By Dr. Avishek Kumar, Founder and Chairman of Sunkonnect
India has demonstrated that it can add renewable energy capacity at scale. The more difficult phase is now beginning: operating a power system in which generation is increasingly distributed, weather-dependent and digitally connected. Across renewable energy projects, manufacturing facilities, inspections and asset operations, the central problem is rarely a complete absence of data. It is that the data are fragmented, inconsistent and disconnected from the engineering and commercial decisions they are meant to support. Thereffore, the next challenge is therefore not capacity addition alone. It is converting a rapidly expanding collection of assets into an intelligent, reliable and responsive energy system.
India does not lack data; it lacks connected decisions
Solar plants, wind farms, substations, meters and factories already generate large volumes of information. Yet performance data may sit in a supervisory control and data acquisition (SCADA) system, maintenance history in a service provider’s software, inspection findings in reports and procurement records in spreadsheets. Each dataset may be useful on its own, but the engineering picture remains incomplete when they cannot be viewed together.
This fragmentation has a direct operational cost. A decline in energy yield may be visible before a failure occurs, but diagnosing it requires context: weather conditions, equipment history, alarms, cleaning records, component batches and earlier inspection observations. Similarly, a module defect found during inspection may have originated in material selection, process conditions or handling several stages earlier. If those records are not connected, teams investigate each incident as an isolated problem and repeat work that could have been avoided.
This is where aritificial intelligence (AI) and Internet of Things (IoT) can create real value. IoT makes dispersed assets observable; AI can identify weak signals and relationships that a conventional dashboard may miss. The International Energy Agency has estimated that digitalisation could reduce operations and maintenance costs across power plants and networks by around 5 per cent. Its more recent analysis indicates that widespread use of AI in power-plant operations and maintenance could generate annual savings of up to $110 billion globally by 2035. But these gains depend on reliable data, engineering context and a clear decision pathway – not on algorithms alone.
Project management: Earlier warnings, not more dashboards
Renewable projects have traditionally relied on manual site inspections, fragmented progress reports and decisions made after a delay has already become visible. Digitalisation often begins by replacing these reports with dashboards. That improves visibility, but visibility is not the same as intelligence.
The more valuable system connects engineering progress, procurement, logistics, contractor productivity, weather, quality and finance, and then highlights where intervention is required. A late shipment matters differently depending on the construction sequence; a productivity decline matters differently if weather or equipment availability is the cause. AI should help project teams distinguish a warning from noise, identify the likely consequence and decide what to do next. It should strengthen the project manager’s judgement, not attempt to replace it.
Digital twins can add another layer when they are linked to real operating data. Before construction, engineers can test layouts and operating scenarios. During operation, they can compare expected and actual performance, investigate losses and evaluate an intervention before applying it. The practical test is simple: does the digital twin help the team make a faster or better engineering decision? If not, it risks becoming an impressive demonstration with little operational value.
The same principle applies to drone inspections. Thermal and visual images can identify hotspots, soiling, vegetation encroachment or structural damage across large sites far faster than a purely manual survey. However, an image classification is not yet a maintenance decision. It becomes useful only when the finding is verified, prioritised by risk, connected to equipment history and assigned to a responsible team.
Forecasting must become an operating discipline
Forecasting is now central to India’s increasingly dynamic power system. Errors affect scheduling, procurement, dispatch and grid balancing; generators may also face deviation charges when actual output differs materially from schedule. AI can improve forecasts by combining satellite imagery, numerical weather predictions, historical generation, cloud movement and seasonal patterns, and by updating estimates as new data arrive.
But the objective is not merely a lower statistical error. A forecast creates value only when it changes an operating decision – when to charge a battery, commit reserve, schedule a flexible load or prepare for a ramp in renewable output. Machine-learning models can be retrained as observations and forecast errors accumulate, but they do not eliminate uncertainty, particularly during extreme or rapidly changing weather. Operators therefore need confidence ranges, clear escalation rules and human oversight, not a single number presented with false precision.
Smart grids: Turning visibility into coordination
Traditional grids were designed largely around centralised power stations sending electricity towards consumers. Rooftop solar, behind-the-meter batteries and electric vehicles (EVs) are changing that architecture. Consumers are becoming producers, loads are becoming flexible and power can increasingly flow in both directions.
Managing this system requires more than connecting millions of devices. Advanced metering infrastructure, automated substations, real-time monitoring and grid-management systems must work together so that utilities can see conditions, locate faults, manage voltage, reroute power and restore service faster. The essential capability is coordination: translating a distributed stream of data into reliable action across the network.
Smart metering illustrates both the scale of the opportunity and the implementation challenge. Under the Revamped Distribution Sector Scheme, approximately 203.3 million smart meters have been sanctioned, including about 197.9 million consumer meters. By early August 2026, around 57.3 million had been installed under the scheme, with about 72.4 million installed nationwide across all schemes.
These meters can improve energy accounting, load forecasting, billing and collection, while giving consumers better visibility into use. Yet installing the device is only the first step. Data quality, communications reliability, interoperability, consumer trust, privacy and cybersecurity will determine whether the metering programme becomes an intelligent operating layer or simply a very large hardware rollout.
Optimisation must extend beyond generation
Clean electricity delivers less value if it is lost in networks, stored inefficiently or consumed at the wrong time. AI-enabled energy-management systems can analyse consumption across factories, offices and campuses, but the goal should not be another report on energy use. The system should recommend an action: adjust a process schedule, change an operating set-point, shift a flexible load or investigate an abnormal consumption pattern.
IoT-enabled building-management systems can regulate heating, ventilation and air-conditioning, lighting, storage and EV charging in response to occupancy, tariffs and renewable availability. In industry, digital systems can shift flexible, energy-intensive processes towards periods of lower prices or greater renewable supply. The best results come when generation, storage and demand are designed as one integrated system and measured against outcomes such as cost saved, downtime avoided, yield improved or emissions reduced.
Storage is an asset; intelligent dispatch creates its value
Pumped storage projects (PSPs) and battery energy storage systems (BESS) projects provide flexibility by moving electricity across time. AI-based controls can coordinate charging and discharging using prices, weather forecasts, demand and grid conditions. But storage value does not come from capacity alone. It comes from deciding when to charge, when to discharge and which service – energy shifting, peak management, reserves or network support – creates the greatest system value at that moment.
India’s National Electricity Plan projects a need for about 73.9 GW/411.4 GWh of storage by 2031-32, comprising roughly 26.7 GW/175.2 GWh of PSPs and 47.2 GW/236.2 GWh of BESS. Building this physical capacity will be only part of the task. Market design, dispatch rules, forecasting, degradation-aware controls and interoperability will determine whether those assets deliver their expected technical and economic value.
To build for Indian operating conditions
India combines a rapidly expanding renewable sector, broad digital infrastructure and deep software and engineering talent. This creates an opportunity not merely to adopt global digital energy tools, but to develop systems suited to Indian conditions: diverse equipment fleets, varying communications quality, multiple languages, cost-sensitive operators and assets spread across very different climates and grid environments.
As other emerging markets confront the same integration and operating challenges, solutions proven at Indian scale can become globally relevant. India’s competitive advantage will not come from building the largest dashboard. It will come from combining field data, power-system knowledge, manufacturing insight and operational accountability – and turning them into repeatable decisions that improve performance.
Why digital-energy deployments still fail
Many deployments begin with the technology instead of the operating problem. Sensors are added without agreeing on data definitions; models are trained on incomplete history; systems do not interoperate; and outputs are delivered to teams that have neither the authority nor the workflow to act. Greater connectivity also expands the cybersecurity attack surface. A successful deployment should therefore begin with a measurable question: can we improve yield, reduce downtime, accelerate commissioning, lower inspection cost or improve grid reliability? It then needs trusted data, secure-by-design infrastructure, engineering validation, a responsible decision owner and continuous monitoring of whether the promised outcome was achieved.
Looking ahead
India’s next energy chapter will not be judged by GWs alone, but by how reliably and efficiently those GWs are planned, integrated and used.
AI can make forecasting and maintenance more anticipatory. IoT can make dispersed assets visible. Smart grids can coordinate distributed resources, and storage can provide flexibility across time. None of these technologies works in isolation, and none creates value simply because it has been deployed.
The real transformation occurs when trusted field data are connected to engineering context and operational responsibility. That is what turns an abnormal signal into preventive maintenance, a forecast into a dispatch decision and a meter reading into a more efficient grid. The leaders of India’s digital energy transition will not necessarily be those with the most sophisticated AI models. They will be those that connect data to decisions – and can demonstrate, in measurable terms, that those decisions make the energy system more reliable, affordable and productive.
