Major tech companies are urging the government to restructure UK energy prices in order for the nation to benefit from data centre infrastructure and not miss out on the AI boom.

Amazon and OpenAI have commissioned a study from the Social Market Foundation (SMF), which found that the high cost of industrial electricity, planning restrictions, and long delays in grid connection are slamming the breaks on data centre development, potentially risking the UK’s position in the AI and digital innovations market.

The SMF detailed some of the issues surrounding the UK’s energy supply and data centre infrastructure, and urged for zonal pricing of energy resources.

This would mean that energy pricing would vary regionally, and depend on supply and demand in each zone – places where green energy is more readily available, and population pressures are at a minimual, would see cheaper energy prices.

According to Tortoise Media, the UK is ranked fourth for their overall AI capacity, falling to 17th worldwide for its AI infrastructure.

The UK is behind the US, China, and Singapore in terms of overall AI capacity, and its top competitors hold the highest rankings in terms of AI infrastructure. Data centres are vital to unlocking the power of AI and running their energy hungry systems.

While the study acknowledged the importance of the UK government’s AI Action Plan in helping the UK reach its AI ambitions, SMF also pointed out the bottlenecks in the UK energy supply which it says are incentivising data centres to operate outside the UK.

Analysis by the foundation found that the UK’s industrial energy prices typically cost four times more than what they are in the US. Further, the UK’s electricity supply is also limited compared to peer economies, and varies sharply by region, with slow grid connections often coinciding with regions where data centres are concentrated.

The SMF also found that European competitors can also supply greener electricity at lower prices compared to the UK as a whole.

Zonal pricing would attempt to mitigate some of these issues, and would drive data centre infrastructure development to more energy independent regions, like Scotland, which has a greater degree of renewable energy and a less dense population.

To compliment this, the SMF said that the locational pricing would need reforms in planning rules that would allow users and generators to change location on the basis of the price signals they face. Certain areas would be designated ‘AI Growth Zones’ in line with the AI Action Plan’s objectives by using Special Development Orders to bypass regular planning procedures, allowing for rapid development.

Recommended reading

• Microsoft Commits $80 Billion to Power AI Data Centres
• How Badly Will AI Data Centres Be Impacted By Power Shortages?
• UK Gov: Data Centres Are Now Critical National Infrastructure

The SMF also pushed for developing an abundance of clean energy, saying that the UK government should opt for a ‘fleet-based approach’ to building nuclear reactors while encouraging the development of small modular reactors.

“Data centres are the critical infrastructure that underpins the digital economy and emerging technologies like AI,” Sam Robinson, senior researcher at SMF, said.

“Without urgent action to address skyrocketing energy costs and planning delays, the UK risks losing its position as a global leader in tech innovation and adoption. A forward-thinking strategy that makes energy affordable accelerates planning, and boosts green energy investment is essential to securing our future as an AI superpower.”

Data centres are also guzzling electricity, and water, at alarming rates, research shows. The World Economic Forum estimated that AI’s requirement for power is just about doubling every 100 days. By 2023, data centres were already consuming about 4% of the US’s total electricity.

Expanding AI use has seen strains on electricity grids in the US as tech companies and governments scramble to find new sources of energy.

Further, the water needed to cool these data centres continually expands: asking ChatGPT to write an 100 word email uses 519 millilitres of water – just over your average bottle. Ten to 50 queries from the chatbot uses about two litres of water, analysis says.