Amid the AI boom, compute power is emerging as an indispensable resource. By 2030, McKinsey expects data centers will need $6.7 trillion in global investments to meet the surging demand, say Senior Partner Mark Patel and coauthors. But for companies across the compute power value chain, making investment decisions is challenging and hinges on forecasting global demand for data center capacity. McKinsey research shows that the capacity needs for AI and non-AI workloads could almost triple by 2030, with AI capacity increasing 3.5 times and making up roughly 70 percent of the total.
Image description:
A stacked bar chart displays the estimated global data center capacity demand from 2025 to 2030, broken down by AI and non-AI workloads, under a “continued momentum” scenario. The chart shows a significant increase in total capacity demand over this period. In 2025, the total capacity demand is 82 gigawatts, with 44 gigawatts attributed to AI workloads and 38 gigawatts to non-AI workloads. This total increases to 219 gigawatts by 2030, with non-AI workloads reaching 64 gigawatts and AI workloads reaching 156 gigawatts. The chart also shows the incremental AI capacity added each year, ranging from 13 gigawatts in 2025 to 31 gigawatts in 2030, totaling 124 gigawatts from 2025 to 2030. A callout indicates a 3.5-fold change in total capacity demand between 2025 and 2030.
Note: This image description was completed with the assistance of Writer, a gen AI tool.
Source: Gartner reports; IDC reports; Nvidia capital markets reports; McKinsey Data Center Demand Model.
End of image description ends.
To read the article, see “The cost of compute: A $7 trillion race to scale data centers,” April 28, 2025.