MongoDB (MDB)
Runs a developer data platform across managed cloud and self-managed document databases.
Something off on this page? Send us feedback.
MongoDB sells the place an application keeps its information. The same software comes two ways: a service MongoDB runs for customers on rented cloud machines, and a package customers install on their own. The cloud version brings in most of the money. What is changing is that the run-it-yourself side has started growing again — and both are now sold as the memory AI applications search.
Item facts: FY2026 · year ended January 31, 2026, from filings, earnings calls and company pages.
Judgment weights, not filed revenue
The band summarizes business focus and direction. ~ marks estimates.
6 in detail · 13 more below
MongoDB Atlas
The cloud database MongoDB runs for customers, $1.81B last year and 73.4% of revenue. Existing customers spent about 22% more than a year earlier. Watch whether usage growth holds near the ~27% pace management expects this year.
Competes with Amazon DocumentDB (Amazon) · Azure Cosmos DB for MongoDB (Microsoft) · Amazon Aurora PostgreSQL (Amazon)
In plain English
Every app — a bank's, a newspaper's, a delivery service's — needs somewhere to keep its information: users, orders, articles. Atlas is that somewhere, except MongoDB runs it, on machines rented from Amazon, Microsoft and Google.
Customers pay the way they pay for electricity: by how much their app actually draws. A quiet app costs little; one that takes off costs more without anyone signing anything new. So growth comes as much from existing apps getting busier as from new customers arriving — about 69,300 of them by the end of July.
Atlas Search and Atlas Vector Search
Word search and meaning search running on the same data, with no second system to keep in step. Among Atlas customers spending $100K a year or more, 48% now use two or more platform features, up from 42% — management credits search for most of that.
Competes with Elasticsearch / Elastic Cloud (Elastic) · Pinecone vector database (Pinecone) · pgvector on PostgreSQL (PostgreSQL)
In plain English
Two ways of finding things, built into the database itself. One matches words the way a search box does. The other matches meaning, so a question about cancelling a booking can surface a document that only ever says refund policy.
Before this existed, customers copied their data into a second, separate search product and kept the two in step. Doing it in one place removes that chore. On the cloud side there is no bill of its own — the work simply shows up as more usage, and a larger monthly total. Customers running the software themselves have paid extra for it since June.
Voyage AI
The models behind meaning-based search, bought for $220M in 2025 and folded into the platform rather than sold on their own. Rivals charge pennies for the same thing, so the value here is the traffic it pulls onto Atlas.
Competes with text-embedding-3 (OpenAI) · Embed v4 (Cohere) · Google embedding models (Google)
In plain English
Bought for $220 million in early 2025: a small team whose models turn documents into long strings of numbers, so a computer can tell which two texts mean roughly the same thing. That translation is what makes meaning-based search work at all.
There is no separate bill for it — it comes with the platform. The payoff is meant to be arrivals rather than fees: the number of customers using Voyage roughly doubled in each of the last two quarters, from a small base, and most of the newcomers were AI startups with no prior MongoDB relationship.
Multi-year enterprise commitments
Long commitments from large enterprises, covering both the cloud and run-it-yourself sides. The signed pile is roughly double a year ago, and about half of it turns into revenue within twelve months. No single customer reached a tenth of last year's revenue.
Competes with Database folded into a cloud-wide deal (Amazon) · Database folded into an Azure deal (Microsoft) · Multi-year Oracle Database licences (Oracle)
In plain English
Big organisations rarely buy a database the way you buy an app. They negotiate once, commit to several years of spending, and sign.
That signature is not money yet — MongoDB counts it only as the software and support are actually delivered, so the promises stack up unearned. The stack stood at $1.52 billion at the end of July, and the two largest contracts in the company's history were both signed in the quarter that ended in January: about $90 million of Atlas with a large technology company, and more than $100 million of the self-run package with a large financial institution. Neither buyer was named.
MongoDB Enterprise Advanced
The install-it-yourself version, counted together with a few smaller subscriptions at $578.1M last year. Guided to grow around 11% this year — its first double-digit year in three. Watch whether the new paid search add-on keeps it there.
Competes with Oracle Database (Oracle) · SQL Server (Microsoft) · Db2 (IBM)
In plain English
Some customers will not put their data on somebody else's computers — banks, telecom operators, public agencies, often because a rule says they may not. For them MongoDB ships the same database as a package to install in their own building, with support staffed around the clock.
They pay a subscription, usually locked in for several years. Because a slice of each multi-year deal counts as revenue the moment it is signed, the quarterly figures jump around: in the latest quarter this line grew about 36%, while the steadier measure of what customers pay per year grew about 11%.
Professional Services
Consulting and training that gets customers running, $77.8M last year and right around 3% of revenue for four straight years. Its job is to start subscriptions rather than to earn a margin of its own.
Competes with Database modernization practice (Accenture) · Oracle Consulting (Oracle) · AWS Professional Services (Amazon)
In plain English
Somebody has to do the moving. When a company shifts a decades-old system onto MongoDB, the plan, the tricky parts and the retraining all have to happen — and MongoDB sells that help by the hour, plus courses.
It is kept deliberately small, right around 3% of revenue every quarter for four years, and run to get subscriptions started rather than to make money itself. Outside consultancies deliver most large migrations. The one recent expansion was buying Clarity Business Solutions in May 2026: roughly $10 million of annual services revenue, at about breakeven, and staff cleared for classified U.S. government work.
MongoDB AtlasThe cloud database MongoDB runs for customers, $1.81B last year and 73.4% of revenue. Existing customers spent about 22% more than a year earlier. Watch whether usage growth holds near the ~27% pace management expects this year.The cloud database MongoDB runs for customers, $1.81B last year and 73.4% of revenue. Existing customers spent about 22% more than a year earlier. Watch whether usage growth holds near the ~27% pace management expects this year.
In plain English
Every app — a bank's, a newspaper's, a delivery service's — needs somewhere to keep its information: users, orders, articles. Atlas is that somewhere, except MongoDB runs it, on machines rented from Amazon, Microsoft and Google.
Customers pay the way they pay for electricity: by how much their app actually draws. A quiet app costs little; one that takes off costs more without anyone signing anything new. So growth comes as much from existing apps getting busier as from new customers arriving — about 69,300 of them by the end of July.
Competes with Amazon DocumentDB (Amazon) · Azure Cosmos DB for MongoDB (Microsoft) · Amazon Aurora PostgreSQL (Amazon)
Atlas Search and Atlas Vector SearchWord search and meaning search running on the same data, with no second system to keep in step. Among Atlas customers spending $100K a year or more, 48% now use two or more platform features, up from 42% — management credits search for most of that.Word search and meaning search running on the same data, with no second system to keep in step. Among Atlas customers spending $100K a year or more, 48% now use two or more platform features, up from 42% — management credits search for most of that.
In plain English
Two ways of finding things, built into the database itself. One matches words the way a search box does. The other matches meaning, so a question about cancelling a booking can surface a document that only ever says refund policy.
Before this existed, customers copied their data into a second, separate search product and kept the two in step. Doing it in one place removes that chore. On the cloud side there is no bill of its own — the work simply shows up as more usage, and a larger monthly total. Customers running the software themselves have paid extra for it since June.
Competes with Elasticsearch / Elastic Cloud (Elastic) · Pinecone vector database (Pinecone) · pgvector on PostgreSQL (PostgreSQL)
Voyage AIThe models behind meaning-based search, bought for $220M in 2025 and folded into the platform rather than sold on their own. Rivals charge pennies for the same thing, so the value here is the traffic it pulls onto Atlas.The models behind meaning-based search, bought for $220M in 2025 and folded into the platform rather than sold on their own. Rivals charge pennies for the same thing, so the value here is the traffic it pulls onto Atlas.
In plain English
Bought for $220 million in early 2025: a small team whose models turn documents into long strings of numbers, so a computer can tell which two texts mean roughly the same thing. That translation is what makes meaning-based search work at all.
There is no separate bill for it — it comes with the platform. The payoff is meant to be arrivals rather than fees: the number of customers using Voyage roughly doubled in each of the last two quarters, from a small base, and most of the newcomers were AI startups with no prior MongoDB relationship.
Competes with text-embedding-3 (OpenAI) · Embed v4 (Cohere) · Google embedding models (Google)
Multi-year enterprise commitmentsLong commitments from large enterprises, covering both the cloud and run-it-yourself sides. The signed pile is roughly double a year ago, and about half of it turns into revenue within twelve months. No single customer reached a tenth of last year's revenue.Long commitments from large enterprises, covering both the cloud and run-it-yourself sides. The signed pile is roughly double a year ago, and about half of it turns into revenue within twelve months. No single customer reached a tenth of last year's revenue.
In plain English
Big organisations rarely buy a database the way you buy an app. They negotiate once, commit to several years of spending, and sign.
That signature is not money yet — MongoDB counts it only as the software and support are actually delivered, so the promises stack up unearned. The stack stood at $1.52 billion at the end of July, and the two largest contracts in the company's history were both signed in the quarter that ended in January: about $90 million of Atlas with a large technology company, and more than $100 million of the self-run package with a large financial institution. Neither buyer was named.
Competes with Database folded into a cloud-wide deal (Amazon) · Database folded into an Azure deal (Microsoft) · Multi-year Oracle Database licences (Oracle)
MongoDB Enterprise AdvancedThe install-it-yourself version, counted together with a few smaller subscriptions at $578.1M last year. Guided to grow around 11% this year — its first double-digit year in three. Watch whether the new paid search add-on keeps it there.The install-it-yourself version, counted together with a few smaller subscriptions at $578.1M last year. Guided to grow around 11% this year — its first double-digit year in three. Watch whether the new paid search add-on keeps it there.
In plain English
Some customers will not put their data on somebody else's computers — banks, telecom operators, public agencies, often because a rule says they may not. For them MongoDB ships the same database as a package to install in their own building, with support staffed around the clock.
They pay a subscription, usually locked in for several years. Because a slice of each multi-year deal counts as revenue the moment it is signed, the quarterly figures jump around: in the latest quarter this line grew about 36%, while the steadier measure of what customers pay per year grew about 11%.
Competes with Oracle Database (Oracle) · SQL Server (Microsoft) · Db2 (IBM)
Professional ServicesConsulting and training that gets customers running, $77.8M last year and right around 3% of revenue for four straight years. Its job is to start subscriptions rather than to earn a margin of its own.Consulting and training that gets customers running, $77.8M last year and right around 3% of revenue for four straight years. Its job is to start subscriptions rather than to earn a margin of its own.
In plain English
Somebody has to do the moving. When a company shifts a decades-old system onto MongoDB, the plan, the tricky parts and the retraining all have to happen — and MongoDB sells that help by the hour, plus courses.
It is kept deliberately small, right around 3% of revenue every quarter for four years, and run to get subscriptions started rather than to make money itself. Outside consultancies deliver most large migrations. The one recent expansion was buying Clarity Business Solutions in May 2026: roughly $10 million of annual services revenue, at about breakeven, and staff cleared for classified U.S. government work.
Competes with Database modernization practice (Accenture) · Oracle Consulting (Oracle) · AWS Professional Services (Amazon)
Named in filings, launches and programs
- MongoDB AMPPlatformLaunched September 2025: AI tools plus MongoDB engineers who rewrite old table-based applications. Bendigo Bank cut core-banking migration development time by 90%.
- MongoDB Community EditionProductThe free download. It earns nothing directly; it is how developers meet the database before anyone at their company pays for anything.
- MongoDB Atlas for GovernmentProductThe U.S. public-sector version, carrying the government's mid-level cloud clearance; higher clearances for sensitive and defense workloads are being pursued.
- Clarity Business SolutionsBrandBought May 2026 — a partner since 2021 bringing about $10M of yearly services revenue and staff cleared for classified U.S. government systems.
- MongoDB Atlas Managed MCP ServerProductLaunched August 2026: a hosted connection that lets coding assistants such as Claude Code and Devin read live data sitting in Atlas.
- Atlas Stream ProcessingProductHandles information arriving in a continuous flow rather than sitting still; gained meaning-based search in August 2026.
- voyage-context-4, voyage-code-4 and Native RerankingProductModel releases from June and August 2026 that sharpen which documents an AI application pulls back when it answers a question.
- MongoDB Relational MigratorProductA tool that converts a database built from rows and tables into MongoDB's document shape.
- MongoDB CompassProductThe point-and-click window onto the database, for people who would rather not type commands.
- Ops Manager, Cloud Manager and the Kubernetes OperatorProductThe control panel customers use to run the database themselves; part of the Enterprise Advanced package.
- Queryable EncryptionProductLets customers search data that stays scrambled the whole time; shipped in the self-run package from version 8.0.
- Time Series collectionsProductBuilt-in handling for readings stamped with a time, such as sensor or market data.
- MongoDB Checkpointer for LangGraph and the Claude Code pluginEcosystemHooks announced in May 2026 that let AI agent frameworks keep their working memory in MongoDB.
MongoDB AMPPlatform
Launched September 2025: AI tools plus MongoDB engineers who rewrite old table-based applications. Bendigo Bank cut core-banking migration development time by 90%.
MongoDB Community EditionProduct
The free download. It earns nothing directly; it is how developers meet the database before anyone at their company pays for anything.
MongoDB Atlas for GovernmentProduct
The U.S. public-sector version, carrying the government's mid-level cloud clearance; higher clearances for sensitive and defense workloads are being pursued.
Clarity Business SolutionsBrand
Bought May 2026 — a partner since 2021 bringing about $10M of yearly services revenue and staff cleared for classified U.S. government systems.
MongoDB Atlas Managed MCP ServerProduct
Launched August 2026: a hosted connection that lets coding assistants such as Claude Code and Devin read live data sitting in Atlas.
Atlas Stream ProcessingProduct
Handles information arriving in a continuous flow rather than sitting still; gained meaning-based search in August 2026.
voyage-context-4, voyage-code-4 and Native RerankingProduct
Model releases from June and August 2026 that sharpen which documents an AI application pulls back when it answers a question.
MongoDB Relational MigratorProduct
A tool that converts a database built from rows and tables into MongoDB's document shape.
MongoDB CompassProduct
The point-and-click window onto the database, for people who would rather not type commands.
Ops Manager, Cloud Manager and the Kubernetes OperatorProduct
The control panel customers use to run the database themselves; part of the Enterprise Advanced package.
Queryable EncryptionProduct
Lets customers search data that stays scrambled the whole time; shipped in the self-run package from version 8.0.
Time Series collectionsProduct
Built-in handling for readings stamped with a time, such as sensor or market data.
MongoDB Checkpointer for LangGraph and the Claude Code pluginEcosystem
Hooks announced in May 2026 that let AI agent frameworks keep their working memory in MongoDB.