Question marks’ over whether AI workloads will be deployed in UK
The announcement of the AI Growth Zones in the AI Opportunities Action Plan show a commitment by government to use data centres to benefit areas outside London and the Southeast, but there is doubt over whether the largest AI workloads will be deployed in Britain at scale, a legal expert has said.
Kirsty Harrower, a partner on the real estate team at Norton Rose Fulbright, who has been advising customers and operators in relation to data centres since 2010, said the government was “aiming for the regions to be brought into the data centre conversation”. But difficulty getting connected to the grid, copyright regulations and the cost of power meant there were “question marks over whether AI workloads are going to be deployed in the UK at scale”.
Because AI workloads can be location agnostic, they provide an opportunity to develop the economy in relatively remote areas, she noted. But Harrower said that the regulatory environment in Britain, the high cost of power and the grid connections queue were potentially off-putting to AI data centre developers. “To my mind, the government has to decide whether to regulate, deregulate, or attract. It seems to me that we have more power availability challenges and regulatory hurdles versus the US.” She said data centre projects in the UK usually tended to be of the order of tens of megawatts compared to hundreds in the United States as a result.
One of the issues in the UK is copyright law, which could benefit from being harmonised with EU regulations if the government wants to make AI use cases more attractive, Harrower said. Energy prices in the UK also made it a difficult market for some. Harrower said she hoped theNESO connections reform process would make it easier for data centres to access the electricity network. “We’ll have to wait and see whether they deliver on their promise of increasing speed to market.”
More widespread use of Simplified Planning Zones by local authorities could also help spur data centre development. The planning regime in the UK remains a real hurdle for data centre developments. “I have developer clients who are finding the process painful.” There is not a specific data centre use class available currently, Harrower pointed out. Developers with sui generis planning consent might also feel it is necessary to get a certificate of lawful use to derisk the process, slowing down progress and increasing cost.
There had been a focus on London and Slough for a long time when it came to data centre development, Harrower noted. Because cloud infrastructure needs to be developed close to the population centres it serves, outside London, cities such as Birmingham, Manchester and Leeds could be good locations to develop cloud data centres.
If the above challenges could be overcome, more remote locations such as Scotland would be suitable for AI workloads.
Want more data centre news? Data Centre Demand has got you covered.
Kirsty Harrower, a partner on the real estate team at Norton Rose Fulbright, who has been advising customers and operators in relation to data centres since 2010, said the government was “aiming for the regions to be brought into the data centre conversation”. But difficulty getting connected to the grid, copyright regulations and the cost of power meant there were “question marks over whether AI workloads are going to be deployed in the UK at scale”.
Because AI workloads can be location agnostic, they provide an opportunity to develop the economy in relatively remote areas, she noted. But Harrower said that the regulatory environment in Britain, the high cost of power and the grid connections queue were potentially off-putting to AI data centre developers. “To my mind, the government has to decide whether to regulate, deregulate, or attract. It seems to me that we have more power availability challenges and regulatory hurdles versus the US.” She said data centre projects in the UK usually tended to be of the order of tens of megawatts compared to hundreds in the United States as a result.
One of the issues in the UK is copyright law, which could benefit from being harmonised with EU regulations if the government wants to make AI use cases more attractive, Harrower said. Energy prices in the UK also made it a difficult market for some. Harrower said she hoped theNESO connections reform process would make it easier for data centres to access the electricity network. “We’ll have to wait and see whether they deliver on their promise of increasing speed to market.”
More widespread use of Simplified Planning Zones by local authorities could also help spur data centre development. The planning regime in the UK remains a real hurdle for data centre developments. “I have developer clients who are finding the process painful.” There is not a specific data centre use class available currently, Harrower pointed out. Developers with sui generis planning consent might also feel it is necessary to get a certificate of lawful use to derisk the process, slowing down progress and increasing cost.
There had been a focus on London and Slough for a long time when it came to data centre development, Harrower noted. Because cloud infrastructure needs to be developed close to the population centres it serves, outside London, cities such as Birmingham, Manchester and Leeds could be good locations to develop cloud data centres.
If the above challenges could be overcome, more remote locations such as Scotland would be suitable for AI workloads.
Want more data centre news? Data Centre Demand has got you covered.
