Generative AI adoption contingent on green energy: Gartner

Gartner predicts that by 2027, 70 percent of businesses will cite sustainability and digital sovereignty as two pieces of top criteria when choosing public cloud generative AI services. 

In its forecast, Predicts 2024: Unravelling Tomorrow’s Cloud Computing Landscape, Gartner said a company’s generative AI services will need to consider nontechnical issues related to sustainability when looking at the public cloud. 

Demands from the likes of investors, customers, regulators and governments will lead organisations down a path of seeking out the best way to optimise IT carbon emissions. 

It predicts new processes, capabilities and tools will be introduced, designed around the management of energy consumption and carbon emissions for cloud-based generative AI workloads.

Kristin Moyer, distinguished VP analyst at Gartner told Digital Nation, “In the short-term, we are taking two steps backwards with AI and one step forward with sustainability (the reverse of how the saying usually goes)."

“This is because ICT electricity requirements are currently growing at roughly 25 percent per year, fuelled by the increasing demands of not just AI, but cloud, data, devices, chips and more,” Moyer said.

Moyer said in the mid to longer term, “We are going to take many more steps forward than backward.” 

“AI can advance environmental sustainability progress nearly beyond comprehension when used at the right time and in the right way. 

“For example, humans can sort 50 to 80 pieces of recyclable materials per minute. Optical sorters leveraging AI can sort 1000 pieces per minute,” Moyer said. 

“Technology will become radically more energy efficient with innovations like optical computing, post-silicon chips and DNA storage.”

Michael Warrilow, VP analyst at Gartner said, “The problem facing Australia is to find enough green energy to power local AI data centres at hyperscale. 

“Otherwise, AI workloads are likely to run offshore and we will face the twin challenges of digital sovereignty and network latency,” Warrilow said. 

Meanwhile, Sid Nag, vice president analyst at Gartner. “Because of its scale and shared services model, cloud technology is best-suited for the delivery of generative AI-enabled applications at scale and the development of general-purpose foundation models.”

“However, certain aspects must be addressed, including digital sovereignty, or the ability to control where data is stored and where operations are executed, and sustainability issues so that organisations can operationalise generative AI,” Nag said. 

Nag said specialty cloud providers will become an important consideration for many enterprise cloud architectures as organisations extend their cloud operations to cover diverse locations and use cases.

“Digital sovereignty will drive the need to include cloud providers that can meet the evolving and unique requirements of sovereign operations no matter the region they operate in,” said Nag.