AN artificial intelligence company boss has hit out at “misinformation” about the scale of the booming tech sector’s water use.
Michael Hunter, in charge of Apatura’s AI data centre projects in Scotland, told The National that new technology meant the energy-intensive computers only used as much water as an Olympic-sized swimming pool.
The firm is behind plans for 10 AI data centres across Scotland, though it has yet to apply for or receive planning permission for any.
AI data centres use water to cool down the supercomputers housed in the sites. Previously, Hunter said, this was done by a method known as evaporative cooling, which meant that the water could not be reused.
READ MORE: Douglas Alexander derides First Minister's Scotch whisky tariffs campaign
However, recent innovations mean that AI data centres can be filled up once with one megalitre (one million litres) of water, topped up by 2% of that volume annually, he said.
Hunter said: “There’s been a huge amount of misinformation.”
He said that while the evaporative cooling method – which pumps air cooled by water, which must be continually replenished, through the system – used vast quantities of water, the new method used an almost fixed volume that was better for the environment.
“We would be using what is known as a closed-loop system,” Hunter said. It works in “very much the same way as your car engine cools itself, or your fridge cools itself”, he added.
“The likes of our [planned] 250 megawatt facility in West Hermiston, that would be filling the system up as about the same amount of water as an Olympic-sized swimming pool,” said Hunter.
“Then you’d be topping up that water on an annual basis of about 2%, so there’d be about 2% losses.
“Fundamentally, the narrative has been completely wrong.”
Hunter also played down concerns over the amount of energy used by AI data centres, saying that Scotland’s energy production outpaced demand.
READ MORE: King Charles flogs off chance to catch endangered Scottish salmon at Balmoral
He said: “I don’t think that we can get away from the fact that these facilities do use large amounts of electricity.”
But Hunter added: “From a Scottish perspective, I think it’s really important to note that, when looking at electricity, we’re producing more electricity on an annual basis than we can, than the UK can use.
“I think last year we produced 130% of electricity that Scotland needed.”
Because the infrastructure is not in place to take the excess amount down south, that means Scots are currently paying for power generation to be switched off, he said.
“That’s the curtailment charges that you might be familiar with that are passed back to us as billpayers,” said Hunter. “That’s currently sitting at £1 billion, projected to rise to £3bn by 2030.”
He added: “From a Scottish perspective, electricity is not something that is in short supply.”
It has been argued that Scotland could be at the forefront of AI growth in the UK, with the country’s cooler climate better for hosting supercomputers, which must be closely guarded against overheating.
However there are significant concerns about the energy-intensive industry. We revealed earlier in the month how planned data centres in Edinburgh would require the same amount of electricity as would building five cities the same size as the capital within its boundaries.
And developers are not required to submit environmental impact assessments to local authorities which decide on awarding planning permission, which campaigners say means that companies are getting away with not revealing the true impact of projects.
Concerns are also mounting that the AI bull market may be a “bubble” – and one due to burst soon.
A recent study by the Massachusetts Institute of Technology (MIT) found that 95% of firms using generative AI pilots – capable of producing text and images – reported that the technology had no impact on their overall profits.
Investors were spooked by the news but the market has since rallied.
Hunter told The National that AI was only “going one way” and added: “To me it feels like we are still very much at the start of this AI journey.”
