TCS rebuilds ABB networks around AI operations

Tata Consultancy Services has signed a multi-year agreement to transform the global network operations of ABB through a centrally managed network-as-a-service model supported by artificial intelligence.

Building on a technology relationship spanning more than two decades, TCS will take responsibility for end-to-end network operations, service integration, cybersecurity functions, and the modernisation of ABB’s local-area, wide-area, and software-defined wide-area networks.

A global network operations centre will monitor performance across the group’s distributed estate, while a common service layer will coordinate several technology suppliers. The programme forms part of ABB’s Future Network Model, which is intended to standardise infrastructure across regions, business units, production sites, and corporate systems.

AI-assisted monitoring will analyse network telemetry, correlate alarms, and automate selected operational tasks. Consolidating those functions should give engineers a clearer view of congestion, failing links, configuration drift, and service degradation that would otherwise remain divided between local tools and supplier platforms.

Across a diversified industrial group, network availability reaches far beyond office applications. Manufacturing systems, engineering data, remote service platforms, enterprise planning, connected products, cloud workloads, and supplier communications all depend on stable connectivity, often across sites containing several generations of equipment.

Standardisation can reduce the number of architectures and management tools that local teams have to maintain, while common telemetry allows recurring faults to be compared across locations. It can also simplify policy enforcement when factories, laboratories, offices, and service centres have developed different configurations over time.

Industrial sites cannot, however, be treated as interchangeable corporate branches. Production equipment may rely on fixed addressing, proprietary protocols, deterministic timing, or older operating systems, while maintenance windows are limited by process schedules and safety requirements.

A routine enterprise change can interrupt a production cell if it alters routing, multicast behaviour, name resolution, or firewall rules relied upon by control equipment. Central configuration templates therefore need plant-specific testing, staged deployment, and a clear division between global authority and local engineering control.

The same discipline applies to automation within the network operations centre. Correlation engines can process more events than a human team and identify patterns spread across devices or regions, but their output depends on accurate inventories, complete telemetry, consistent time synchronisation, and models that recognise normal production variation.

Automated remediation requires carefully bounded permissions. Re-routing traffic around a failed corporate link may carry limited operational risk, whereas changing a rule at an industrial firewall or isolating a network segment connected to production demands stronger approval, audit, and rollback controls.

Cybersecurity sits at the centre of the programme because consolidated management platforms hold credentials and control paths extending across the organisation. Administrative segmentation, privileged-access management, signed configuration changes, resilient authentication, and independent recovery mechanisms will be needed around the new operating model.

Closer integration between industrial data and shared digital infrastructure is already reshaping large manufacturing groups. ArcelorMittal’s work with AWS has similarly linked production data, analytics, and common cloud services across steelmaking operations, increasing the dependence of plant performance on well-governed network architecture.

As those connections expand, information-technology and operational-technology responsibilities increasingly overlap. Central analytics require access to plant information, yet production systems need predictable availability and protection from faults or attacks elsewhere in the enterprise.

Segmentation must therefore support controlled data exchange rather than create either a flat network or an isolated plant. Industrial demilitarised zones, application proxies, one-way transfer where appropriate, and tightly governed remote access can preserve the required flows without exposing controllers directly to wider corporate traffic.

Operational results will provide the clearest measure of the programme: mean time to detect and repair failures, successful change rates, application latency, site availability, security-event containment, and the number of manual interventions should show whether centralisation has reduced complexity rather than merely relocated it.

ABB’s scale gives the project an extensive test of AI-supported network operations across manufacturing and engineering environments. Common controls and global visibility can improve consistency, provided that site-level resilience, tested recovery, and disciplined separation between enterprise and production systems remain embedded in the design.