ReportGem ReportGem 中文

REAL-TIME GLOBAL RESEARCH

Booming Datacenter pipeline and AI spending boom - the pivotal role of energy: POWERING UP EUROPE

Published: 2026-06-17Institution: Goldman SachsPages: 15Original language: EnglishEvidence page: 2

Research evidence excerpt

Booming Datacenter pipeline and AI spending boom - the pivotal role of energy: POWERING UP EUROPE

Goldman Sachs Powering Up Europe

Executive summary

Booming Datacenter pipeline and AI spending boom - the pivotal role of energy

Europe’s datacenter pipeline – measured as the number of connection requests to

power grids across Europe – has surged to nearly 500 GW, equivalent to 1.5x current

European power demand, up sharply from our most recent projection of c.290 GW six

months ago and just 170 GW in January 2025. While some of this pipeline may be

speculative, we believe its rapid growth and its scale signal a major datacenter

construction phase ahead, spread fairly evenly across the continent. The EUDCA expects

+20 GW of additions by 2031, nearly tripling existing capacity, with around 60% already

under construction or FID. This would drive +1.5% power consumption growth pa from

2028-29, on our estimates. We continue to believe Europe will reach 65-80 GW of IT

capacity by 2035, with datacenters driving 1.5-2% power consumption pa later in the

decade, depending on AI Agents and AGI adoption. Meanwhile, Big Tech AI spending

keeps climbing, with the five main hyperscalers implying c.$700 bn of capex in 2026,

rising to nearly $1 trillion in 2027. Next year’s AI infrastructure spending would be

around ten times larger than in 2023, underscoring that the datacenter boom continues

in full force. We believe utilities will be pivotal to this process by providing energy,

scouting for land and grid connections, and reinforcing existing power grids to support

the incremental consumption.

Exhibit 1: The EU DC pipeline has more than doubled in one year

EU 28 GSe datacenter pipeline evolution, 2025-26 (GW)

c.480 GW

2025 EU-28 Consumption: c.320 GW

c.290 GW

c.280 GW

c.230 GW

c.170 GW

The English excerpt is extracted automatically from the cited source page and may contain layout or recognition errors. It is never batch translated.

Open report viewer