India loses nearly Rs 1.8 lakh crore every year because electricity disappears between power plants and homes. This amount is almost twice the Central Government’s annual health spending. In Uttar Pradesh alone, about 15% of electricity purchased by Uttar Pradesh Power Corporation Limited (UPPCL) is lost before it can be billed, resulting in losses of nearly Rs 10,000 crore every year.
Besides financial losses, these losses result in power cuts, voltage fluctuations, and mounting debt, making it harder for electricity distribution companies (DISCOMs) to improve service quality.
Uttar Pradesh is now taking a decisive step to address this challenge. UPPCL has approved a major pilot project in Varanasi district to build a comprehensive digital twin of a power distribution network using artificial intelligence. The pilot will cover all major substations across both urban and rural areas of the district.
Dr Ashish Goel, Chairman of UPPCL, says, “AI and ML can reshape DISCOM operations by transforming large-scale data into fast, actionable insights.”
Funded by the Ministry of Power, the project will be implemented by Pravāh, an AI startup founded by Mohak Mangal and Dhruv Suri, who met at Stanford University. Instead of deploying large survey teams on the ground for years, Pravāh uses street-view and satellite imagery, combined with computer vision, to map every pole, wire, transformer, and electrical connection down to individual homes and businesses. What traditionally takes years can now be completed in a matter of months.
Why Varanasi matters
Varanasi offers a unique and challenging environment for such a pilot. The district includes dense urban neighbourhoods, remote rural feeders, underground cables in commercial zones, and areas with a high concentration of rooftop solar installations.
Like many DISCOMs across India, UPPCL has seen its network grow more complex over time, faster than its mapping systems could keep up. Without a complete and accurate digital twin, it becomes difficult to know how much electricity each transformer is actually supplying or where infrastructure upgrades are most urgently needed, even for experienced engineering teams.
DISCOM executives who have worked for decades with evolving grid conditions see clear value in gaining this new level of visibility.
According to Jitendra Nalwaya, Director Technical at Purvanchal Vidyut Vitaran Nigam Limited, “Digital twin and physics-based model will provide visibility to DISCOM in terms of planning and maintenance, and improve overall network health and reliability.”
Artificial Intelligence at scale
Traditionally, creating detailed network maps for a utility as large as UPPCL could take five to six years and require hundreds of crores in investment. AI-driven mapping dramatically shortens timelines and reduces costs, making large-scale network digitization practical even for resource-constrained utilities.
Pravāh’s approach has already been demonstrated in Maharashtra, where it mapped poles, transformers, and electricity connections across Sangli district in just a few weeks, without deploying a single team on the ground.
The Varanasi pilot goes a step further. In addition to mapping physical assets, Pravāh will analyze how electricity flows through the network. This will help identify areas with high losses, voltage problems, and the impact of rooftop solar generation on grid stability.
From pilot to statewide transformation
If the Varanasi pilot is successful, UPPCL and the Ministry of Power plan to scale the approach across the state, positioning Uttar Pradesh as a leader in modernizing power distribution networks.
The initiative has already drawn national attention. Pravāh won the Ministry of Power’s Powerthon competition and was recognized by the Hon’ble Minister of Power, Shri Manohar Lal Khattar, in December. As part of this recognition, the company was awarded funding to carry out the Varanasi pilot. At a recent Ministry-led AI/ML conference, senior officials emphasized the urgency of scaling such technologies to help utilities manage increasing grid complexity nationwide.
For Uttar Pradesh, the timing is especially important. With significant government funding flowing into power sector modernization and smart meters being rolled out rapidly, this pilot offers a way to ensure those investments deliver maximum value. Accurate network maps can make smart meter data far more actionable and help utilities fully realize the benefits of RDSS programs.
The Varanasi pilot is expected to begin in the coming weeks, with results anticipated by mid-2026.
Disclaimer
Views expressed above are the author’s own.
END OF ARTICLE
