SKN | Goldman Sachs Raises Global Data Center Capacity Forecast to 273 GW by 2030 as AI Infrastructure Spending Accelerates

Key Takeaways:

• Goldman Sachs raised its global data center capacity forecast to 273 gigawatts by 2030, up from its previous estimate of 217 GW, reflecting stronger artificial intelligence infrastructure demand.
• The projected expansion could require approximately $8.5 trillion in cumulative capital expenditure over five years, based on an estimated $50 billion per gigawatt, including computing infrastructure.
• Power availability, permitting delays, community opposition and higher financing costs could constrain construction, potentially favoring projects with secured electricity supplies and regulatory approvals.

Goldman Sachs has significantly increased its outlook for global data center development, reflecting the accelerating infrastructure requirements of artificial intelligence and cloud computing. The bank now expects worldwide capacity to reach 273 gigawatts by 2030, compared with its previous forecast of 217 GW. The revision highlights the scale of investment required to support AI adoption while drawing attention to the practical constraints that could determine which projects ultimately proceed.

Goldman Sachs Sees a Larger Global Infrastructure Expansion

In a November 2 research note, Goldman Sachs increased its global data center capacity projection by 56 GW. The revised forecast implies approximately 170 GW of additional capacity from 2025, when worldwide supply stood at around 103 GW.

The projected expansion could require approximately $8.5 trillion in cumulative capital expenditure over five years, using an estimated cost of $50 billion per gigawatt, including computing infrastructure. Annual global capacity additions are expected to reach 25 GW in 2026, 32 GW in 2027 and 35 GW in 2028. Goldman also increased its longer-term estimates to account for projects that have not yet been formally announced.

The scale of the forecast illustrates how AI infrastructure is becoming a major capital-intensive development cycle, extending beyond semiconductor production into electricity, buildings, networking and computing systems.

United States Remains the Largest Growth Market

Goldman Sachs raised its estimate for US data center capacity in 2030 to 152 GW from 125 GW. This represents approximately 103 GW of additional capacity after 2025, with the United States expected to account for roughly 60% to 70% of worldwide additions through the end of the decade.

Expanding cloud and neocloud infrastructure, alongside facilities supporting graphics processing unit demand, contributed to the upward revision. These categories added 11.8 GW to the bank’s cumulative US capacity forecasts for 2026 through 2030.

The research also points to the growing financial commitments of major technology and cloud companies. Consensus projections for selected US technology and cloud businesses indicate that annual capital expenditure could reach approximately $1.74 trillion by 2030, compared with $457 billion in 2025.

Large AI Facilities Drive Construction Requirements

High-performance computing facilities designed for AI workloads are expected to represent a substantial portion of new development. Projects with capacity of at least 75 megawatts could account for 66% of global additions and 75% of US additions by 2028.

This concentration in larger facilities reflects the infrastructure demands of advanced computing workloads. However, the scale of these developments also increases their dependence on reliable electricity, substantial financing and timely regulatory approvals.

Power and Financing Constraints Could Reshape Project Economics

Goldman Sachs identified several obstacles that could affect the pace of construction, including limited electricity availability, permitting delays, community opposition and rising financing costs. These challenges are particularly important because high-performance computing developments typically rely on debt financing for approximately 80% to 90% of project costs.

Higher borrowing expenses could therefore place pressure on project economics, especially when combined with construction delays or uncertainty over electricity connections. Developments with secured power supplies and clear regulatory approvals are considered better positioned to advance as financing conditions tighten and scrutiny of new facilities increases.

Over the coming years, the difference between announced capacity and operational infrastructure will become increasingly important. Goldman Sachs’ revised forecast signals substantial long-term demand, but the realization of that growth will depend on whether developers can secure electricity, financing and permits at the pace required. For investors assessing the AI infrastructure cycle, execution capacity and access to essential resources may prove as important as demand itself.

For a confidential discussion regarding your cross-border banking structure and global wealth management arrangements, contact our senior advisory team.