E D I T O R ' S Q U E S T I O N

ata centre investors

At the same time , investing in grid resilience , battery storage and strategic partnerships with utility providers will help meet the growing AI-driven demand while maintaining reliability and scalability .

Ramprakash Ramamoorthy , Director of AI Research at ManageEngine , said : “ The AI boom is creating an unprecedented demand for energy in the computing sector . Massive GPU installations require vast amounts of electricity to run and – critically – to keep cool . This isn ’ t just a concern from a business point of view , but from an environmental one . Maintaining uptime requires a huge operational expenditure on energy , while innovative approaches will be required to ensure the new technology doesn ’ t derail efforts to reduce IT infrastructure ’ s carbon footprint .

“ Many data centre operators are making strides to optimise current systems with both of these needs in mind . They are investing in cuttingedge cooling technology and GPU hardware that reduces energy consumption , both of which can have a significant impact on overall usage . Larger providers have also been investing in renewable energy sources such as wind , solar and geothermal , powering their data centres through means other than fossil fuels . Another possible approach is to build data centres in colder locations , where the atmospheric chill can be used to help control the centre ’ s temperature .

“ However , the sheer scale of AI energy usage means that these efforts on their own are unlikely to meet the demand . As a result , many hyperscale data centre operators are turning to highercapacity innovations – particularly hydrogen fuel cells and nuclear power generation . Those might sound like outlandish solutions to this particular problem , especially as hydrogen continues to be a niche power source despite its widespread availability .

“ However , executed correctly , both offer high-capacity , cleaner alternatives to oil or natural gas , making them worthy of serious attention . Nuclear power ’ s risk profile and political sensitivity do pose challenges , though , which may ultimately lead to slower adoption of this method across the industry as a whole .

“ The other key area of development is in the efficiency of AI models themselves . The quicker and smarter their operation becomes , theoretically the lower their power consumption will be for any given task . AI developers and coders therefore also have a role to play in controlling energy usage – not to mention the potential for AI models themselves to analyse and identify possible solutions to the problem , given the right training and time .

“ Ultimately , maintaining stability and uptime for AI installations while balancing the need to reduce harmful emissions is a real challenge . Achieving it will require collaboration across the IT and energy industries , and with governments . AI use cannot come at the cost of increased global heating or deforestation – alternatives to fossil fuels must be prioritised , along with R & D on energy use reduction .”

www . intelligentdatacentres . com 31