Investing.com -- Google’s former lead of global data center strategy, Kathryn Ross, highlighted shifting investment trends, infrastructure challenges, and AI-driven demand in a Bernstein-hosted call on Feb. 13.

Ross estimated AI-driven data center demand could reach 210-315 GW by 2030, a sharp rise from the 12-20 GW hyperscalers had in early 2024. She noted that 80% of current AI infrastructure is used for training, but expects inferencing demand to dominate by 2030.

She underscored energy constraints in the U.S., with power load growth stagnating at 0-2% over two decades. Europe’s $200 billion AI investment plan and Malaysia’s $17 billion data center expansion reflect global shifts in infrastructure priorities.

Ross also discussed the potential for training data centers to be repurposed for inferencing, Microsoft’s AI data center expansion outside the U.S., and the industry's move toward smaller, modular data centers near urban centers.