The worldwide expansion of digitalization has made data centers the beating heart of modern technological infrastructure. From cloud computing to data management, from social platforms and video streaming to Artificial Intelligence, everything passes through this vast IT ecosystem that hosts servers, data storage units, computers, and networking equipment. This dense concentration of devices in controlled environments makes it clear how closely the digital transition is tied to the energy transition: data centers consumed 415 TWh of global energy in 2024 (about 1.5% of total demand), a figure expected to double to 945 TWh by 2030 (3% of global consumption). Half of that increase will come from AI servers alone.

The energy impact of digital infrastructures must therefore be managed carefully. At the same time, automation and advanced data analytics systems can optimize consumption, reduce waste, and boost industrial and infrastructural efficiency. Digitalization could help cut emissions by up to 53% by 2050. This issue has grown even more urgent following the wars in Ukraine (2022) and the Persian Gulf (2026), which amplified the energy crisis, disrupted traditional supply sources, and pushed up consumer prices. From an infrastructure perspective, the high energy intensity required for AI training and the localized concentration of demand risk overloading power grids.

An Energy-Hungry Boom

By 2030, 80% of the increase in data center energy consumption will be concentrated in the United States, China, and the European Union. Data from the Digitalization & Decarbonization report by the Energy & Strategy Observatory of Politecnico di Milano outline the picture:

Electricity consumption in data centers, 2024 → 2030:

• USA: 233 TWh (4.4% of total) → +130% forecast by 2030
• China: 129 TWh (1.1% of total) → +170%
• EU: 49.7 TWh (2.3% of total) → +70%
• Italy: 5.8 TWh (1.9% of total) → 24.5–42.1 TWh by 2035 (7.4–12.7% of total)

In Italy, high-voltage connection requests for new data centers reached 69 GW in December 2025, nearly 13 times the level recorded in 2023. Installed capacity is expected to climb from an estimated 609 MW in 2025 to 2.3–4.6 GW by 2035.

Data center emissions currently exceed 1 million tons of CO₂ (2024). If the Italian energy mix remains unchanged, emissions will climb to 4.87–8.37 million tons by 2030. Even if the grid is 65% decarbonized (as per the National Energy and Climate Plan), emissions would still range between 2.9 and 5 million tons. This goal looks ever harder to achieve after renewable installations fell by 6% in 2025 — 7.2 GW of new capacity versus 7.6 GW in 2024 (Renewable Energy Report 2026).

Decarbonizing the Digital World

Digitalization itself remains a lever for decarbonizing business processes. Among companies listed in the MIB ESG Index, 350 recent digital initiatives helped achieve a 46% reduction in direct emissions (Scope 1). Yet total emissions still rose 9%, driven by a 13% increase in indirect emissions (Scope 3), which account for roughly 95% of the total — involving suppliers, logistics, and end users.

However, data center electricity demand will keep rising regardless of renewable growth, and AI’s contribution to sustainability won’t come automatically — in fact, emissions could increase. Integrating on-site renewable generation is key to reducing stress on the power grid, cleaning up the energy mix, and curbing CO₂ output, while improving sector resilience. The EU’s AI Continent Action Plan (incorporated into Italy’s AI Bill) envisions data centers that are more efficient, powered by clean energy, and capable of shrinking their carbon footprint through energy storage, self-consumption practices, and advanced cooling systems — the latter representing about 25–30% of a data center’s electricity use.

Cutting-edge cooling systems, such as direct liquid cooling, can lower energy consumption by over 90% compared to traditional air-based systems. Immersion cooling, rear-door heat exchangers, and AI-driven thermal management — which could dynamically optimize cooling through machine learning — offer additional efficiency gains.

Barriers to AI Expansion

Energy intensity, along with geographical and environmental impact, could slow the growth of AI. Energy efficiency will define both the trajectory and accessibility of AI innovation worldwide. To address the challenges of consumption and democratize development, researchers at the Observatory point to three key strategies: algorithmic and hardware innovation, energy transition, and localized policies.

Ideally, computational expansion should not be bound by energy constraints. Yet developers often face trade-offs — prioritizing energy efficiency over sheer computational performance — limiting algorithmic complexity and slowing breakthroughs. This dynamic risks concentrating innovation within a few well-funded organizations. Greater efficiency, in turn, could democratize access to AI tools, opening opportunities for startups and smaller players.

Significant investments in networks, generation capacity, and flexibility are needed. In Lombardy alone, high-voltage connection requests for new data centers have already hit 34 GW. “When networks in cities like Frankfurt, London, Amsterdam, and Dublin start to congest, cloud and server providers look elsewhere,” explains Federico Zucco, Project Manager at the Energy & Strategy Observatory. “In the United States, heavy investment in data centers came before network upgrades, causing congestion, higher local electricity prices, and mixed reactions. In Europe, Madrid and Milan stand out as areas with strong potential — networks that still have capacity to host new facilities.”

Environmental and Cost Pressures

The environmental impact of data centers varies depending on local energy matrices. Regions relying on renewables — such as Iceland, with its geothermal and hydroelectric supply — can operate low-carbon data centers and enjoy naturally cooler climates that cut cooling costs. By contrast, regions dependent on fossil fuels (notably parts of the U.S., China, and India) face higher carbon footprints and rising operational costs.

Finally, electricity prices play a major role. Spain, with 50–60% of its power coming from renewables, saw wholesale prices up to 45% lower than the European average in early 2026, outperforming countries like Italy. This has reduced gas dependence and boosted competitiveness.

In Italy, however, LNG imports from Qatar have proven “a fragile replacement for Russian gas,” warns Davide Chiaroni, Deputy Director of the Energy & Strategy Group. “Unless we revive renewable growth and make it the price-setting factor in the energy market — within a balanced mix that also includes nuclear power, which has longer deployment timelines — we’ll never build a truly resilient, sustainable energy system, either environmentally or economically.”

[Photo by Fré Sonneveld on Unsplash]