MongoDB enters the race to power AI

MongoDB announced a vector search capability for its database, bolting the feature on to an existing commercial product.

The vector database space—which were already a somewhat divisive one—is about to get a lot more complicated.

MongoDB announced this week that it was getting into vector search with a new product called Atlas Vector Search. It’s one of the first major database providers to explicitly enter the vector search space after a lot of hand-wringing about who would be the one to do it. And it’s also one of the first major commercial entities to bolt vector database functionality onto a core product as a feature, rather than have a purpose-built vector database like some emerging startups.

“There are some unique benefits of storing vectors alongside your operational data not to mention all the things you get from hosting it inside a battle-tested developer platform what you’ve been building for however long,” Benjamin Flast, MongoDB lead product manager for vector search, told me. “We see those natural synergies as one additional benefit to us running vector search. It’s also a fast evolving space.”

MongoDB’s entry was long-expected from people I talk to in the industry as its infrastructure is better suited for it—though some didn’t expect it to happen this quickly. The emergence of large language models sparked a frenzy around vector databases early this year around the time Pinecone and Weaviate raised large funding rounds.

But investors questioned around that same window in April this year how much time it would take for MongoDB to launch a product while those companies (as well as Chroma) were out raising. It turned out to be just a few months.