In a significant shift within the datacenter landscape, hyperscale operators are projected to claim 61% of global datacenter capacity by 2030. This growth is fueled by the robust demand for cloud services and the increasing need for computational power to support AI developments, according to Synergy Research Group.

Despite continued investment in GPU servers for AI processing, on-premises infrastructure is seeing reduced growth relative to this hyperscale boom. In 2018, on-premises capacities accounted for 56% of the total, a figure expected to drop to 22% by the decade’s end as hyperscale capacities soar.

Currently, hyperscale entities like AWS, Microsoft, and Google represent 44% of worldwide capacity. Notably, this includes a rise in self-owned facilities from these companies, transitioning away from leased spaces.

By 2030, hyperscale datacenter infrastructure is forecasted to triple, although exact data from Synergy remains undisclosed. Conversely, both on-prem and colocation setups are predicted to become less prominent as their market shares decline.

John Dinsdale, Synergy Research’s chief analyst, emphasizes that the push for cloud infrastructure and AI capabilities is central to this expansion, with regional variances in hyperscale prevalence noted, particularly stronger in the US.

Dinsdale projects all regions will witness substantial growth, with hyperscalers’ portion increasing by at least 20% annually across the board. Yet, this rapid growth raises concerns over potential market saturation, with industry leaders wary of repeating past market bubbles.