Over the past two years, we have witnessed an extraordinary acceleration in the adoption and deployment of Artificial Intelligence (AI) across nearly every sector of the economy. From finance and healthcare to manufacturing, education, and energy, AI technologies are being increasingly embedded into core operations. With each passing month, the list of AI applications expands, often replacing or augmenting functions that were once entirely human-driven. Whether in automating decision-making, optimizing logistics, or enhancing predictive maintenance, AI is proving to be not just a supportive tool, but a transformative force.
This rapid growth is expected to continue unabated, and its potential is indeed promising. Yet, as with every major technological shift, the expansion of AI comes with far-reaching implications—some of which are only just beginning to be understood. One of the most immediate and potentially disruptive side effects relates to energy demand. As AI systems proliferate, so too does the infrastructure needed to power them. Large-scale machine learning models and real-time processing capabilities require immense computing power, translating into a rising number of high-capacity data centres. These facilities, often operating continuously with demanding cooling and security requirements, are among the most energy-intensive installations globally.
Already, a typical hyperscale data centre has an installed capacity exceeding 5 MW, with several under development aiming for capacities in the hundreds of megawatts—and even gigawatt (GW) scale. As pointed out in the recent (May 20) Webinar “AI and Energy Transition” organised by IENE (https://www.iene.eu/en/congress/81/Webinar-IENE-AI-and-Energy-Transition), this is not a future concern; it is an emerging reality. And the sheer scale of planned development raises fundamental questions. How many such data centres will be built globally in the next five to ten years? What will their cumulative electricity consumption be? More critically, how will this trend affect regional power systems that are already under pressure from the twin challenges of decarbonisation and electrification?
Southeast Europe is no exception. Countries such as Greece, Bulgaria, Romania, and Serbia have already entered the race to attract data centre investment. These countries offer competitive electricity costs, strategic geographic locations, and increasingly favourable digital infrastructure. Licensing and construction for several major data centres are underway, with expected commissioning dates by 2026 or 2027. However, there is still little publicly available information on the cumulative installed capacity of these projects, or their projected energy draw. If even a modest number of high-capacity data centres come online within the same timeframe, the impact on national grids could be significant—perhaps even destabilising.
The critical question then becomes: how much will electricity demand rise due to the expansion of AI and its physical infrastructure in Southeast Europe? While precise forecasts remain elusive, preliminary studies suggest that global data centre electricity use could double by 2030. If Southeast Europe mirrors global trends, the increase in electricity demand could be substantial— especially in markets that are smaller and already facing grid challenges. These are countries still navigating the shift from fossil-fuel-based generation to intermittent renewables, with thermal systems being phased down without always having sufficient backup or grid flexibility in place.
A sharp and unpredictable increase in electricity demand from AI and data centres could expose systemic vulnerabilities. Most of the region’s electricity systems were not designed with such exponential and often geographically concentrated loads in mind. Without substantial upgrades in grid infrastructure, storage capacity, and demand-response mechanisms, there is a real risk of localised grid stress, reliability concerns, and delayed energy transitions. The rise in base load demand could also slow efforts to retire carbon-intensive plants, particularly if renewables are unable to match new load growth in terms of dispatchability and timing.
Thus, while the growth of AI promises economic and technological benefits, it also presents an emerging challenge for energy planners and policymakers. Southeast Europe must begin now to model, plan, regulate and factor in this demand surge. Coordinated cross-border strategies, investment in grid resilience, and clearer energy forecasting frameworks will be essential. Otherwise, the AI revolution could become yet another stressor on already fragile energy systems— potentially delaying the region’s progress toward cleaner, more secure, and more efficient electricity markets.
[from the Editorial of IENE’s “Energy Weekly Report” (No 454) of May 23, 2025]
