Schneider Electric agreed on June 30, 2026 to acquire Cognite, the Norwegian industrial data-and-AI software maker, for $3.1 billion in cash — one of the year's largest industrial-AI deals and a roughly 18-times multiple on Cognite's 2025 revenue.

The deal

Schneider is buying 100% of Cognite Holding B.V. from Norway's Aker ASA and other investors, with completion expected in the coming quarters subject to regulatory approvals. Founded in 2017, Cognite employs more than 800 people across the Americas, Europe and Asia-Pacific and reported 2025 revenue above $170 million, with annual recurring revenue bookings growing 36%.

What Cognite built

Cognite's core product, Data Fusion, imposes a unified data model and knowledge graph on the sprawl of industrial data — sensor streams, engineering diagrams, maintenance logs — so it can be queried in context. On top sits Atlas AI, an agentic layer that can reason over that graph and take autonomous action across industrial processes. That combination, contextualized data plus agents, is precisely what asset-heavy industries have found hardest to build themselves.

Why Schneider wants it

Cognite will be consolidated into AVEVA, Schneider's wholly owned industrial-software business, with Data Fusion and Atlas AI absorbed into AVEVA's CONNECT platform — aiming at end-to-end coverage of how industrial assets are designed, run and improved. "Cognite has built something rare, a truly industrial grade AI platform," said Schneider CEO Olivier Blum, adding the ambition to "give systems the ability to think, adapt, and act."

The read

Industry analysts framed the deal as a data-layer land grab: ARC Advisory Group called it a reinforcement of "the data foundation for industrial AI," arguing that whoever owns the contextualized data model owns the platform industrial agents will run on. For Aker, it is a landmark software exit for Norway; for the market, an 18x revenue multiple on a 36%-growth asset is a confident price that signals automation giants would rather buy the industrial-AI data layer than spend years building it.