The real gold in industrial AI lies in the models: Vinod Krishnan, ESSAR

Time of India

30-07-2025

In the age of digital disruption, where BFSI and retail race ahead with customer-facing innovations, industrial enterprises stand quietly in the background, seemingly slower, yet working on challenges that are infinitely more complex and consequential.In a candid conversation with ETCIO DeepTalks, Vinod Krishnan, Chief Digital and Information Officer at Essar, shared his hard-won lessons from decades of leading digital strategy, transformation, and innovation in India's industrial and infrastructure sectors. He offered a window into how industrial companies engineer resilience and growth, balancing ambition with pragmatism, and technology with business realities.
'In industry, you're solving a 10-year-old problem that will matter 5 years into the future. You need patience, perspective, and persistence, because the impact you create is profound and long-lasting,' said Vinod Krishnan, CDIO, Essar.
The untapped canvas of industrial digitization
'There's a book I recommend: Fusion Strategy. It says we've tapped only 25% of digitization's potential and the remaining 75% lies in industrial domains. That's our canvas,' said Krishnan.
Krishnan led with a fascinating observation: The real opportunity for digital transformation lies not in consumer sectors, but in the untapped potential of industries like mines, refineries, steel plants, and other operationally intensive environments.
Unlike retail and BFSI, industrial companies don't compete on instant gratification. Instead, their focus is on operational optimization, cost reduction per unit, and codifying decades of expertise into data-driven processes.
'Our business models have remained unchanged since the paleolithic era. Take mining, the process of extracting resources hasn't fundamentally changed. But how we optimize that process, that's where digital comes in,' he said.
Where the investments go
'If you ask me where the biggest shift has happened, it's in the value we place on data. Data comes first, everything else follows,' Krishnan said.
While cloud, AI, and cybersecurity are important, Krishnan was emphatic that data infrastructure forms the bedrock.
'Previously, expertise lived in the minds of people who'd been around for 30 years. But people retire, and knowledge walks out the door. Digitization is about codifying that knowledge, making efficiency systemic, not accidental,' he said.
Industrial organizations are channeling their budgets into:
Data acquisition and contextualization from OT and remote sitesEdge computing to bring computation closer to operationsInfrastructure to integrate local and remote data streams
'We're very ROI-driven. Every rupee must justify itself. But thanks to democratization of tools and lower costs, what used to seem out of reach is now feasible,' he said.
ROI lies in operational efficiency, not glamour
'In our world, the best ROI comes from areas with inefficiencies, because small changes ripple through the system,' he said.
For example, Krishnan recalled a transformation at a telecom infrastructure company where automated billing reduced manpower from 200 to just 20, in a complex, asset-heavy environment where each line item carried unique conditions.
Similarly, predictive maintenance and energy billing optimization have delivered tangible outcomes. 'We don't need a new payment gateway. We just need to shave off a few rupees here and there, and because we repeat these operations a million times, the value compounds massively,' he said.
Why edge trumps cloud
'We're an edge market. Mines and oilfields aren't in metropolitan hubs. OT data is local. Decision-making happens where the well is drilled, not at headquarters,' he said.
Krishnan explained why industrial enterprises rely heavily on edge computing:
Remote sites lack reliable, affordable connectivityOT systems use proprietary protocolsLocal autonomy is critical for uptime
'Cloud may evolve with better satellite connectivity, but today, edge is where the action is,' he said.
Why genAI & data monetization have limited appeal
Unlike B2C sectors, industrial companies don't see immediate, large-scale benefits from GenAI or data monetization. 'GenAI has use cases, like document analysis or dashboards, but it's not transformative here. Nice-to-have, not must-have,' Krishnan said.
Instead, the focus is on building reusable models, not selling data. 'Our competitive edge lies in the models we develop, not the data itself. Predictive maintenance, for instance, can be deployed in 50 places once perfected.'
Why does OT deserve more than just IT?
Given the unique vulnerabilities of OT environments, security investments lean heavily toward securing operational technology and the IT/OT interface. Krishnan said, 'Depending on the setup,
OT security
could account for 60%-80% of the security spend. IT is relatively straightforward, OT is where risks are greater.'
Talent and vendors
'Attracting top talent is hard. You can't keep a Nobel laureate on the bench for months. I call it the laser surgeon approach, pay top dollar, use them intensively, then move on,' he said.
Krishnan advocated for a pragmatic mix of in-house and external expertise.
Build a Center of Excellence (CoE) for model maintenanceOutsource model creation to specialists
'In industry, downtime can be catastrophic, even fatal. Resilience is about seamless load transfer and minimizing downtime, not self-healing gimmicks,' he said. Redundancy, hot standbys, and tested failover mechanisms take precedence over fancy 'chaos testing' or 'self-healing' narratives.
'Don't be discouraged when it feels like everyone else is moving faster. Industrial change is slower, but deeper. You're solving problems that have existed for decades, and your impact lasts longer,' he concluded. 'When a long-term industrial project finally delivers, the satisfaction is far greater than any quick, flashy win elsewhere.'
His advice to peers:
Focus on ROI-driven, operationally relevant use casesBe patient, industrial transformation is a long gestation processCodify expertise before it walks out the doorBuild a resilient edge infrastructure
Invest in OT security and pragmatic AI use cases