For about fifteen years, digital transformation has been a major challenge for industrial companies — but it cannot succeed without genuine IT/OT convergence, connecting the information system to the shop floor. How to approach it, how to run the project, which standards to adopt? At Tech For Industry Show 2026 (Paris Expo Porte de Versailles), a panel brought together Patrick Lambolet (Schneider Electric, president of the OPC Foundation France hub), Rémy Verot (Bel Group), Laurent Mismacque (CEO, Siemens France) and Thierry Dano (Renault Group).

IT/OT convergence means making two long-separated worlds work together: operations (OT) and information systems (IT). As Siemens recalls, back then "the thing that knew everything about the factory was the machine, not the people around it." The automation engineer held all the line's data but stopped using it once production started; quality, production and other functions needed it but couldn't access it.

These worlds keep different logics: IT wants up-to-date, interchangeable systems with short life cycles; OT wants durability — "a machine is installed for 15 years." Hence the image, echoed by several panelists, of two "cousins" who both do programming but with opposite constraints.

Everything rests on data. But data alone isn't enough: it must, Schneider Electric stresses, be made reliable, contextualized (under what conditions, where was it captured?) and usable. Otherwise you accumulate unused data lakes. The goal of IT/OT convergence: OT makes data available and IT can actually use it — for maintenance, advanced control, quality or energy, always securely.

In automotive, one vehicle is produced per minute: every minute of downtime hurts profitability. Stuck around 2000-2010 with machines "full of data but locked away," Renault began in 2017 to collect its process data (torque values, setpoints…) and tie it to each manufactured vehicle. The benefit: on a customer incident, correct the process and recall only the strictly necessary vehicles. Today, 25,000 connected machines and de-siloed data usable by 100% of functions.

A food group with a heterogeneous installed base (4,000 to 6,000 controllers of all brands), Bel made a structural choice: deploying an Edge platform in each factory as a bridge between IT and OT. The same packaging machine is exposed in the same format regardless of brand — any functional project thus accesses an operational data model, without revamping the entire controller base. Two effects: creating a new role (tool manager in each factory) and a virtuous economic model — connectivity, recognized as technical debt and strategic investment, is funded separately and deducted from functional projects, which increases the ROI of each project (MES, energy…) and accelerates the transformation.

For both Siemens and Schneider, the leading standard is OPC UA: vendor-agnostic, natively secure, it carries not just the data but its associated meaning — its structure and contextualization. Created about thirty years ago (20 years for OPC UA), backed by a global association of over 1,000 members, it is now mature and widely used. Renault chose it for two reasons: the interoperable protocol, but above all the data model — a unified "enterprise dictionary" (finding the same "start cycle" on every machine, for a transverse, scalable cycle-time calculation).

MQTT has its place for volume and bandwidth reasons, particularly with edge computing — processing information close to the source to create added value before capitalizing on it. Siemens' guiding principle: never "collect data for the sake of data," but start from the use case, which determines the architecture.

Renault illustrates OPC UA's flexibility: a 6-axis robot exposes ~1,500 parameters, but only ~30 data points are actually collected depending on use cases. You expose everything and "open the pipe" when a new need arises. A telling example: a predictive maintenance project on robot gearboxes, closed for lack of results for three years, then relaunched six months later thanks to a PhD student — the "pipes" reopen on demand.

Bel Group leads, with GIMÉLEC and other manufacturers (within the food sector's strategic contract and ANIA), the Agro Digital Ready standard. The finding: Bel, Renault, Danone or Avril need the same data (machine status, fault codes, breakdowns, energy consumption) but worked in scattered order, each with its own specifications. The solution: a star rating (from 1 to 5), based on OPC UA, defining a common data-exchange nomenclature — simple and cheap at one star, up to a "technology festival" at five stars including European machine-connectivity regulation. The goal: not reinventing the wheel, and helping the sector's more fragile SMEs too. Around fifty members, standardization expected by year-end.

The panelists' consensus: it's not a technology problem — "all the building blocks are there" — but a matter of organization, governance and culture. Some concrete advice:

What is IT/OT convergence? It's the connection between the world of operations (OT — machines, controllers) and information systems (IT), so that shop-floor data is collected, secured and actually used by control systems.

What is OPC UA used for? OPC UA is a vendor-agnostic, secure interoperability standard that lets equipment and software from different vendors communicate — not only carrying data, but associating its structure and meaning (contextualization).

What's the difference between OPC UA and MQTT? OPC UA structures and contextualizes data for interoperability; MQTT is a lightweight protocol favored for volume and bandwidth, often used at the edge. The two are complementary.

Is IT/OT convergence only for large groups? No: according to the panelists, it's replicable in SMEs, which even benefit from greater agility. The key is starting from a method, one or two use cases, and clear data governance.

This panel was recorded at Tech For Industry Show 2026, the B2B trade show for industrial transformation (Smart Manufacturing, AI, Supply Chain) at Paris Expo Porte de Versailles. Watch the full discussion above, and join us on May 19–20, 2027, Hall 7, for the next edition.