Realm
Summer 2011AcquiredBetter data structures = no DB.
Pressure-test this opportunity
Explore the risks and possibilities with a prompt for ChatGPT, Claude, or your agent.
Better data structures = no DB.
Explore the risks and possibilities with a prompt for ChatGPT, Claude, or your agent.
Realm rebuilt mobile persistence around application objects and offline operation. Founded in 2011 by former Nokia engineers Alexander Stigsen and Bjarne Christiansen, the Summer 2011 YC company spent years building a mobile-native database core before launching publicly in 2014.[1]
MongoDB agreed to acquire Realm for $39 million in cash in April 2019 and closed in May.[2][3] The product had two afterlives. Realm's local database remained open source, while MongoDB retired Atlas Device Sync and related App Services on September 30, 2025.[4] The acquisition was strategically coherent; the hosted service's later retirement does not mean all Realm technology failed.
Stigsen and Christiansen founded the company under the tight.db identity after working at Nokia.[1] They joined YC in 2011, then stayed in stealth for several years to build a new database engine for mobile devices.
“Building a database takes a long while,” Stigsen said in 2015.[5] The delay reflected the scope of the bet. Realm was not a wrapper around an existing service. It was an embedded database designed to fit inside applications, remain available offline, and expose application objects without the mapping and persistence boilerplate associated with SQLite or Core Data.
The public product launched in 2014. Its small footprint, local speed, and developer ergonomics helped it spread through mobile projects before the company added cloud synchronization.
The founding sequence matters. Realm first earned adoption as a local component that developers could use without a hosted dependency. The later Object Server and Synchronization Platform added a commercial cloud layer, but the local database retained an independent technical identity.
Realm embedded a sub-one-megabyte object database inside mobile applications. Data remained available locally and offline without calls to an external service.[5] Developers worked with application objects directly instead of translating between object models and relational tables.
This put Realm against SQLite and Apple's Core Data at the persistence layer. It competed on developer experience, performance, and offline behavior rather than a cloud API alone.[1]
Realm later added Object Server and a Synchronization Platform. The hosted layer moved changes between devices and cloud, supporting an offline-first application that could reconnect and converge.[6] That broadened the business from an embedded component into a cloud-to-device stack.
MongoDB integrated Realm with Atlas and Stitch, then renamed the SDKs as Atlas Device SDKs.[7] When hosted sync retired, local database code continued through open-source and community branches. Atlas-specific sync code required removal or migration. The distinction is essential: embedded storage continued; the managed synchronization control plane did not.
Realm targeted mobile developers who needed fast local persistence, offline operation, and less boilerplate. The sync platform later targeted teams building applications that needed cloud-to-device convergence.
No audited revenue, margin, retention, or customer-concentration figures were found. In March 2015, Realm said it ran on 100 million devices and named Groupon, BuzzFeed, Intuit, Zynga, Coinbase, and Expensify.[5] At acquisition, MongoDB reported more than 100,000 active developers and more than two billion downloads.[2] These are company-reported figures.
Realm competed with SQLite, Core Data, Firebase, Couchbase Lite, Parse, and custom persistence and synchronization stacks.[6] Its wedge was local object ergonomics plus offline behavior.
MongoDB offered the strategic adjacency of Atlas storage and a serverless control plane. Realm could bring mobile developers to MongoDB; MongoDB could give Realm a hosted backend and enterprise distribution.
Realm combined free or open local SDK adoption with paid synchronization, platform, and enterprise support. MongoDB sought to monetize the integrated workflow through Atlas.[8]
The primary acquisition announcement states $39 million cash, subject to adjustments.[2] TechCrunch reported that Realm had raised just over $40 million before the sale.[8] Invested capital does not establish investor or founder returns: preferred terms, ownership, adjustments, and retention arrangements were not observed.
No ARR, gross margin, paid usage mix, or customer concentration was found.
Realm's company-reported footprint grew from 100 million devices in 2015 to more than 100,000 active developers and two billion downloads by the 2019 acquisition announcement.[5][2]
MongoDB's acquisition and multi-year integration show strategic traction. The later EOL of hosted sync is a separate product decision. Continued local SDK branches show that adoption at the embedded layer outlived the managed service.
MongoDB called Realm and its own products “a natural fit.”[2] Realm supplied an embedded database, sync engine, and developer community. MongoDB supplied Atlas, serverless services, and distribution. The combination filled a missing device edge in MongoDB's platform.
That fit also coupled hosted sync to the economics and priorities of a larger cloud portfolio. MongoDB deprecated Atlas Device Sync in September 2024 and ended it, along with linked App Services capabilities, on September 30, 2025.[4]
The local database did not disappear. MongoDB had stated that Device SDKs would remain open source and usable with or without Atlas.[7] Community branches continue local functionality, while Atlas sync code was removed or requires migration.
The durable boundary was open client technology versus a proprietary hosted control plane. Local persistence could survive through source and community maintenance. Authentication, permissions, subscriptions, conflict rules, and offline queues depended on a service owner.
Realm might have sustained synchronization independently, charged differently, or found another owner. Its developer footprint and local adoption gave it a credible platform base.
Missing ARR, margins, usage mix, and MongoDB's internal retirement record prevent a confident verdict. The $39 million price versus reported invested capital also cannot establish a poor return. The evidence supports strategic integration followed by selective retirement, not a simple failure story.