We’ve sent a lot of earnings reviews this season. All of those and my current portfolio/performance can all be found here.

Housekeeping:

I hear you. The 101 sections are important, but they're too long to include in quarterly earnings reviews. As suggested by readers, I will be making a 101 archive that I can link back to after this season. That will make it easy to access for those needing a refresher (or new readers) and easy to skip for those not needing the info.

MongoDB is a key player in data storage and analytics, with a document-oriented setup. This differs from legacy relational-style databases and next-gen versions like Snowflake’s. How so? Relational databases store data in rigid rows and columns linked by pre-set relationships. These databases look like giant Excel spreadsheets and use structured query language (SQL) to work. The datasets are fixed, with formatting and filtering more limited. 

Per MDB leadership, relational databases cannot seamlessly handle unstructured data like MongoDB’s “not only SQL” (NoSQL) format can. MDB's document-oriented setup organizes and uses unstructured data more effectively, which is vital in the age of GenAI and agentic AI. Relational and SQL are fine for traditional machine learning, but less effective for agentic, multi-step, goal-oriented tasks. It’s worth noting that Snowflake and Databricks argue that their upgraded relational databases can handle scalable unstructured and semi-structured data ingestion. Still, MongoDB is widely believed to be a leader in the space.

Atlas’s cloud-native database architecture uses a group of servers (or a cluster) to actually store data for eventual product creation within its platform. The nature of MongoDB’s Atlas product enables easy cluster additions and subtractions as needs fluctuate. That greatly helps control costs and improve performance. It also offers MongoDB Search for data querying with a highly capable Vector Search offering for AI use cases as well. Vector Search allows clients to conveniently scrape insights from data via theme-based querying rather than just word-based. MongoDB’s retrieval-augmented generation (RAG) perfectly complements these search products, as it enriches large language model (LLM) outputs with external context as needed to improve querying precision. This way, search products deliver optimally relevant and accurate responses. 

More Core Atlas Products:

Product Expansion & AI:

Beyond its aforementioned Vector Search offering, MongoDB excels in automating the cumbersome data modernization work for migrations. Its Relational Migrator (RM) handles and takes client headaches out of this process. Now, it’s focused on leveraging RM, allowing companies to actively re-code their legacy applications with end-to-end support. They call this the “modernization factory.” With this holistic strategy, MongoDB becomes a more prevalent and relied-upon vendor, as it directly handles data and software modernization through its singular platform.

Voyage AI (recently acquired) is a key player in AI model accuracy. While models routinely create delightful, jaw-dropping experiences, they’re also wrong a lot. Hallucination rates (wrong answer rates) can often reach 25%, making models really only useful when you know what a right answer should look like and are double-checking. That limitation puts a tight ceiling on utility and is what Voyage aims to fix. 

It has two products. First is its set of Vector Embedding Models. As leadership puts it, these are the “bridge between models and a client's private data.” They allow for meticulous information transfer into models and significantly overlap with its RAG offering with the aim of ensuring maximum answer accuracy. Voyage's embedding models are what make data usable for RAG and their database offerings by numerically organizing standardized insights. All of this helps with machine learning accuracy rates and uncovering patterns to reduce model mistakes. Secondly, it offers reranking models. As an LLM query yields a batch of sources to begin forming a final response, these reranking models (as the name indicates) rank the available sources to curate the final answer from a better batch of information. It’s a second filter that makes sure models are using only the best, most accurate information. With Voyage, MDB gains a broader suite of GenAI services to round out a more cohesive platform.

More things to Know:

Atlas App Services is a serverless toolkit for developers to build modern apps. It provides a slew of templates, integrations and guardrails to diminish friction. MongoDB AI Applications Program (MAAP) is an AI adoption acceleration initiative. It helps customers effectively use GenAI tools (with Voyage AI giving them more confidence in doing so). It frees customers to embrace new AI-enabled technologies like Vector Search.

Enterprise Advanced (EA) refers to its on-premise (Atlas is cloud-based), database and app bundle. It allows companies to purchase licensing for subscription-based usage (rather than paying for consumption under Atlas).

Most of MongoDB’s growth is based on acquiring new customers and migrating their workloads onto the platform, as well as consumption-based revenue from customers using its data products and growing workloads.

A material part of this outperformance was powered by non-Atlas, or on-premise, demand. Non-Atlas ARR rose by 8% Y/Y. This crushed -20% Y/Y growth estimates, as the company enjoyed much better than expected multi-year licensing revenue. They’re often aggressively pessimistic when it comes to forecasting for this segment, as revenue recognition is lumpy and unpredictable. There will continue to be plenty of on-premise opportunities for highly sensitive workloads, but the lion's share of the workload opportunity will be for its cloud-based Atlas offering.

And that’s why the sizable Atlas beat is even more encouraging than the non-Atlas result. Strong demand for the segment was driven by consistent consumption growth across the globe. This momentum was felt within new customer wins and existing workload expansions. There was nothing temporary or weird about this success… just general and structural strength.

Lastly, year-to-date new customer additions have grown by 65%. That is related to its decision to allow mid-market demand to access self-service channels instead of going through formal procurement processes. This has shrunk sales cycles, removed customer friction and accelerated growth. Great decision.

GPM pressure is related to Atlas growing as a percentage of the overall business. Atlas GPM specifically improved Y/Y, but it’s still a large drag and one that they’re happy to accept. Cloud-based demand has a far more compelling runway than on-premise.

Operating expense growth was held back by some delayed investments. Those investments will now be partially pushed to next quarter and FY 2027.

Cash flow was helped by income statement profit beats and some cash collection favorability. While that second item is not typically structural, they think improving cash flow conversion should be a durable theme as they can make working capital improvements sustainable.

The company has begun settling restricted stock unit (RSU)-related taxes via cash instead of new shares. That should help slow down the pace of dilution, along with its ongoing buyback program.

Guidance now includes 27% Y/Y revenue growth for Atlas vs. about 25% previously expected. It also includes 7%-9% non-Atlas growth, implying non-Atlas will do much better than its previous annual segment guidance calling for a roughly 5% Y/Y decline. As this is a consumption-based business model, they remain highly prudent in forecasting assumptions.

MDB also offered preliminary thoughts on FY 2027 results. Despite the aforementioned note on some investment delays, they still expect to expand the EBIT margin by 1-2 points Y/Y and generate a roughly 90% FCF conversion rate. They reiterated all targets outlined in the September Investor Day, which was awesome to hear amid the CEO change. Leverage next year is expected to come from strong revenue growth, as OpEx is expected to grow in line with revenue.

Finally, for FY 2027, they believe non-Atlas revenue won't be an especially large headwind or tailwind for the overall business. That was a pleasant surprise that led to 2027 revenue estimates rising by 4%. Their FY 2027 starts after next quarter.

MDB trades for 81x forward EPS. EPS is expected to grow by 32% this year (we’re now in their Q4 for FY 2026), 13% next year and 21% the year after. It also trades for 100x forward FCF. FCF is expected to more than triple this year to $377M. Growth is expected to be 19% next year and 36% the year after.

Introducing New CEO CJ Desai:

CJ Desai took over as MongoDB’s new CEO last month. His resume is fantastic. He has been the President and COO of ServiceNow, the President of Dell’s Emerging Technologies Division, a Director of Engineering at Oracle and most recently Cloudflare’s Product & Engineering President. Casual. Outgoing CEO Dev Ittycheria leaves MongoDB in a fantastic spot. CJ Desai was handed a monster beat-and-raise quarter, an ability to reiterate investor day targets and an operational engine that needed no overhauling. In the case of this company, “if it ain’t broke, don’t fix it” deeply applies. Desai can focus on subtle optimization and continuation of strong performance. No disruptive overhaul needed. As CEO, his priorities will therefore entail more of the same. These include: cultivating close customer relationships, scaling go-to-market, advancing the product roadmap for the AI era and “making MongoDB a foundational data platform” for that era. Again… more of the same in the best of ways. It’s rare the CEO transitions are as seamless as this has been, and the credit for that goes to Ittycheria.

Desai had opportunities to lead several other technology companies, but he chose MongoDB. He did so based on his highly upbeat view on MDB’s positioning. Their document-oriented database architecture is ideally suited for online transaction processing (OLTP) and helping customers modernize their data and app architecture. This foundation’s specialty is also unstructured data, which makes it a perfect partner for GenAI and agentic AI workloads. As Desai rightfully said, AI models, agents and apps need to solve “informational retrieval problems” to unlock value-enhancing use cases. These tools are only as good as the data they have access to. MongoDB provides scalable and organized storage of all they need, with vector search to actionably query from these data repositories and Voyage’s models to ensure this process is as precise and model hallucination-minimizing as possible. Customers don’t want to use different vendors to power their AI workloads vs. their modern general compute workloads if they don’t need to. And with MongoDB… they don’t need to.

“MongoDB is uniquely positioned at the center of the AI platform shift.” – New CEO CJ Desai

Core Business Health:

Even before AI tailwinds ramp, MongoDB’s core business is performing very well. As a reminder, this company will enjoy most of its AI success from the app layer of that opportunity. That layer is still nascent for the vast majority of companies, meaning AI is not carrying this financial engine today. These strong results could not have occurred without its main products being very healthy. And they view this non-AI runway as being very long. Most cloud data and app migrations still have not happened. You hear other cloud companies like AWS talk about it all the time. This migration journey is still in its early innings and MongoDB is a clear beneficiary. Desai is very high on the new customer pipeline as well as ample potential to expand existing contracts. 

AI Data & App Workload Opportunity:

The document-oriented foundation is an unstructured data specialist that is perfect for AI. It can massively scale as needed and field a mess of data types while rapidly making sense of it all. As this technological explosion vastly accelerates data processing, app creation and complexity for both of those things, Atlas will be a heavily relied-upon ally for many companies.

“AI is dramatically increasing app building speed and iterating. Fixed database schemas (relational databases) simply cannot keep pace… Our model is built for diverse, fast-changing, and interdependent data. Our integrated search, vector search, and voyage embeddings remove the need for brittle bolt-ons, and we are seeing industry-leading results.” – New CEO CJ Desai

Its Voyage AI decision is also looking like a very good one. This is greatly improving data pipelines, data usability and model response processes all in an effort to lower model hallucination rates. Enterprises rightfully won’t accept 20%-25% incorrect answer rates on models or AI applications. Can you imagine a customer service rep messing up 1 in 4 times? You’d fire them immediately, and that’s where models are today. Companies like MongoDB have a vested interest in accelerating AI adoption to nurture their data and app-based products. Voyage is a great way to accomplish this and give enterprises more confidence in this emerging technology. MDB can profit from direct sales of these models, which are turning into a great top-of-funnel for its other products. And? They can also enjoy the indirect benefit of more AI adoption supporting its entire business. 

While it’s early, there is traction. Mercor (AI native hiring company) is using Voyage AI and Vector Search to support 50% month-over-month client growth without crippling performance bottlenecks or significant incremental engineering talent needs. An established global media conglomerate displaced their Elasticsearch-based offering in favor of Vector Search and Voyage AI models. Within a few weeks, they were live, with 65% lower costs, 90% lower latency and 35% higher click-through rates via better content recommendations.

They’re quickly building an end-to-end data and app development stack for the AI era. Relational Migrator and MAAP make it easier to modernize products and set them up to embrace AI. Its core operational database scales as needed and ingests whatever unstructured data is needed. Voyage makes sure this process is done in a way to maximize context relevancy. Vector Search ensures all of this context is actionably organized and ready for querying. Not all of these products are solely for AI (which is a good thing), but all of them gear MDB up to greatly enjoy AI’s future tailwinds.

More Notes: