Databricks buys $900M backed machine learning startup Tecton

Databricks is buying Tecton, a machine learning startup backed by Sequoia Capital and Kleiner Perkins, as part of its plan to build out full-scale AI tools for large companies.

The deal was confirmed Friday by Databricks CEO Ali Ghodsi, who spoke with Reuters. No price was shared, but the deal is being done using Databricks’ private shares. Tecton was last valued at $900 million back in 2022 and has around 90 employees today.

This is just the latest in a growing list of deals. Databricks is using its massive private valuation to snatch up companies that plug right into its broader AI strategy.

This specific acquisition is focused on real-time machine learning infrastructure. Tecton sells tools that allow companies to deploy and serve AI models at scale, with low latency, something Databricks wants to bake directly into its own platform.

Databricks targets real-time AI agents with Tecton deal

Tecton wasn’t just some outside partner. It already had ties with Databricks. In 2022, Tecton announced official partnerships with both Databricks and rival Snowflake, and since then, both of those companies have invested in Tecton.

Many of Tecton’s users, including Coinbase, already use Databricks services in their AI stack. Now Databricks wants to tighten that loop. Ghodsi said Tecton’s software and team will help scale up Agent Bricks, which is the company’s product for building and automating workflows with AI agents.

Speed is the goal. “It’s really the real-time building block to feed real-time information into the agents,” he said. He explained that things like voice-based AI systems require instant feedback. “Many of the use cases are directly user-facing and human-facing, and humans hate to wait.”

Tecton was founded in 2020 by engineers who used to work at Uber. These engineers helped build Michelangelo, the internal AI system Uber uses to run features like real-time pricing.

That experience turned into a commercial product focused on helping other companies serve AI systems that need fast, live data. Since launching, Tecton raised $160 million from backers like Andreessen Horowitz and Bain Capital Ventures.

But it didn’t go public or get acquired, until now. The company stayed private while building out a solid user base, with enterprise customers already putting its tools into real-world AI systems.

The new deal doesn’t say what happens to Tecton’s product line, but the clear signal is that its technology and engineers are now going to be absorbed into Databricks’ AI platform.

Databricks is making these acquisitions on the back of a growing valuation. Just this week, the company said it signed a term sheet for a fresh funding round at over $100 billion, a jump of more than 60% from eight months ago.

That gives Databricks the ability to issue more private shares and keep buying startups that strengthen its AI muscle.

This year alone, they’ve bought Neon, a serverless database startup, for $1 billion, and Tabular, which brings in the creators of the Apache Iceberg data format. Last year, they paid $1.3 billion for MosaicML, a generative AI platform.

The smartest crypto minds already read our newsletter. Want in? Join them.