AI Boom to Fuel 165% Surge in Data Center Power Demand by 2030: Goldman Sachs

The global demand for data center power is poised for a significant surge, driven by the increasing adoption of artificial intelligence, according to a report by Goldman Sachs. The investment firm predicts a 50% increase by 2027 and a staggering 165% increase by 2030, compared to 2023 levels.

"Recent Chinese developments, and particularly the AI model known as DeepSeek, have raised concern about the returns on current and projected AI investment," Goldman Sachs stated in a February 4th post. However, the firm also acknowledges uncertainties surrounding DeepSeek's training, infrastructure, and scalability.

The report highlights the key factors driving this projected demand. Large "hyperscale" cloud providers and corporations are building sophisticated AI models that require extensive data processing, placing immense power demands on data centers. Simultaneously, these same entities are investing heavily in constructing new, high-capacity data centers to accommodate these workloads.

Goldman Sachs Research forecasts a tightening of the supply-demand balance for data center infrastructure in the coming years. Occupancy rates are projected to rise from roughly 85% in 2023 to potentially exceed 95% in late 2026. This will likely be followed by a moderation starting in 2027 as new data centers come online and AI-driven demand growth slows.

The report also delves into the evolving landscape of data center types. While traditional corporate and telecom-owned data centers currently make up a significant portion of the market, the emergence of AI-dedicated facilities is reshaping the sector. These specialized centers, typically owned by hyperscalers or wholesale operators, are designed to handle the unique power requirements and high-density workloads associated with AI applications.

Regionally, Asia Pacific and North America currently dominate the data center market, particularly in areas like Northern Virginia, Beijing, Shanghai, and the San Francisco Bay Area. However, Goldman Sachs predicts that North America will witness the most new capacity coming online in the next five years.

Despite the projected growth, the report highlights potential constraints. "Data center supply – specifically the rate at which incremental supply is built – has been constrained over the past 18 months," notes James Schneider, a senior equity research analyst at Goldman Sachs. These constraints stem from permitting delays, supply chain bottlenecks, and the high costs and lengthy timelines associated with upgrading infrastructure.

The report concludes that the growing power demands of data centers will necessitate substantial investment in the electrical grid, estimating that approximately $720 billion in grid spending will be required through 2030. "These transmission projects can take several years to permit, and then several more to build, creating another potential bottleneck for data center growth if the regions are not proactive about this given the lead time," Schneider adds.