The $600 Billion Surge: S&P Global Forecasts 40% Growth in Cloud Capex as AI Infrastructure Peaks

As of January 23, 2026, the global cloud economy has entered a second, even more intensive wave of capital deployment. According to a landmark report from S&P Global, cloud service providers—led by the "Big Five" of Microsoft, Amazon, Google, Meta, and Oracle—are set to increase their capital expenditures (capex) by nearly 40% in 2026, bringing the annual total to a staggering $600 billion.

This forecast follows a historic 2025, where these hyperscalers spent $437 billion (a 68% year-over-year increase). The continued aggression in spending highlights a fundamental shift: AI is no longer a "feature" of the cloud; it is now the primary engine of its growth and the largest single driver of infrastructure investment in the world.

1. The Hyperscaler Capex Explosion
S&P Global identifies an "insatiable" demand for AI infrastructure as the reason for this continued spending spree. The transition from AI training (building the models) to inferencing (running the models for users) is requiring a massive geographic expansion of data centers.

Financial Divergence: While revenue growth for the top hyperscalers remains robust—expected to stay above 20% throughout 2026—the sheer volume of capex is beginning to impact free operating cash flows, leading to a more complex credit risk environment for these tech giants.

Custom Silicon Shift: To mitigate the high costs of third-party chips, hyperscalers are using this budget to accelerate "self-sufficiency," developing in-house AI processors to reduce their long-term reliance on the current semiconductor supply chain.

2. AI’s Share of the Global IT Pie
The S&P report notes that AI-related expenditures now constitute almost one-quarter of all IT spending globally. This concentration of capital is reshaping how corporations prioritize their budgets.

Propelled IT Growth: Total global IT spend is projected to increase by 9% in 2026. Without the AI infrastructure buildout, that growth rate would be significantly lower.

The Software-to-Silicon Loop: Most of this spending is being funneled directly into hardware—specifically AI-optimized servers, networking switches, and high-bandwidth memory (HBM)—with 2026 supplies already largely sold out to the largest cloud providers.

3. The Enterprise ROI Paradox
While the "plumbers" of the AI world (cloud providers) are spending at record levels, S&P Global points to a growing tension in the enterprise sector.

The "Nascent" Gain: Although AI features are now embedded in most enterprise tools, actual revenue gains from these investments are described as "nascent." S&P notes that many enterprises are tightening non-AI IT budgets to fund their cloud migrations, waiting for a clear return on investment.

The 2030 Vision: Citing industry sentiment, S&P highlights that nearly 8 in 10 executives believe AI will not significantly contribute to their bottom-line revenue until 2030. Despite this, the "fear of missing out" (FOMO) ensures that investment remains high to avoid becoming a "laggard" in AI readiness.

4. Energy and Power: The New Gating Factor
Perhaps the most significant bottleneck identified for 2026 is not the chips, but the power required to run them. S&P Global Energy predicts that global data center power demand will increase 17% by the end of this year.

Grid Capacity: In the U.S. alone, spending on data centers is nearing $500 billion for 2026, but projects are increasingly being delayed by the "interconnection queue"—the time it takes to hook new facilities into the power grid.

Sustainability Strain: The explosive demand is testing the sustainability commitments of major tech firms. While companies like Microsoft and Google remain committed to net-zero goals, the 14% annual increase in power demand through 2030 is making those targets harder to reach without radical breakthroughs in clean energy storage.