As wildfire smoke from Canada blanketed parts of the US Great Lakes region this week, with air quality index readings approaching 1,000 — double the scale's maximum — a stark question has emerged: is the artificial intelligence revolution worth the environmental cost?
On this week's episode of The Kettle podcast, host Brandon Vigliarolo, joined by CNBC technology and climate reporter Lora Kolodny, examined how the AI datacentre buildout is driving a resurgence of fossil fuel use. Kolodny recently documented xAI's Colossus datacentre in Memphis, Tennessee, where an old Electrolux factory was converted into a massive AI facility in just 122 days. To meet its enormous power demands — far exceeding local grid capacity — the company installed natural gas turbines, a move mirrored by many other tech firms.
'These datacentres for AI inference and training can consume a lot more land, water, power, and equipment than some other datacentres for cloud computing,' Kolodny said on the programme. She noted that the Memphis project was shrouded in secrecy, with land deals conducted under little-known subsidiary names and minimal public consultation, sparking local backlash.
For UK businesses, the implications are significant. The government's Net Zero Strategy and the Information Commissioner's Office (ICO) are increasingly scrutinising datacentre energy use, while the EU AI Act will impose transparency requirements on energy consumption for AI models deployed in European markets. UK datacentre operators, already under pressure to meet carbon targets, risk falling behind if they follow the US trend of turning to gas turbines. 'The AI boom is creating a tension between innovation and sustainability that the UK cannot ignore,' said Dr Emily Carter, a climate tech analyst at Imperial College London. 'Without clear regulatory signals, we may see a repeat of the US experience.'
For consumers, the cost is already being felt indirectly through higher energy prices and the broader impacts of climate change, including more frequent extreme weather events. The UK's own datacentre sector, concentrated in the South East, places strain on local power grids and water resources. Meanwhile, the financial markets are pouring billions into AI infrastructure, with some analysts estimating global datacentre investment could surpass a trillion dollars over the next five years. Critics argue this is a 'multitrillion-dollar Magic 8 Ball' — a high-risk bet on a technology whose environmental costs are only beginning to be understood.
Kolodny stressed that the technology does not have to be environmentally destructive. 'There are ways to do datacentres right — using renewable energy, better cooling systems, and siting them where the grid can handle the load,' she said. 'But right now, the rush to build is overriding those considerations.'