Data Centre Power: Where will the Industry be in 2030?

In 2025, AI is prolific in the digital world. There are very few softwares operating without it, whether that be tucked away in the background or an AI-generated image on social media.

Data centres are no exception. According to the BBC, it is estimated that around £2.2tn (US$3tn) will be spent on AI-supportive data centres across the world within the next four years.

That said, while big tech companies were rushing to build AI-driven data centres, they are now working cautiously to build high quality centres with strong AI services. Despite this, according to Bain & Company’s latest 2030 global data centre forecast, the industry will struggle with power availability and construction delays.

AI and data centre power growth

With the amount of money being poured into AI, many are fearing an ‘AI bubble’ but Bain & Company (Bain) claims that AI will continue to grow rapidly. 

The report highlights that by 2030, the world will need twice as much data centre power as we do today, and it predicts demand will soar to 163GW. With AI driving the majority of the increase, the US data centre electricity demand could double to 409TWh.

Aaron Denman, Partner at Bain and Company and Americas Utilities & Renewables Practice Lead, said: "We expect there will be sufficient energy supply to meet demand. However, power access is now the critical gatekeeper of growth. Even as GPU and construction constraints ease, more flexible and independent sources of power will be needed. As such, the behind-the-meter (BTM) power generation has become the go-to source shifting timelines and decision-making."

Impact of US data centres

According to Bain, by 2030 US data centres could account for around 9% of the country’s total electricity, which is over double the amount of today's consumption and around 150TWh above the US Energy Information Administration’s basic forecast.

In order to meet the demand, close coordination between utilities, data centre operators and regulators is required. However, alongside building traditional transmission infrastructure and power generation, a unique course of action is required.

Initial steps include flexible demand programmes which shift consumption to off-peak periods, battery storage to maintain and manage load volatility and BTM such as natural gas, nuclear unit restarts and rooftop solar.

Extended term solutions are reliant on grid modernisation, renewable integration and transmission expansion.

The flexible BTM power generation sources can support smaller power sources like local generators or solar panels and enable the running of smaller data centres that handle inference workloads. By the end of the 2020s, the large majority of AI compute will be from inference workloads. The biggest and most powerful AI models will still require huge data centres, with each using at least 1GW of power.

Padraic Brick, Partner at Bain & Company says: "The general prediction that hyperscalers would scale back investments didn't happen in 2025. 

“However, we are seeing more deliberate investments by hyperscalers as they scale capacity focusing more on capital efficiency and getting more selective on locations for new deployments, particularly for AI."

Moving towards 2030

In the next four years, Bain predicts that North America will still account for around half of the largest concentration of data centre capacity, which is fueled by hyperscalers’ capital expenditures.

Simultaneously, government AI programmes and growing business use are leading the charge as more data centres are being constructed in places like Europe and Asia-Pacific as well. 

Bain states that companies are now looking for freedom to choose where their computers are so that they can manage speed, data rules and access clean energy.

"The AI data centre race is no longer just about scale. Winners are taking deliberate and careful approaches to capacity investments, while at the same time, actively securing fit-for-purpose power generation and mitigating build delays," said Peter Hanbury, Senior Partner; Technology and Americas Manufacturing Excellence leader at Bain & Company.

The physical build-out of data centres, in conjunction with power constraints, is another essential challenge with developers needing to navigate challenges like regulatory requirements and local legislative demands. 

The report highlights that proven actions can reduce construction timelines by up to a year.

Time-consuming permitting processes and equipment lead times ranging from eight to 24 months are slowing projects as well as labour shortages adding strain. The most significant obstacle is the electric utility connection with delays of up to five years.

According to Bain, the next four years will be critical for data centre power. Taking consistent and immediate action to meet the demand of AI-powered data centres is crucial for creating more power availability and improving construction efficiency.