AI data centres challenge aging power grids

Technical hurdles impacting electrification

Today, AI has become a key technology that can be utilised across various fields, including consumer applications and industrial automation. The growth of AI has resulted in the increased construction of large data centres. These facilities, which can cover extensive areas and contain high-powered processors, need a continuous and significant supply of electricity. Compared to traditional data centres, AI-focused centres are designed to meet this high level of demand. 

Over the coming decade, this demand for power is expected to double, especially in areas that are highly populated and undergoing significant digital transformation. New and/or expanded grid connections by hyperscale data centres are on the rise and frequently exceed the capacity of existing local and regional grids. Combined with other electrification trends, including the increased number of electric vehicles (EVs), and the decarbonization of heating, for example, further impact this growing demand. 

Power grids in developed economies were constructed in the mid-20th century. The philosophy has been to deliver a one-way flow of electricity from large, centralised power stations to consumers and businesses. Now, these grids are expected to manage complex bi-directional flows, integrate intermittent renewable resources, and support new loads that far exceed the norms for which the infrastructure was originally designed. 

Alongside the physical aging of wires, transformers, and switchgear, all of which can lead to increased risk of failure and blackouts or brownouts under high loads there are other concerns. Some of the main vulnerabilities include limited transmission capacity. Here, the existing transmission lines, substations, and transformers cannot carry the sustained high loads that AI-driven data centres require. Further adding to the challenge, upgrading, or expanding such assets is capital-intensive and often takes several years to complete. 

As an example, Sweden’s main grid operator, Svenska kraftnät states in the report Grid development plan 2024 - 2033; “Over the next decade, we plan to renew almost 50 of our current substations and to construct approximately 30 new substations.” However, the complexities in developing the grid have grown substantially since the 1960s, as illustrated in the following diagram published in the Svenska kraftnät report. 

Implementing smart grids to support increased energy demand is not without challenges. Upgrading the electricity network to manage this added demand, operators can implement digital communication technology, sensors, and computer processing. These tools help to better manage the power supply and demand in real-time. Smart grids help to enable a two-way flow of both electricity and information between the utility and consumers, more efficiently when compared to traditional grids. Smart grids enhance existing power grids with features like automated fault recovery, real-time monitoring, integration of renewable energy, and improved consumer energy insights.

As Matti Sjögren, Nordic Liability Risk Management Specialist, highlighted in his 2019 article, “Smart power grids are taken into use, even while the hardware may be old. This leads to new, possibly unidentified risk exposures to societies and companies.” He highlights, that from an insurance perspective, the “digitalisation of earning methods and the increase of immaterial values in new business models aren’t less exposed to interruptions of the common infrastructures like communication or cloud services. “ 

This is especially concerning if the International Energy Agency (IEA) estimates materialise. According to the IEA report Energy and AI; “Data centres accounted for around 1.5% of the world’s electricity consumption in 2024, or 415 terawatt-hours (TWh). The United States accounted for the largest share of global data centre electricity consumption in 2024 (45%), followed by China (25%) and Europe (15%).” The report further estimates that data centre electricity consumption is set to more than double to around 945 TWh by 2030.  ¹⁾ 

To address these multi-layered challenges, utilities and grid operators are considering a range of solutions. Modernising the existing grid is vital, upgrading transmission and distribution infrastructure, including advanced conductors, smart transformers, and grid-scale battery storage, for example. Power companies can also construct redundancy into critical portions of the grid, including looped distribution networks and backup power sources, to prevent single points of failure. 

Data centres can also do their part in securing power supply to their operations by, for example, shifting non-essential loads to off-peak periods, and by participating in demand response programs, or deploy on-site generation and storage. 

Other opportunities lie in the establishment of microgrids and decentralised power generation, e.g. for campus-style data centres, which can operate independently during grid stress or outages. 

In Finland, according to Fingrid’s most recent Main grid development plan (2025), investments into electricity transmission links continue to grow. The report highlights that, “Finland’s main grid is currently being reinforced more than ever before, and at an accelerating pace: investments will reach EUR 1.7 billion in the period 2025 - 2028.” 

AI can also have a role in tackling the risk of blackouts and brownouts. This can be done for example by continuing to advance forecasting to better predict and manage both supply and demand at granular levels to further improve efficiency. 

The rising power needs of AI data centres are straining outdated grids, creating major infrastructure challenges. The expansion and modernisation of power grids require substantial upgrades to meet growing digital, and electrification demands, however the challenge goes beyond power companies and grid operators. Addressing these issues that stand in the way of an electrified future will require close cooperation among utilities, regulators, tech firms, and policymakers. 

_{Source: 1) Key Questions on Energy and AI – Analysis - IEA}