performance there as well. and in line with what we saw elsewhere, but they're still below our original expectations. So I think that's probably the key couple things there.
Tyler Radke of Citi
Thank you.
Operator
Thank you. One moment for the next question. And our next question for the day will be coming from Brad Reback of Stifel. Your line is open.
Brad Reback of Stifel
Great. Thanks very much. Dave, on your commentary around the modernization pilots, how should we think about timing as that being a bit of a tailwind to the overall growth rate?
Dave
Yeah, so again, just to make sure everyone understands, the legacy relational application market or database market is quite large. It's over $80 billion, and it's a massive opportunity for us. And since day one, since our IPO, we've been getting customers to migrate off relational to MongoDB. But one of the biggest friction points has been that while it's easy to move the data, you can map the schema from a relational schema to a to a document schema and you can automate that, the biggest stumbling block is that the customer has to, or some third party, has to rewrite the application, which by definition creates more costs, more time, and in some cases more risk, especially for older apps where the development teams who built those apps no longer exist. So what's been compelling about AI is that AI has finally created a shortcut to overcome that big hurdle And so essentially you can start basically diagnosing the code, understand the code, recreate a modern version of that code, and generate test suites to make sure the new code performs like the old code. So that definitely gets people's interest because now all of a sudden what may take years or multi-years you can do in a lot less time. And the pilots that we have done, the time and cost savings have been very, very compelling. That being said, we're in the very early days. There's a lot of interest. We have a growing pipeline of customers across, frankly, all parts of the world, from North America to EMEA and even the PACRIM. And so we're quite excited about the opportunity. But again, I would say it's very early days. But there's a number of reasons why I would say that customers are... very focused on this, what the cost of licensing and maintaining legacy apps is becoming too high to bear. In many cases, the regulatory and compliance requirements are forcing customers to upgrade. There's a whole end of life of critical technologies, notably Sybase, that's forcing customers to act. There's a ton of technical depth on these legacy platforms that limits the organization's flexibility to do things with AI. And candidly, customers have also soured on the traditional approach of using large systems integration projects that are very costly and take a long time. So this whole approach is definitely getting their attention.
Brad Reback of Stifel
That's great. And then one fast follow-up. It feels like we've spoken about EA more than typical on this call, which is great. As we think about the back half pipeline, Is the composition similar in so much as it's predominantly existing customers, or are you beginning to see an uptick in net new customers there as well? Thanks.
Dave
Yeah, I would say that it's predominantly existing customers that are doing this, you know, who are maybe doubling down on EA and expanding the footprint to EA, and that's typically the driver for our EA business.
Brad Reback of Stifel
Thanks very much.
Dave
Thank you.
Operator
Thank you. And one moment for the next question, please. And our next question will be coming from Carl Kirsvedt of UBS. Your line is open.
Carl Kirsvedt
Okay, great. Hey, Dave, at the beginning of the call, you mentioned that Mongo's slow start to the year was, quote, purely operational. But on the Q1 call, when you were describing what happened, you described it as very much a broad-based macro issue. So I'm curious, should I interpret that comment as you, Mike, and the team have done a bit of a rethink and might be changing modestly at least your explanation about the Q1 results?
Dave
Yeah, actually, just to be clear, I think just to make sure everyone understands, what we called out was that the consumption of existing workloads was broad-based. The slowdown was broad-based across both geos and channels. it was a new business, it just got off to a slower start. So that was what we called out in our Q1 call. And with our Q2 call and results, as you can see, new business performed better than expected on both Atlas and EA, which is why we believe the Q1 issue was an operational issue, not a Q1 new business, sorry, was an operational issue, not a macro issue. The macro issue was all about the slowdown in consumption because Q1 typically tends to be a seasonally strong quarter for us and that didn't happen.
Michael
Yeah, I think that's really important, Carl, just to keep in mind when we talk about macro, it could have those two different effects, right? The new business piece and then the expansion of existing workloads. And so What we were talking about, as Dave said, in Q1 was the existing workload expansion. The operational piece was on the new business side. And I think the key thing that we've observed really in a wide range of macroeconomic conditions in our almost seven years as a public company is with the exception of that one quarter, we've been able to execute quite well in the new business opportunity really in most environments. given that we have a huge market, relatively low share, strong product, and a talented team that executes well. And so I think that's probably helpful just to understand as we're sort of using the terms just to make sure we don't trip ourselves up.
Carl Kirsvedt
Yep. Okay. That's great. And then my follow-up for you, Mike, is when you were pressed in prior questions about the second half, most of the variables that you brought up were frankly headwinds, things like tough compares. But If I look objectively at your 3Q total revenue guide of 497, and this is my assumption, but I assume kind of a normal beat, that's going to result in sequential total revenue growth that's actually the highest, I think, that MongoDB has ever put up. Clearly, there's something good that you're embedding in that 3Q guide. What is that specifically?
Michael
Yeah, so I'm not sure I'd follow all the math on the fly, and I'm happy to follow up, obviously, with any of that. But if I just think about the Q3 guide, I won't belabor the headwinds, but they exist, and we've sort of been through those, as you called out. But maybe to talk about the couple of positive things that we see that lead to sort of our improved outlook or increased revenue forecast for the balance of the year. I'd say it's really two things. In Atlas, it's reflecting the stronger than expected Q2, and therefore the starting ARR at the beginning of Q3 is higher. Because we haven't seen any change in the underlying macro, the actual growth assumptions for Q3 and Q4 we have remain unchanged, and so you're just applying growth to a higher base, but as the base keeps getting bigger and bigger, that does matter, and that flows through. And then secondly was the positive impact of the increased strength in the EA pipeline for the second half of the year and the impact on that revenue. So those would probably be the two things that I'd call out on the positive side of the revenue in addition to, of course, having everyone keep in mind the tough compare and some of the headwinds we talked about.
Q4
Okay. That's helpful, Mike. Thanks a lot.
Operator
Thank you. Thank you. And one moment for the next question. Our next question will be coming from Jason Adler of William Blair. Your line is open.
Jason Adler
Yeah, thank you. I'm trying to understand the Q4 implied revenue guide, Michael. You know, just the implied sequential growth there is only 1.5% if I look at your full year relative to what you said specifically on Q3. That would be well below the historical seasonal pattern in Q4. So I just want to understand, is there something going on in Q4 that's different, just sequentials? I understand the year-over-year is tough, but the sequentials seem well below where you've normally been. Is there something specific that we should think about there?
Michael
Yeah, I think there are two things. So obviously, it sounds like you get the year-over-year, but just to make sure everyone's sort of on the same page is making sure that people understand you know, the tough compare for Atlas specifically around the unused commitments and then EA, the multi-year, and then more broadly, you know, we had one, we had less growth in Q1. That compounds, and then also we talked about how we have a, we're assuming just in the same way that we saw in Q1, we're assuming less of a seasonal rebound in Q3 That has implications for Q4.
Jason Adler
Gotcha. Okay. And then just one quick follow-up on gross margins. Did you talk about Atlas gross margins in the quarter? And then how material will the impact be from the prepayments and the IPv4 purchases on Atlas gross margins over time?
Michael
Yeah, so we didn't specifically break out Atlas gross margin in the quarter, but they continue to be lower, but obviously we've pretty significantly shrunk the delta. That does explain some of the comparison on the year-over-year basis. We didn't give specific guidance going forward. If you think about the two changes that we called out on the cash flow side, those will benefit us in terms of gross margin. There won't be a significant benefit in fiscal 25, so we'll obviously talk about that more when we get to the fiscal 26 guide, but it's a very excellent use of our cash and good ROI in terms of improving those economics further. Thank you.
Operator
Thank you. One moment for the next question. And our next question will be coming from Brad Sills of Bank of America. Your line is open.
Brad Sills
Oh, great. Thank you so much. Great to hear the continued strength in new workloads here. That's been a very consistent theme here throughout all of this. I did want to ask about some of the newer services like vector and stream processing. How are those contributing to the strength you're seeing or sustained strength you're seeing in new workloads?
Dave
Yeah, Brad, thanks for the question. So in terms of search, we're seeing solid momentum in search. We're having success with that business. It's starting to grow. And we just introduced a new capability called Search Nodes, which allows customers to optimize their search deployments by asymmetrically scaling specific nodes dedicated for search versus the rest of the nodes on their cluster. It also helps dealing with use cases that are very search intensive. One of the largest gaming companies in the world re-platformed the content moderation platform from DocDB, Elastic, and Dynamo to Atlas and Atlas Search, and using Atlas Search nodes for workload isolation and high performance. On Vector, we're continuously seeing growth and adoption, and we see Vector is effective in attracting new customers to the MongoDB platform. A world-renowned financial news organization, which was already running on Atlas, migrated from Elastic Search to Atlas Search using search nodes to take advantage of our vector search capabilities to build a site search that combines lexical search with semantic search to find the most relevant articles for user query. And a European energy company built a geospatial search application using Atlas and search and vector search. And the Atlas was built on-prem and to clouds to vectorize geospatial data and facilitate research and discovery. And then we recently, you know, announced streaming, or stream processing, I should say, you know, GA and .local New York in May, and we've seen strong interest. It's still early days, but we're seeing interest from a variety of industries, ranging from automotive to retail to transportation, who all want to work with streaming data and want to be able to take actions on that data to, you know, drive their business. And so... So customers are very pleased with the performance of the product and how easy it is to use. But again, it's just early days since we just launched it, only in May.
Brad Sills
Wonderful. Great to hear it. One more, if I may, please. On the last earnings call, you mentioned how the slowdown in consumption was broad-based across industries and workload types, which led you to believe that it was a macro-related impact. With some of the improvement you've seen this quarter, I know it's early, but was that also broad-based, and could you deduce that maybe that might mean some improvement in the underlying macro, or was it more outsized to certain types of industries and services?
Michael
Thank you. I would describe it as broad-based, but I would describe it within sort of a reasonable range of outcomes, and so no clear indication that macro is improving or deteriorating, and just a good quarter.
Brad Sills
Wonderful. Thank you.
Operator
Thank you. One moment for the next question. And our next question will be coming from Rashi Jaleria of RBC Capital Markets. Your line is open.
Rashi Jaleria
Wonderful. Hey, this is Rashi Jaleria from RBC. Two questions. Maybe, Dave, I wanted to first start out going back to some of what you talked about, which is the MongoDB versus Postgres debate. One of the kind of popular theories out there or debates out there, is which architecture is better suited to new generative AI applications, especially if we truly do believe AI will lead to a replatforming of software similar to what we saw with the cloud. I know you talked a little bit about reference architecture earlier in Q&A, but maybe could you walk us through why you believe MongoDB is better suited to new generative AI native applications versus Postgres? And then I've got a quick follow-up.
Dave
Yeah, sure. So very quickly, the reason we believe that we're well positioned to win these new workloads is that AI-driven workloads require the underlying database to be capable of processing queries against rich and complex data structures. As you know, with AI, the data structures can be very, very complex. So that means that the data can be large and obviously not in any consistent size. MongoDB is designed to handle these different data structures, and I talked about we can help unify metadata, operational data, vector data, and generate it all in one platform. Relational databases, and Postgres is one of them, have limitations in terms of how they can handle different types of data. When the data gets too large, these relational databases have to do what's called off-row storage, and it creates a performance overhead on these relational platforms. Postgres has this thing called TOAST, which stands for the Oversized Storage Attribute Storage Technique, and it's basically a way to handle these different data types but it creates a massive performance overhead. So we believe that we are architecturally far better for these more complex AI workloads than relational databases. And we believe that ultimately we are the ideal data layer for AI applications. Obviously it's early days. As I said, we haven't seen a lot of inference workloads in production, but architecturally we feel very good about our ability to compete against relational databases and Postgres in particular. That being said, Postgres is popular because they're the beneficiary of people who want to stay on relational and who've kind of lived in relational for the last 30, 40 years. But head-to-head, we feel like we can win our share of business.
Rashi Jaleria
Got it. Thank you. That's really helpful. And then maybe just sticking on the theme of AI, we've talked before in the past that AI is just driving a lot of new code, making developers significantly more productive. Have you seen that behavior in any of your existing customers on Atlas where maybe their utilization rate goes up or the number of applications built per customer goes up? Anything like that, that understanding it super early might at least give you some early indication that the increased developer productivity could turn into a real tailwind in terms of consumption of the underlying Atlas for you. Thank you.
Dave
Yeah, so this is a common question I ask our customers when I meet with them in terms of what code generation tools are they using and what benefits they're gaining. The answers tend to be a little bit all over the map. Some people see 10%, 15% productivity improvement. Some people say 20%, 25% productivity improvement. Some people say it helps my senior developers be more productive. Some people say it helps my junior developers become more like senior developers. So the answers tend to be all over the map. There's no question in my mind that with this platform shift, like previous platform shifts, the cost of building apps will come down. And so by definition, more apps will be produced, which will generate more data, which requires more databases. Now, obviously, we're in the early days. And much like I've lived through other platform shifts, like going to the internet, a lot of people built out these. There's all this dark fiber that was built out to service all this potential demand. It didn't happen. People thought, It was maybe overinflated, and then before we knew it, the Internet transformed the way we work, the way we live, the way we socialize. And so I think that the same thing is happening here. I think it's all on the come, but I think we're well positioned for that opportunity. Wonderful. Thank you.
Operator
Thank you. One moment for the next question. And our next question will be coming from Brent Braceland of Piper Sandler.
spk14
Your line is open.
Atlas
Thank you. This is Brent Braceland, Piper Sandler. Dave, I wanted to go back to Atlas. And the question here is, given all the moving parts here around accounting, tough compares, unused commitments, Atlas growth looked really strong. I think 30% plus on a normalized basis after adjusting for unused commitments. That's before any benefit from AI. Can you just go back to the growth algorithm for Atlas here? I ask because 30% normalized growth would be 3x faster than industry growth, implying a meaningful share capture of new software builds. maybe be worthwhile if you just go walk through what is driving the momentum on the Atlas side that is, again, two, three X faster than the industry?
Dave
Well, what I would say is obviously we're pleased with our results. What I would say is one, Atlas is the most widely available cloud service by definition because it runs across the largest hyperscalers in 118 regions around the world. So no matter who you are, you can run on Atlas almost anywhere in the world. Second, because MongoDB is a general purpose database, you can use us for almost any conceivable use case, so we're not limited to some narrow set of use cases or narrow set of users who have a very particular set of needs. That helps. Three, we have a large community of developers. We have millions of developers all around the world from Asia to North America to EMEA who are using MongoDB to address some problem. And so that also obviously, because of the popularity of MongoDB, we are the world's most popular modern database. I think that also speaks to the strength of the product. And then I would say we try and execute well. Obviously, we got off to a slower start than we liked in Q1, but we are very, very focused on execution. We have a really good team, both in terms of our go-to-market team as well as our product and the supporting ecosystem teams, our partner team, engineering team. And so we all try and put our heads down, and we were obviously disappointed by Q1, and we're really focused on growing the business as fast as we can.
Atlas
Perfect. And then, Michael, just a quick follow-up on op margins. Guidance implies 10% plus op margins in the second half, particularly as the revenue run right here gets to $2 billion exiting the year. Can you continue to invest in sales capacity while balancing double-digit margins at this scale?
Michael
Yeah, so specifically around the second half, we've obviously increased the full year guide, and that flows through. We did talk about some of the timing issues between halves. We are incrementally, on a marginal basis, investing in the back half of the year in our AI initiatives. But I think if you look at the two-year view, We're still making 500 basis points of margin progress over the last two years. So, yeah, I think it's a balancing act. Brent, I think the key thing is we have such a large opportunity and such attractive investment areas that we'll try and continue to do both.
Operator
Helpful. Thank you. Thank you. One moment for the next question. And our next question will be coming from Eric Heath. of KeyBank Capital Markets. Your line is open.
Eric Heath
Hey, good afternoon. Thanks for taking the question. Just a couple of quick ones for me. Mike was curious to get a better understanding on just maybe some of your assumptions on why you're expecting a slower than typical rebound coming out of 3Q. And then, Dave, thinking about the third quarter here and the Fed fiscal year end, just what are some of your expectations for that article, which I assume is usually a big EA opportunity? And thanks.
Michael
Yeah. So quickly just on, on Q3, I think we're informed by our recent experience, including the fact that the seasonal rebound in Q1 of this year was slower than what we've typically seen. And so, uh, that's, uh, what we're taking into account when we think about Q3 and that's really what's driving it.
Dave
Yeah. And on the, I think you said, if I heard you correctly, you're talking about the fed, the public sector and the fed in particular. Yes. I mean, obviously their fiscal year ends in September. Um, we, uh, We have a good team there. We have FedRAMP moderates, so we are well-suited for workloads that have that requirement. We are not yet FedRAMP high certified, so that is something that we're working on as well to try and expand the envelope of openings we can pursue. And then, as you implied, there's also an opportunity to sell EA to these customers, and we're doing that as well. Obviously, all that is baked into our guide for how we think about Q3 and the second half of the year. But we feel good about the opportunities in that segment.
Operator
Thank you. And this does conclude today's Q&A session. I would now like to turn the call back to Dave Itacheria for closing remarks. Please go ahead.
Dave
Thank you. And again, I want to thank everyone for joining us today. We are pleased with our Q2 results and our continued execution against our large market opportunity. While the macro environment remains mixed, our ability to win new business remains strong across both Atlas and EA. And as we look ahead, we believe we are the ideal data layer for AI applications, and we continue to see increased demand from customers to modernize their legacy applications. And last but not least, we will continue investing judiciously and focusing on our execution to capture this long-term opportunity. Thank you for joining us today, and we'll talk to you soon. Take care.
Operator
This does conclude today's conference call. You may all disconnect. Hello. Thank you. Bye. Thank you. Good day, and thank you for standing by. Welcome to the MongoDB's second quarter fiscal year 2025 conference call. At this time, all participants are in a listen-only mode. After the speaker's presentation, there'll be a question and answer session. To ask a question during the session, you will need to press star 11 on your telephone. You will then hear an automated message advising your hand is raised. To withdraw your question, please press star 11 again. Please be advised that today's conference is being recorded. I would now like to turn the call over to your speaker for today. Brian, then you. Please go ahead.
Brian
Thank you, Lisa. Good afternoon, and thank you for joining us today to review MongoDB's second quarter fiscal 2025 financial results, which we announced in our press release issued after the close of market today. Joining me on the call today are Dave Itacheria, President and CEO of MongoDB, and Michael Gordon, MongoDB's COO and CFO. During this call, we will make forward-looking statements, including statements related to our market and future growth opportunities, our expectations for the macroeconomic environment in fiscal 2025, and the impact of AI, the benefits of our product platform, our competitive landscape, customer behaviors, our financial guidance, and our planned investments and growth opportunities in AI. These statements are subject to a variety of risks and uncertainties, including the results of operations and financial conditions that could cause actual results to differ materially from our expectations. For discussion of the material risks and uncertainties that could affect our actual results, please refer to the risks described in our quarterly report on Form 10-Q for the quarter ended April 30th, 2024, filed with the SEC on May 31st, 2024. Any quarterly statements made in this call reflect our views only as of today, and we undertake no obligation to update them except as required by law. Additionally, we will discuss non-GAAP financial measures in this conference call. Please refer to the tables in our earnings release on the investor relations portion of our website for reconciliation of these measures to their most directly comparable GAAP financial measure. With that, I'd like to turn the call over to Dave.
Dave
Thanks, Brian, and thank you to everyone for joining us today. I'm pleased to report that we had a good quarter and executed well against our large market opportunity. Let's begin by reviewing our second quarter results before giving you a broader company update. We generated revenue of $478 million, a 13% year-over-year increase against a very difficult year-over-year compare and above the high end of our guidance. Atlas revenue grew 27% year-over-year, representing 71% of revenue. We generated non-GAAP operating income of $52.5 million for 11% non-GAAP operating margin, and we ended the quarter with over 50,700 customers. Overall, we were pleased with our performance in the second quarter. We had a strong new business quarter, and we saw improving sales productivity year over year. We saw strength across the board with both Atlas and Enterprise Advance exceeding our expectations, demonstrating the enduring appeal of our run anywhere strategy. Our Q2 performance reinforced our belief the slow start to the new business Q1 was purely operational, and we feel good about our new business outlook in the second half of the year. Moving on to Atlas consumption, the quarter played out modestly better than our expectations. Michael will discuss consumption trends in more detail. Finally, retention rates remain strong in Q2, demonstrating the quality of our product and the mission criticality of our platform. Our performance this quarter reinforced our confidence in our ability to execute on a long-term opportunity. As we said before, companies today rely on software to express their business strategy. This trend has driven our success for the past decade, and we anticipate it will continue to do so for the foreseeable future. Even with our success to date, We only have a low single digit share in one of the largest and fastest growing markets in all of software. When you combine this foundational tailwind with the opportunities for our customers to incorporate generative AI into their businesses and modernize their legacy application estate, it is clear that MongoDB has multiple long-term growth opportunities. Turning to AI, AI continues to be an additional long-term opportunity for our business. At the start of the fiscal year, we told you that we didn't expect AI to be a meaningful tailwind for our business in fiscal year 2025, which has proven accurate. Based on recent peer commentary, it seems that the industry now mostly agrees with this view. Companies are currently focusing their spending on the infrastructure layer of AI and are still largely experimenting with AI applications. Inference workloads will come and should benefit MongoDB greatly in the long run, but we are still very early when the monetization of AI apps will take time. AI demand is a question of when, not if, and our discussions with customers and partners give us increasing conviction that we are the ideal data layer for AI apps for a number of key reasons. First, more than any other type of modern workload, AI-driven workloads require the underlying database to be capable of processing queries against rich and complex data structures quickly and efficiently. Our flexible document model is uniquely positioned to help customers build sophisticated AI applications because it is designed to handle different data types, your source data, vector data, metadata, and generated data right alongside your live operational data, updating the need for multiple database systems and complex backend architectures. Second, MongoDB offers a high performance and scalable architecture. As the latency of LLMs improve, the value of using real-time operational data for AI apps will become even more important. Third, We are seamlessly integrated with leading app development frameworks and AI platforms, enabling developers to incorporate MongoDB into their existing workflows while having the flexibility to choose the LLM and other specific tools that best suit their needs. Fourth, we meet or exceed the security and compliance requirements expected from an enterprise database, including enterprise-grade encryption, authorization, and auditability. Lastly, customers can run MongoDB Anywhere on-premise or as a fully managed service in one of the 118 global cloud regions across three hyperscalers, giving them the flexibility to run workloads to best meet their application use cases and business needs. We see three main opportunities where we believe AI will accelerate our business over time. The first is that the cost of building applications in the world of AI will come down, as we've seen with every previous platform shift. creating more applications and more data, requiring more databases. The second opportunity is for us to be the database of choice for customers building greenfield AI applications. While we see that there's tremendous amount of interest in and planning for new AI-powered applications, the complexity and fast-moving nature of the AI ecosystem slows customers down. That's why we launched the MongoDB AI Applications Program, or MAP, which became generally available to customers last month. MAP brings together a unique ecosystem, including the three major cloud providers, AWS, Azure, and GCP, as well as Accenture and AI pioneers like Anthropic and Cohere. MAP offers customers reference architectures, an end-to-end technology stack that includes pre-built integrations, professional services, and a unified support system to help customers quickly build and deploy AI applications. The third opportunity is to help customers modernize their legacy application state. As you know, this segment of the market is a massive opportunity for us, as most of the existing 80 billion plus database industry is built on dated relational architecture. Modernizing legacy applications has always been part of our business, and we have taken steps over the years to simplify and demystify this complex process through partnerships, education, and most recently, our relational migrator product. AI offers a potential step function improvement, lowering the cost and reducing their time and risk to modernize legacy applications. For that reason, earlier this year, we launched several pilots with our customers where we worked with them to modernize mission-critical applications, leveraging both AI tooling and services. The early results from these pilots are very exciting as our customers are experiencing significant reductions in time and cost of modernization. In particular, we have seen dramatic improvements in time and cost to rewrite application code and generate test suites. We see increasing interest from customers that want to modernize their legacy application state, including large enterprise customers. As a CIO of one of the world's largest insurance companies said about our pilot, this is the first tangible return he's seen on his AI investments. While it's still early days and generating meaningful revenue from this program will take time, we're excited about the results of our pilots and the growing pipeline of customers eager to modernize the legacy estate. Finally, I understand that there are a lot of questions about the current business conditions and the macro environment more broadly, so let me give you a sense of what we're seeing across the business. As a reminder, when thinking of the macro influence on our business, it's important to distinguish between consumption of existing workloads and new business. Starting with consumption of existing applications on our platform, this is where we have historically seen a macro impact, as usage of applications impacted by the underlying business conditions of our customers. As we discussed in our last earnings call, in Q1, we did see broad-based consumption growth slowdown, suggesting some macro softening. Our usage trend suggests a similar macro environment in Q2 as in Q1, even though Q2 Atlas consumption growth was modestly ahead of our expectations. Moving on to new business, we generally have not seen the macro environment impact our ability to win new business, and that was true in Q2 as well. We realize that this is different from what you hear from some other software vendors. Ultimately, software application development continues even in uncertain environments as customers know they need to continue investing in internally developed software to run the business as well as to drive competitive differentiation. In addition, we still have relatively low market share in a large market, which means we have an opportunity to gain share in any environment. Now, I'd like to spend a few minutes reviewing the adoption trends of MongoDB across our customer base. Customers across industries and around the world are running mission-critical projects on MongoDB Atlas, leveraging the full power of our developer data platform, including Fanatics, Occidental Petroleum, and Indeed. Fanatics Betting and Gambling, a division of the sports ecosystem company Fanatics, leverages MongoDB to significantly enhance the user experience of their mobile app. Initially, the team launched a platform on Postgres, but faced challenges with scalability, flexibility, and excessive complexity. After migrating to MongoDB Atlas, the team also integrates Atlas Search to provide users with a better experience to find all available betting options. With Atlas handling scaling, partitioning, and operations, developers can focus on writing code and improving the user experience. Looking ahead, Fanatics plans to continue to expand on MongoDB Atlas as they ensure they can operate at scale as they prepare for the start of the NFL season. L'Oreal, McKesson, and Nationwide Building Society are turning to MongoDB to modernize applications. L'Oreal's tech accelerator, a department dedicated to catalyzing digital innovation at L'Oreal, is utilizing MongoDB for an application designed to bring products and solutions to market while quickly improving employee efficiency. The team's previous database solution had limited out-of-the-box functionality and was unable to handle the complex calculations needed to retrieve and restructure large amounts of data from their data warehouse. L'Oreal migrated to MongoDB Atlas to streamline the application architecture and simplify a previously highly complex and time-consuming data access layer. With this migration, L'Oreal achieved a 40-fold performance improvement. On Atlas, the existing code is easier to maintain, more scalable, and more efficient, making life easier for developers. Mature companies and startups alike are using MongoDB to help deliver the next wave of AI-powered applications to their customers, including Delivery Hero, Generali, and Questflow. Delivery Hero, a longtime MongoDB Atlas customer, is the world's leading local delivery platform, operating in 70-plus countries across four continents. Their quick commerce service enables customers to select fresh produce for delivery from local grocery stores. Approximately 10% of the inventory is fast-moving perishable produce that can go quickly out of stock. The company risks losing revenue and increasing customer churn if the customer didn't have viable alternatives to their first choice. To address these risks, they are now using state-of-the-art AI models and MongoDB Atlas Vector Search to give hyper-personalized alternatives to customers in real time if items they want to order are out of stock. With the introduction of MongoDB Atlas Vector Search, The data science team recognized that they could build a highly performant real-time solution more quickly and for less cost than alternative technologies. In summary, we had a healthy Q2 with both Atlas and EA exceeding expectations. We saw a strong new business quarter and improved sales productivity, and we're confident in our ability to keep winning new business in the second half of the year. Looking forward, we see great opportunity to help our customers modernize legacy applications and build the next generation of AI-powered applications. With that, here's Michael.
Michael
Thanks, Dave. I'll begin with a detailed review of our second quarter results and then finish with our outlook for the third quarter and full fiscal year 2025. First, I'll start with our second quarter results. Total revenue in the quarter was $478.1 million, up 13% year-over-year and above the high end of our guidance. As a reminder, while we have difficult compares throughout fiscal 2025, we're facing a particularly difficult year-over-year comparison in Q2, given we had a number of large multi-year partnership licensing deals in Q2 last year. Under 606 accounting rules, we recognize the license component up front, which creates a difficult compare a year later. Shifting to our product mix, let's start with Atlas. Atlas grew 27% in the quarter compared to the previous year and now represents 71% of the total revenue, compared to 63% in the second quarter of fiscal 2024 and 70% last quarter. We recognize Atlas revenue primarily based on customer consumption of our platform, and that consumption is closely related to end-user activity of their applications. Let me provide some context on Atlas consumption in the quarter. First, as a reminder, in Q1, Atlas consumption growth was below our expectations, so we updated our growth assumptions for the remainder of the year. In Q2, Atlas consumption growth was modestly ahead of those updated expectations across the board. While this is encouraging, Consumption growth is still below our original forecast from the beginning of the year. Turning to non-Atlas revenue, non-Atlas came in modestly ahead of our expectations in the quarter, as we continue to have success selling incremental workloads into our existing customer base. On a year-over-year basis, non-Atlas revenue was down 13% due to the especially difficult compare I referenced earlier. Turning to customer growth, during the second quarter, we grew our customer base by approximately 1,500 customers sequentially, bringing our total customer count to over 50,700, which is up from over 45,000 in the year-ago period. Of our total customer count, over 7,300 are direct sales customers, which compares to over 6,800 in the year-ago period. The growth in our total customer count is being driven primarily by Atlas, which had over 49,200 customers at the end of the quarter compared to over 43,500 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 adding incremental Atlas workloads. Continuing on, in Q2, our net AR expansion rate was approximately 119%. The decline versus historical periods is attributable to a smaller contribution from expanding customers. We ended the quarter with 2,189 customers with at least $100,000 in ARR and annualized MRR, up from 1,855 in the year-ago period. Moving down the income statement, I'll be discussing our results on a non-GAAP basis unless otherwise noted. Gross profit in the second quarter was $360.8 million, representing a gross margin of 75 percent, which is down from 78 percent in the year-ago period. Our year-over-year margin decline is primarily driven by a lower mix of high-margin upfront license revenue compared to last Q2, as well as Atlas growing as a percentage of the overall business. Our income from operations was $52.5 million, or an 11% operating margin for the second quarter, compared to a 19% margin in the year-ago period. The primary reason for our more favorable operating income results versus guidance is our revenue outperformance. In addition, Q2 operating income benefited from the timing of certain marketing and other spend, which we now expect to incur in the second half of the year. Net income in the second quarter was $59 million, or 70 cents per share, based on 83.8 million diluted weighted average shares outstanding. This compares to a net income of $76.7 million, or 93 cents per share, on 82.5 million diluted weighted average 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. Operating cash flow in the second quarter was negative $1.4 million, After taking into consideration approximately $2.6 million in capital expenditures and principal repayments of finance lease liabilities, free cash flow was negative $4 million in the quarter. This compares to free cash flow of negative $27.3 million in the year-ago period. In addition, this quarter, we also received $170.6 million in cash from the settlement of the capped calls associated with our 2024 convertible notes. I'd now like to turn to our outlook for the third quarter and full fiscal year 2025. For the third quarter, we expect revenue to be in the range of $493 million to $497 million. We expect non-GAAP income from operations to be in the range of $57 million to $60 million, and non-GAAP net income per share to be in the range of 65 cents to 68 cents, based on 84.6 million estimated diluted weighted average shares outstanding. For the full fiscal year 2025, we expect revenue to be in the range of $1.92 billion to $1.93 billion, non-GAAP income from operations to be in the range of $187 million to $195 million, and non-GAAP net income per share to be in the range of $2.33 to $2.47, based on 84.3 million estimated diluted weighted average shares outstanding. Note that the non-GAAP net income per share guidance for the third quarter and full fiscal year 2025 includes a non-GAAP tax provision of approximately 20%. I'll now provide some context around our updated guidance. First, because of the stronger than expected Atlas consumption in Q2, our starting Atlas ARR for the back half of the year is higher than we anticipated in May. As a result, we're raising our Atlas revenue forecast to reflect that higher starting point. We are not changing our underlying Atlas Q3 and Q4 consumption growth rate assumptions. Second, we are slightly increasing our EA assumptions for the rest of the year to reflect the strength of the second half EA pipeline. Finally, thanks to the performance in Q2 and the increased revenue outlook, we now expect a 10 percent operating margin at the midpoint of our fiscal 25 guidance. We will continue investing to capture our long-term opportunity with a focus on our strategic priorities. Separately, although we don't provide guidance on cash flow, we wanted to call out two items that will impact fiscal 25 cash flow, but benefit our long-term financial profile. In Q2, we started paying some of our cloud provider commitments upfront in exchange for better economics. We see this as a very low risk, high ROI use of our cash, and one that will benefit our gross margin going forward. These prepayments will represent a negative impact to our operating cash flow in the back half of the year, of roughly $20 million per quarter. In addition, in Q3, we will incur approximately $20 to $25 million of CapEx to acquire IPv4 addresses, which will allow us to further reduce our cloud infrastructure costs in the future. To summarize, we're pleased with our second quarter results, and especially our ability to win new business. We have a small share, one of the largest and fastest growing markets in all of software. The secular tailwinds at our back are getting stronger in the age of AI, and we're excited about the future. We'll continue investing judiciously and focusing on our execution to capture this long-term opportunity. With that, we'd like to open it up to questions.
Operator
Operator? Thank you. As a reminder, if you would like to ask a question, please press star 11 on your telephone. We also ask that you limit yourself to one question and one follow-up. As well, please wait for your name and company to be announced before you proceed with your question. One moment while we take the first question. And the first question today is going to come from Sanjit Singh of Morgan Stanley. Your line is open.
spk08
Yes, hi. Thank you for taking the question. Congrats on Q2. Nice to see Mongo beating and raising again. David, I wanted to follow up on some of the changes, the operational changes you're making to the business that you announced last quarter, just in terms of the update, in terms of realigning some of the sales incentives to drive higher workload quality while maintaining some of that record workload growth that you saw last year. Could you sort of walk us through what the changes were you put in place and how much confidence you have that you will drive on better workload growth as we get into early next year?
Dave
Yeah, sure. So, first of all, thanks, Sanjit. Just a reminder for everyone, we made some slight incentive comp changes to really get our reps to focus, to have a little bit more balance on size versus volume of workloads acquired. These changes were well received by the field. We had a good workload quarter, as you just described. Workload starts small, so it's too early. While we're pleased with our results, obviously, it's too early to declare a victory and to really see if this is going to have a material impact and changes, but clearly we're happy with what we've seen so far.
spk08
Right. And then as a follow-up for Michael, I remember at the Q4 last year, you gave us the update on sort of the unused credit dynamics. As we start to think about the back half of this year, is there anything else that we should think about that? I mean, you just mentioned some of the cash flow impacts just now. Is there anything on sort of like the revenue side of the equation that we should keep in mind? I kind of noticed at the end of this year, there's this more holiday days in the quarter in Q4. Does that have an impact? Anything that we should be, that you would like to call out in terms of our modeling when we think about expectations for Atlas more broadly going into the end of the year?
Michael
Yeah, I think there are a few things to take into account that have been throughout the year, but of course are reflected in the updated guide. Specifically on Atlas throughout the year, we have tough compares given the headwinds from the lack of the unused commitments. That's obviously hardest. As the year goes on, that sort of number builds, so that makes for a tough year-over-year comparison. I think the other thing to think about in the context of Atlas is, you know, the ending year dynamic is really a compounding over the course of the year. And as we talked about at the beginning of the year, we got off to a slower start and some of the prior year workloads, those cohorts were growing at slower rates. And so that does affect and kind of compound over the course of the year. And so I think that's important to keep in mind and is factored into our guidance. And then looking at the revenue picture more broadly, moving away from Atlas and thinking about EA, we've had throughout the year a headwind on EA, and we'll have that as a multi-year headwind in the back half as well. The only other things I think I would call out is we saw a slower seasonal rebound in Q1, and so we're expecting that slower seasonal rebound to occur in Q3 as well. And then, as you mentioned, Q4 is a seasonally weaker quarter. And so all those are some of the considerations to think about when you're thinking about the rest of the year.
Atlas
Appreciate that, Michael.
Operator
Thank you. One moment for the next question. And our next question today will be coming from the line of Remo Linschild of Barclays. Your line is open.
metadata
Hey, thank you. Two quick questions from me. Michael, on the EA, you kind of talked about like a slightly better pipeline in the second half. Can you talk a little bit, is that kind of still the continuation of what we saw in Q2, Q3 last year, where people wanted to modernize their kind of self-service footprint ahead of a move to the cloud, or what's the dynamic there? And then one for Dave, like, There's a big debate about reference architecture on AI, and I think it's a little bit too early, but it comes up with investors a lot. What are you seeing there? I know AI infrastructure, we're still in the early innings, but what are you seeing there in terms of people engaging with you? Thank you, and good to see you back on track. Thank you.
Michael
Thanks, Raimo. On the EA question, I would say there are lots of reasons why we're seeing strengths of EA I wouldn't uniquely or exclusively tie it to, you know, AI. I think it's broadly, you know, in support and shows the strength of the run anywhere strategy. The other thing that I'd call out is, you know, given the 606 dynamics as it relates to EA, we're always sensitive to sort of the multi-year aspect and that's sort of what's, you know, provided, you know, given the strength of multi-year in fiscal 24, some of the tough compares here in fiscal 25. I think what we've seen in terms of the pipeline is, is strength in EA broadly. So not just a multi-year phenomenon, but really sort of a volume phenomenon as well that we're seeing when we kind of look out on the horizon, and that is reflected in the updated guide.
Dave
Thanks, and Raimo, on the question about AI, I think in terms of reference architectures, I think it's important to understand, and I said this in the prepared remarks, is that Unlike most of the workloads, AI-driven workloads really require the underlying database to be capable of processing queries against very rich and complex data structures, both quickly and efficiently. And MongoDB is well-positioned to do that. We can basically unify and handle source data, vector data, metadata, generated data from your LLM right alongside your live operational data. And then as the performance of these LLMs and latency of these LLMs increase, accessing real-time data becomes really important. Like, say you're calling and talking to a customer support chatbot, you want that chatbot to have up-to-date information about that customer so that they can provide the most relevant and accurate information possible. You know, there are some questions about LLMs, whether a general-purpose LLM or a fine-tuned LLM, what the tradeoffs are. Our belief is that given the performance of LLMs, you're going to see The general purpose LLMs probably win and will use RAG as the predominant approach to marry generally available data with proprietary data. And then you are starting to see things like advanced RAG use cases where you get much more sophisticated ways to ask complex questions, provide more accurate and detailed answers, and better adapt to different types of information and queries. That's what we're seeing. I think it's a quickly evolving space, but we feel very good about our positioning for AI, even though it's still very early days.
Operator
Okay, perfect. Thank you. Thank you. One moment for the next question. And our next question will be coming from Cash Reagan of Goldman Sachs. Your line is open.
Cash Reagan
That's perfect. I was about to call Raimo and say, can you please explain reference architectures? I'll buy you a nice glass of wine. So hopefully he'll take me up on that offer because that was a very technical question. Thank you. But anyway, coming back to the call here. Dave, a question for you. Why is EA doing so well? So you talked about a pipeline in the second half of the year. Last year you had some significant wins for EA. We were supposed to be on this cloud journey. So Definitely Atlas has reached parity in many senses. It can support big scale applications, whatnot. So curious to get your thought on why is EA still an important piece of the business? And I suppose when you look at AI, could something surprise us? I know that you've had this view that we're building infrastructure first and then the platform, then applications. Could it be by any chance a different way to approach AI? in this cycle and that we don't really need applications, but somehow these LLMs are going to be a perfect replacement for the way we think about old world applications. I'm just curious to see what the devils advocate. You might be somewhat skeptical of the whole AI applications build out on top of the infrastructure and platform. Thank you so much.
Dave
Yeah, so thanks, Cash. What I would just say is, you know, we did have a better than expected both Atlas and EA quarter. There's no question about that. I don't want to suddenly say there's some inflection point on EA and our business. I think I would really attribute it to, one, better execution, and two, customers do really appreciate our run-anywhere strategy. There's still lots of customers, either for regulatory reasons or other reasons, who want to run workloads on-premise, and they're not going away. I mean, we've been in this cloud journey for 10 years, and if Some workloads are just very hard to move to the cloud, or some workloads for many customers don't make sense to move to the cloud, at least not anytime soon. So the fact that they can build it on MongoDB and have the optionality to move it to the cloud later very easily is something that's very compelling for customers. With regards to AI, I mean, we predominantly see most of the AI workloads in the cloud, but there are definitely lots of customers who are looking at using open source LLMs in particular things like LAMA, and running those workloads locally. Obviously, it means that they also have to have access to NVIDIA hardware like GPUs, but we do see some customers do that. Again, I wouldn't suggest that that's also an inflection point or cause for the EA outperformance. I think it's very, very early days, and most of those are experiments that customers are running. But clearly, we feel really good about running anybody's strategy, and as we've said in the past, We are investing in basically introducing search and vector search to our community product, and that will then show up in EA. So EA is definitely an area where we're also investing from a product point of view.
Atlas
Thank you. One moment for the next question.
Operator
And our next question. We'll be coming from the line of Tyler Radke of Citi. Your line is open.
Tyler Radke of Citi
Hi, this is Tyler Radke. I think the question was for me.
Tyler Radke
I couldn't hear the name, but thanks for taking the question. Dave, you talked about a post-gress displacement in the quarter, and I think that's the first time you've talked about that, at least in recent quarters. So I was wondering if you could just sort of James Ingram, LGO Admissions, frame for us what use cases, do you do come across them, I know it is popular in terms of a sequel displacement. James Ingram, LGO Admissions, For migrating our legacy applications, if you could just frame for us sort of the competitive environment in postgres both the open source stuff and some of the new you know venture backed startups in the space, thank you.
Dave
Yeah, sure. So thanks for the question. Yeah, it's important to understand that Postgres has been around for almost 40 years. I mean, Postgres, the name is termed from post-ingress. So that technology has been around a long time. As you said, they're really the beneficiary of lift and shift from Oracle, SQL Server, and MySQL. So they're kind of consolidating the relational market. In terms of why do we compete or why do we win, I would say it's a few things. One, our schema flexibility is MongoDB has a very flexible scheme allowing you to store documents in a JSON-like format, so this is beneficial for application structures that evolve over time. We can horizontally scale, so we're making it very easy to distribute data across multiple servers or virtual servers for applications that require massive amounts of data. Performance of large data sets, again, we can handle that better than Postgres. The built-in sharding allows for automatic data distribution. And then we also, I think, developer productivity, the JSON-like format and flexible schema can lead to faster development cycles, especially for customers who really work in agile environments. So we feel like we compete. Our win rates against Postgres are high. But again, there's lots of decisions being made where we're not party to, where people are just doing a lift and shift off of. you know, a legacy platform, or they just want to stay on relational because that's what they know. And obviously it's our job to educate them on the benefits of MongoDB, but we feel good about our competitive position against Postgres.
Tyler Radke of Citi
Thank you. And follow-up question for Michael.
Tyler Radke
You talked about how consumption in Atlas in the quarter tracked a bit better than planned. Sounds like it was not a macro-related improvement. What do you think the driver that was and are you seeing any improvement in some of the recently acquired? workloads those starting to ramp up better or maybe they have different seasonal patterns than any thought any any color on sort of that driver of Outperformance would be helpful.
Michael
Thank you Yeah, sure, I would say that yes q2 consumption growth was better than our expectations that was great to see and I would describe it as within a reasonable or typical range of outcomes, and so there are no signs that we've seen that would specifically point to any material changes in the underlying macroeconomic environment, better or worse. We've certainly seen, to your question on the workloads and some of the cohorting, we did see a little better performance there as well. and in line with what we saw elsewhere, but they're still below our original expectations. So I think that's probably the key couple things there.
Operator
Thank you. Thank you. One moment for the next question. And our next question for the day will be coming from Brad Reback of Stifel. Your line is open.
Brad Reback of Stifel
Great. Thanks very much. Dave, on your commentary around the modernization pilots, how should we think about timing as that being a bit of a tailwind to the overall growth rate?
Dave
Yeah, so again, just to make sure everyone understands, the legacy relational application market or database market is quite large. It's over $80 billion, and it's a massive opportunity for us. And since day one, since our IPO, we've been getting customers to migrate off relational to MongoDB. But one of the biggest friction points has been that while it's easy to move the data, you can map the schema from a relational schema to a to a document scheme and you can automate that, the biggest stumbling block is that the customer has to, or some third party, has to rewrite the application, which by definition creates more costs, more time, and in some cases more risk, especially for older apps where the development teams who built those apps no longer exist. So what's been compelling about AI is that AI has finally created a shortcut to overcome that big hurdle of And so essentially you can start basically diagnosing the code, understand the code, recreate a modern version of that code, and generate test suites to make sure the new code performs like the old code. So that definitely gets people's interest because now all of a sudden what may take years or multi-years you can do in a lot less time. And the pilots that we have done, the time and cost savings have been very, very compelling. That being said, we're in the very early days. There's a lot of interest. We have a growing pipeline of customers across, frankly, all parts of the world, from North America to EMEA and even the PACRIM. And so we're quite excited about the opportunity. But again, I would say it's very early days. But there's a number of reasons why I would say that customers are very focused on this, what the cost of licensing and maintaining legacy apps is This is becoming too high to bear. In many cases, the regulatory and compliance requirements are forcing customers to upgrade. There's a whole end of life of critical technologies, notably Sybase, that's forcing customers to act. There's a ton of technical depth on these legacy platforms that limits the organization's flexibility to do things with AI. And candidly, customers have also soured on the traditional approach of using large systems integration projects that are very costly and take a long time. So this whole approach is definitely getting their attention.
Brad Reback of Stifel
That's great. And then one fast follow-up. It feels like we've spoken about EA more than typical on this call, which is great. As we think about the back half pipeline, is the composition similar in so much as it's predominantly existing customers, or are you beginning to see an uptick in net new customers there as well? Thanks.
Dave
Yeah, I would say that it's predominantly existing customers that are doing this, you know, who are maybe doubling down on EA and expanding the footprint to EA. And that's typically the driver for our EA business.
Brad Reback of Stifel
Thanks very much.
Dave
Thank you.
Operator
Thank you. And one moment for the next question, please. And our next question will be coming from Carl. Chris Fett of UBS. Your line is open.
Carl Kirsvedt
Okay, great. Hey, Dave, at the beginning of the call, you mentioned that Mongo's slow start to the year was, quote, purely operational. But on the Q1 call, when you were describing what happened, you described it as very much a broad-based macro issue. So I'm curious, should I interpret that comment as you, Mike, and the team have done a bit of a rethink and might be changing modestly, at least your explanation about the Q1 results?
Dave
Yeah, actually, just to be clear, I think just to make sure everyone understands, what we called out was that the consumption of existing workloads was broad-based. The slowdown was broad-based across both geos and channels. It was new business just got off to a slower start. So that was what we called out in our Q1 call. And with our Q2 call and results, as you can see, new business performed better than expected on both Atlas and EA, which is why we believe the Q1 new business was an operational issue, not a macro issue. The macro issue was all about the slowdown in consumption because Q1 typically tends to be a seasonally strong quarter for us, and that didn't happen.
Michael
Yeah, I think that's really important, Carl, just to keep in mind when we talk about macro, it could have those two different effects, right? The new business piece and then the expansion of existing workloads. And so what we were talking about, as Dave said in Q1, was the existing workload expansion. The operational piece was on the new business side. And I think the key thing that we've observed really in a wide range of macroeconomic conditions in our almost seven years as a public company is with the exception of that one quarter, we've been able to execute quite well in the new business opportunity really in most environments, given that we have a huge market, relatively low share, strong product, and a talented team that executes well. And so I think that's probably helpful just to understand as we're sort of using the terms just to make sure we don't trip ourselves up.
Carl Kirsvedt
Yep. Okay. That's great. And then my follow-up for you, Mike, is When you were pressed in prior questions about the second half, most of the variables that you brought up were frankly headwinds, things like tough compares. If I look objectively at your 3Q total revenue guide of 497, and this is my assumption, but I assume a normal beat, that's going to result in sequential total revenue growth that's actually the highest, I think, that MongoDB has ever put up. Clearly, there's something good that you're embedding in that 3Q guide. What is that specifically?
Michael
Yeah, so I'm not sure I'd follow all the math on the fly, and I'm happy to follow up, obviously, with any of that. But if I just think about the Q3 guide, I won't belabor the headwinds, but they exist, and we've sort of been through those, as you called out. But maybe to talk about the couple of positive things that we see that lead to sort of our improved outlook or increased revenue forecasts. for the balance of the year. I'd say it's really two things. In Atlas, it's reflecting the stronger-than-expected Q2, and therefore the starting ARR at the beginning of Q3 is higher. Because we haven't seen any change in the underlying macro, the actual growth assumptions for Q3 and Q4 we have remain unchanged, and so you're just applying growth to a higher base, but as the base keeps getting bigger and bigger, that does matter, and that flows through. And then secondly was the positive impact of the increased strength in the EA pipeline for the second half of the year and the impact on that revenue. So those would probably be the two things that I'd call out on the positive side of the revenue in addition to, of course, having everyone keep in mind the tough compare and some of the headwinds we talked about.
Q4
Okay. That's helpful, Mike. Thanks a lot.
Operator
Thank you. Thank you. And one moment for the next question. Our next question will be coming from Jason Ather of William Blair. Your line is open.
Jason Adler
Yeah, thank you. I'm trying to understand the Q4 implied revenue guide, Michael. You know, just the implied sequential growth there is only 1.5% if I look at your full year relative to what you said specifically on Q3. That would be well below the historical seasonal pattern in Q4. So I just want to understand, is there something going on in Q4 that's different, just sequentials? I understand the year-over-year is tough, but the sequentials seem well below where you've normally been. Is there something specific that we should think about there?
Michael
Yeah, I think there are two things. So obviously, it sounds like you get the year-over-year, but just to make sure everyone's sort of on the same page is making sure that people understand you know, the tough compare for Atlas specifically around the unused commitments and then EA, the multi-year, and then more broadly, you know, we had one, we had less growth in Q1. That compounds, and then also we talked about how we have a, we're assuming just in the same way that we saw in Q1, we're assuming less of a seasonal rebound in Q3 That has implications for Q4.
Jason Adler
Gotcha. Okay. And then just one quick follow-up on gross margins. Did you talk about Atlas gross margins in the quarter? And then how material will the impact be from the prepayments and the IPv4 purchases on Atlas gross margins over time?
Michael
Yeah, so we didn't specifically break out Atlas gross margin in the quarter, but they continue to be lower, but obviously we've pretty significantly shrunk the delta. That does explain some of the comparison on the year-over-year basis. We didn't give specific guidance going forward. If you think about the two changes that we called out on the cash flow side, those will benefit us in terms of gross margin. There won't be a significant benefit in fiscal 25, so we'll obviously talk about that more when we get to the fiscal 26 guide, but it's a very excellent use of our cash and good ROI in terms of improving those economics further. Thank you.
Operator
Thank you. One moment for the next question. And our next question will be coming from Brad Sills of Bank of America. Your line is open.
Brad Sills
Oh, great. Thank you so much. Great to hear the continued strength in new workloads here. That's been a very consistent theme here throughout all of this. I did want to ask about some of the newer services like vector and stream processing. How are those contributing to the strength you're seeing or sustained strength you're seeing in new workloads?
Dave
Yeah, Brad, thanks for the question. So in terms of search, we're seeing solid momentum in search. We're having success with that business. It's starting to grow, and we just introduced a new capability called Search Nodes, which allows customers to optimize their search deployments by asymmetrically scaling specific nodes dedicated for search versus the rest of the nodes on their cluster. It also helps dealing with use cases that are very search-intensive, One of the largest gaming companies in the world re-platformed the content moderation platform from DocDB, Elastic, and Dynamo to Atlas and Atlas Search, and using Atlas Search nodes for workload isolation and high performance. On Vector, we're continuously seeing growth and adoption, and we see Vector is effective in attracting new customers to the MongoDB platform. A world-renowned financial news organization, which was already running on Atlas, migrated from Elastic Search to Atlas Search using search nodes to take advantage of our vector search capabilities to build a site search that combines lexical search with semantic search to find the most relevant articles for user query. And a European energy company built a geospatial search application using Atlas and search and vector search. And the Atlas was built on-prem and to clouds to vectorize geospatial data and facilitate research and discovery. And then we recently, you know, announced streaming, or stream processing, I should say, you know, GA and .local New York in May, and we've seen strong interest. It's still early days, but we're seeing interest from a variety of industries, ranging from automotive to retail to transportation, who all want to work with streaming data and want to be able to take actions on that data to, you know, drive their business. And so... So customers are very pleased with the performance of the product and how easy it is to use. But again, it's just early days since we just launched it, only in May.
Brad Sills
Wonderful. Great to hear it. One more, if I may, please. On the last earnings call, you mentioned how the slowdown in consumption was broad-based across industries and workload types, which led you to believe that it was a macro-related impact. With some of the improvement you've seen this quarter, I know it's early, but was that also broad-based, and could you deduce that maybe that might mean some improvement in the underlying macro, or was it more outsized to certain types of industries and services?
Michael
Thank you. I would describe it as broad-based, but I would describe it within sort of a reasonable range of outcomes, and so no clear indication that macro is improving or deteriorating, and just a good quarter.
Brad Sills
Wonderful. Thank you.
Operator
Thank you. One moment for the next question. And our next question will be coming from Rashi Jaleria of RBC Capital Markets. Your line is open.
Rashi Jaleria
Wonderful. Hey, this is Rashi Jaleria from RBC. Two questions. Maybe, Dave, I wanted to first start out going back to some of what you talked about, which is the MongoDB versus Postgres debate. You know, one of the kind of popular theories out there or debates out there, is which architecture is better suited to new generative AI applications, especially if we truly do believe AI will lead to a replatforming of software similar to what we saw with the cloud. I know you talked a little bit about reference architecture earlier in Q&A, but maybe could you walk us through why you believe MongoDB is better suited to new generative AI native applications versus Postgres? And then I've got a quick follow-up.
Dave
Yeah, sure. So very quickly, the reason we believe that we're well positioned to win these new workloads is that AI-driven workloads require the underlying database to be capable of processing queries against rich and complex data structures. As you know, with AI, the data structures can be very, very complex. So that means that the data can be large and obviously not in any consistent size. MongoDB is designed to handle these different data structures, and I talked about, you know, we can help unify metadata, operational data, vector data, and generate it all in one platform. Relational databases, and Postgres is one of them, have limitations in terms of how they can handle different types of data. In fact, When the data gets too large, these relational databases have to do what's called off-row storage, and it creates a performance overhead on these relational platforms. Postgres has this thing called TOAST, which stands for the Oversized Storage Attribute Storage Technique, and it's basically a way to handle these different data types but it creates a massive performance overhead. So we believe that we are architecturally far better for these more complex AI workloads than relational databases. And we believe that ultimately we are the ideal data layer for AI applications. Obviously it's early days. As I said, we haven't seen a lot of inference workloads in production, but architecturally we feel very good about our ability to compete against relational databases and Postgres in particular. That being said, Postgres is popular because they're the beneficiary of people who want to stay on relational and who've kind of lived in relational for the last 30, 40 years. But head-to-head, we feel like we can win our share of business.
Rashi Jaleria
Got it. Thank you. That's really helpful. And then maybe just sticking on the theme of AI, we've talked before in the past that AI is just driving a lot of new code, making developers significantly more productive. Have you seen that behavior in any of your existing customers on Atlas or where maybe their utilization rate goes up or the number of applications built per customer goes up. Anything like that, that understanding it super early might at least give you some early indication that the increased developer productivity could turn into a real tailwind in terms of consumption of the underlying Atlas for you. Thank you.
Dave
Yeah, so this is a common question I ask our customers when I meet with them in terms of what code generation tools are they using and what benefits they're gaining. The answers tend to be a little bit all over the map. Some people see... 10, 15% productivity improvements. Some people say 20, 25% productivity improvements. Some people say it helps my senior developers be more productive. Some people say it helps my junior developers become more like senior developers. So the answers tend to be all over the map. There's no question in my mind that with this platform shift, like previous platform shifts, the cost of building apps will come down. And so, by definition, more apps will be produced, which will generate more data, which requires more databases. Now, obviously, we're in the early days, and much like, you know, I've lived through other platform shifts, like going to the Internet. A lot of people built out the, you know, there's all this dark fiber that was built out to service all this potential demand. It didn't happen. People thought, It was maybe overinflated, and then before we knew it, the Internet transformed the way we work, the way we live, the way we socialize. And so I think that the same thing is happening here. I think it's all on the come, but I think we're well positioned for that opportunity.
Operator
Wonderful.
Dave
Thank you.
Operator
Thank you. One moment for the next question. And our next question will be coming from Brent Braceland of Piper Sandler.
spk14
Your line is open.
Atlas
Thank you. This is Brent Braceland, Piper Sandler. Dave, I wanted to go back to Atlas. And the question here is, given all the moving parts here around accounting, tough compares, unused commitments, Atlas growth looked really strong. I think 30% plus on a normalized basis after adjusting for unused commitments. That's before any benefit from AI. Can you just go back to the growth algorithm for Atlas here? I ask because 30% normalized growth would be 3x faster than industry growth, implying a meaningful share capture of new software builds. maybe be worthwhile if you just go walk through what, what is driving the momentum on, on the Atlas side that is again, two, three X faster than the industry.
Dave
Um, Well, what I would say is obviously we're pleased with our results. What I would say is one, Atlas is the most widely available cloud service by definition because it runs across the largest hyperscalers in 118 regions around the world. So no matter who you are, you can run on Atlas almost anywhere in the world. Second, because MongoDB is a general purpose database, you can use us for almost any conceivable use case. So we're not limited to some narrow set of use cases or narrow set of users who have a very particular set of needs. That helps. Three, we have a large community of developers. We have millions of developers all around the world from Asia to North America to EMEA who are using MongoDB today. to address some problem. And so that also obviously, because of the popularity of MongoDB, we are the world's most popular modern database. I think that also speaks to the strength of the product. And then I would say we try and execute well. Obviously, we got off to a slower start than we liked in Q1, but we are very, very focused on execution. We have a really good team, both in terms of our go-to-market team as well as our product and the supporting ecosystem teams, our partner team, engineering team. And so we all try and put our heads down, and we were obviously disappointed by Q1, and we're really focused on growing the business as fast as we can.
Atlas
Perfect. And then, Michael, just a quick follow-up on op margins. Guidance implies 10% plus op margins in the second half, particularly as the revenue run right here gets to $2 billion exiting the year. Can you continue to invest? in sales capacity while balancing double-digit margins at this scale?
Michael
Yeah, so specifically around the second half, we've obviously increased the full year guide, and that flows through. We did talk about some of the timing issues between halves. We are incrementally, on a marginal basis, investing in the back half of the year in our AI initiatives. But I think if you look at the two-year view, We're still making 500 basis points of margin progress over the last two years. So, yeah, I think it's a balancing act. Brent, I think the key thing is we have such a large opportunity and such attractive investment areas that we'll try and continue to do both.
Operator
Helpful. Thank you. Thank you. One moment for the next question. And our next question will be coming from Eric Heap. of KeyBank Capital Markets. Your line is open.
Eric Heath
Hey, good afternoon. Thanks for taking the question. Just a couple of quick ones for me. Mike was curious to get a better understanding on just maybe some of your assumptions on why you're expecting a slower than typical rebound coming out of 3Q. And then, Dave, thinking about the third quarter here and the Fed fiscal year end, just what are some of your expectations for that article, which I assume is usually a big EA opportunity? And thanks.
Michael
Yeah, so quickly, just on Q3, I think we're informed by our recent experience, including the fact that the seasonal rebound in Q1 of this year was slower than what we've typically seen. And so that's what we're taking into account when we think about Q3, and that's really what's driving it.
Dave
Yeah, and I think you said, if I heard you correctly, you're talking about the Fed, the public sector and the Fed in particular. Yes, I mean, obviously their fiscal year ends in September. We... We have a good team there. We have FedRAMP moderate, so we are well-suited for workloads that have that requirement. We are not yet FedRAMP high certified, so that is something that we're working on as well to try and expand the envelope of options we can pursue. And then, as you implied, there's also an opportunity to sell EA to these customers, and we're doing that as well. Obviously, all that is baked into our guide for how we think about Q3 and the second half of the year. But we feel good about the opportunities in that segment.
Operator
Thank you. And this does conclude today's Q&A session. I would now like to turn the call back to Dave Itacheria for closing remarks. Please go ahead.
Dave
Thank you. And again, I want to thank everyone for joining us today. We are pleased with our Q2 results and our continued execution against our large market opportunity. While the macro environment remains mixed, our ability to win new business remains strong across both Atlas and EA. And as we look ahead, we believe we are the ideal data layer for AI applications, and we continue to see increased demand from customers to modernize their legacy applications. And last but not least, we will continue investing judiciously and focusing on our execution to capture this long-term opportunity. Thank you for joining us today, and we'll talk to you soon. Take care.
Operator
This does conclude today's conference call. You may all disconnect.
Disclaimer