Why SCHNEIDER ELECTRIC is paying $3.1 billion to become a player in industrial AI

TL;DR

The announcement initially sparked a reaction of mistrust from the markets, by paying $3.1 billion to acquire Cognite, an industrial software publisher founded in 2017 and generating a little more than $170 million in annual turnover, Schneider Electric immediately revived questions about the valuations of artificial intelligence. The French group’s shares immediately fell on the stock market, while several analysts considered the multiple high, even if they recognized the strategic coherence of the operation.

A debate which, however, misses the point, because Schneider Electric is not buying an additional player in generative AI, but is seeking to control the technological layer which could become the most strategic in the industry over the next decade. Behind this acquisition lies a much deeper change. After building its dominance around electrification and automation, Schneider now intends to become a central player in industrial intelligence.

The industry has entered a new phase of its digital transformation. The first revolution consisted of connecting equipment. Programmable controllers, SCADA systems, IoT sensors and digital twins have gradually made factories observable. Every machine, every engine, every production line has become a source of data.

This stage is now largely completed, manufacturers no longer lack information, and are faced with the opposite problem with an abundance of data coming from heterogeneous systems incapable of naturally communicating with each other. Maintenance, engineering, supervision, asset management, energy consumption or planning data remains locked in technical silos built over several decades.

Artificial intelligence radically changes the nature of the problem. Models are no longer the main limiting factor, and their effectiveness now depends on their ability to understand the operational context in which they operate. The fact remains that an AI capable of generating text provides no value if it ignores the relationship between a pump, a heat exchanger, a safety valve or a maintenance history, and it is precisely on this layer that Cognite has built its technological advantage.

Contrary to the image often associated with AI companies, Cognite does not primarily develop models. His specialty is creating a coherent representation of industrial infrastructures through a unified data model and knowledge graph capable of automatically linking information from hundreds of different systems. This contextualization constitutes the true strategic asset of the company.

In a modern factory, a simple maintenance intervention can simultaneously mobilize asset management software, a supervision system, engineering plans, incident logs, real-time data from sensors and safety procedures. This information already exists. The problem is that they are generally not linked together.

By reconstructing these relationships, Cognite provides the foundation on which artificial intelligence can finally reason reliably.

Another key point, until now, manufacturers have mainly explored the uses of AI as a decision support tool. The models detected anomalies, and made recommendations or produced analyses.

The logic now changes in nature, now the objective is no longer just to identify a risk of breakdown, but to automatically trigger a maintenance order, it is a matter of adjusting the operating parameters in real time. AI gradually ceases to be an observer and becomes an actor in industrial operations.

The Atlas platform developed by Cognite illustrates this transition, its low-code environment makes it possible to build agents capable of interacting directly with industrial data, automating complex workflows and assisting operators in decision-making processes that have until now been largely manual.

This capability complements the CONNECT platform from AVEVA, the software subsidiary of Schneider Electric. And this is probably one of the most important lessons of this acquisition, if Cognite is officially bought by Schneider Electric, the operation is above all designed to strengthen AVEVA. Cognite will be integrated into the British subsidiary in order to enrich CONNECT, the industrial intelligence platform developed by the publisher.

The architecture gradually becomes readable: the electrical and automation equipment produces the data, AVEVA constitutes the software platform, CONNECT orchestrates this information in the cloud, Cognite provides its contextualization and its exploitation by AI, and Atlas provides the agents capable of automating operations.

Schneider assembles all the building blocks of a platform covering the entire industrial value chain, from infrastructure design to their daily operation.

This development is gradually bringing Schneider Electric closer to the major software platform publishers. Tomorrow, differentiation will no longer be based solely on the performance of a PLC or an electrical panel. It will depend on the ability to bring together equipment, software, data and artificial intelligence agents within the same environment.

Siemens, Honeywell, Rockwell Automation and Emerson are also pursuing a comparable strategy aimed at developing their own industrial intelligence platform.

Hyperscalers are also investing heavily in this strategic layer. Microsoft, Amazon Web Services and Google Cloud already offer environments capable of hosting digital twins, industrial data platforms and generative AI services. NVIDIA provides the computing infrastructures and specialized models that power this new generation of applications.

The boundary between manufacturers and digital players is gradually blurring, the former control the physical equipment and processes, while the latter control the computing power, models and cloud infrastructures. The battle will undoubtedly ultimately be played out on a third layer: that of contextualized data capable of powering autonomous agents. And it is precisely this position that Schneider is seeking to secure today.

The market reaction nevertheless reflects understandable caution. With a price representing nearly eighteen times Cognite’s annual revenue, the transaction appears demanding by traditional enterprise software standards.

Schneider Electric probably doesn’t value Cognite based on its current revenue. The group is buying eight years of development of a cloud-native industrial platform, a knowledge graph built for complex environments, hundreds of connectors to main industrial systems, technology already deployed with large international clients and more than 800 industrial data and artificial intelligence specialists. Reconstructing this set internally would probably have required several years of development, with no guarantee of obtaining an equivalent level of maturity.

Fundamentally, this acquisition reflects a much deeper shift in value creation in the industry. If for decades, equipment constituted the main differentiating factor, artificial intelligence opens a new perspective, where value is concentrated around data, its contextualization and its autonomous exploitation.

By paying $3.1 billion for Cognite, Schneider Electric is taking a position on what could become the nervous system of industrial infrastructures. In the factory of tomorrow, equipment will remain essential, but the competitive advantage will increasingly belong to those who know how to understand the data they produce, orchestrate the decisions they enable and automate the resulting actions. It is this ambition, more than the valuation of Cognite, which gives all its strategic significance to this operation.