81% of energy professionals expect an AI-driven surge in demand: data centers are inflating requests, and network costs...

81% of energy professionals expect an AI-driven surge in demand: data centers are inflating requests, and network costs are being shifted to the public

81% of leaders of energy and related companies believe that artificial intelligence will require more electricity than current forecasts anticipate. At the same time, 32% expect a significant overshoot, while only 3% expect lower demand. However, this is not a consumption calculation but a survey by Kekst CNC commissioned for the ADIPEC energy conference, which will take place 2–5 November in Abu Dhabi, reports The National.

The study covered 1832 executives: 1528 representatives from energy, technology, AI, finance, and industry, and another 304 from the oil and gas sector. Additionally, 17 interviews were conducted. The survey ran from August to October 2026, covering, according to the publication, 78 countries. There is no independent calculation of additional electric capacity in the published results.

71% of respondents view AI as an opportunity rather than a threat. Another 43% expect an increase in the influence of tech companies on global energy. Among oil and gas executives, 81% consider AI adoption necessary to remain profitable, 27% expect to increase efficiency, and 25% — to profit from supplying power to data centers. At the same time, a quarter admit the industry is not ready for the new demand.

The survey also reflects the region’s investment interests: 85% consider the energy sector more attractive than five years ago, 93% support developing multiple energy sources, 72% expect an enhanced role for the Middle East, and 74% — its importance in building energy infrastructure. The study was released ahead of ADIPEC, organized by dmg events with participation from ADNOC.

Actual volumes are estimated by the International Energy Agency. According to the IEA study, data centers consumed 485 TWh in 2025, and by 2030 demand could reach 950 TWh — about 3% of global electricity consumption. Consumption by specialized AI centers is expected to triple. In 2025 demand from all data centers rose by 17%, and those focused on AI by 50%.

By the IEA’s energy balance, electricity for data centers was supplied by coal — 30%, renewables — 27%, gas — 26%, and nuclear — 15%. Even with growth in green generation, computing infrastructure remains dependent on gas and coal.

But part of the declared demand may turn out to be fictitious. According to Capgemini’s research, 77% of energy companies have difficulty forecasting load, 67% face speculative requests from data centers, and roughly 19% of requests may never turn into real facilities. The share of AI in data center consumption could rise from 25% to 60% over three to five years.

Ordinary consumers may end up paying for new networks. According to a National Grid Partners survey, nearly 80% of energy companies have implemented AI for connection planning, 87% complain about outdated tariff regulation, and 83% point to the shifting of data centers’ network costs onto households.

81% of energy professionals expect forecasts to be exceeded, the IEA projects 950 TWh by 2030, and Capgemini warns of 19% potentially empty requests. If networks are built for capacities that never materialize, multibillion-dollar costs could remain in ordinary consumers’ tariffs.

Incredible, but true...