​Behind Vertical AI: What AI Is Already Demanding Of Energy And Utilities

Innovation​Behind Vertical AI: What AI Is Already Demanding Of Energy And UtilitiesByDeepak Garg,Forbes Councils Member.for Forbes Technology CouncilCOUNCIL POSTExpertise from Forbes Councils members, operated under license. Opinions expressed are those of the author. | Membership (fee-based)May 20, 2026, 08:45am EDTDeepak Garg is the chairman, co-CEO and founder at SEW.AI, a vertical AI-native platform for global Energy, water and gas providers. gettyOutside the utilities sector, AI is still widely viewed as a digital phenomenon: abstract, virtual and cloud-based. In my work with energy providers, the conversation is focused on something vastly different: AI's physical footprint. Due to the adoption of AI, data centers are expanding rapidly in size, scale and number, bringing with them rows of servers, dense compute infrastructure and unprecedented electricity demand. The land acquired, substations expanded and transformers stretched will have significant implications for utilities and grid reliability. This shift is forcing utilities to rethink how energy systems are planned, operated and experienced. The Scale Utilities Can No Longer Ignore AI-driven data centers represent an altogether different category of electricity demand for the energy and utility industry. ​​According to a 2024 joint report from the U.S. Department of Energy and Lawrence Berkeley National Laboratory, "data centers consumed about 4.4% of total U.S. electricity in 2023." This figure is projected to rise to between 6.7% and 12% by 2028. In less than a decade, data center electricity demand has already tripled, and it could double or even triple again within the next few years. ​​​AI’s future, therefore, is increasingly inseparable from energy’s future. ​​​​AI inference operates at a massive scale with always-on workloads and GPU-dense infrastructure. ​​​Unlike traditional industrial demand, AI workloads are continuous, highly concentrated and capable of scaling faster than gene...