Capex in Context: AI Investment Approaches Banking-Scale Flows

Capital expenditures by major U.S. tech firms on artificial intelligence infrastructure are forecasted to rival net-new bank lending in 2026, highlighting a tectonic shift in economic power and capital allocation. AI's concentrated investment surge signals systemic transformations with wide-reaching consequences, from supply chains to public markets.

Artificial intelligence is reshaping more than just technology; it’s now redrawing financial landscapes. In 2026, projected AI-related capital expenditures, dominated by a handful of U.S.-based tech giants, are set to surge to approximately $646 billion. To put this into perspective, that figure rivals the total net-new bank lending for 2025, which stood at roughly $700 billion, according to the FDIC. Where bank lending channels credit across millions of households and small businesses, AI spending represents high-stakes bets by a narrow group of firms investing in data centers, semiconductor development, networking infrastructure, and energy resources.

“The scale of hyperscaler capex is expected to be approximately $646 billion, or about 2% of U.S. GDP,” said Torsten Slok, chief economist at Apollo Global Management, during a recent analysis. “This is no longer just a sector story—it’s a macroeconomic force.”

Bank lending typically reflects distributed economic activity, supporting everything from home purchases to small business operations. By contrast, AI expenditures funnel concentrated corporate capital into fixed assets that enable compute power. Amazon, Alphabet, Microsoft, and Meta account for the lion’s share of this spending, with combined capital commitments as high as $665 billion for 2026, based on company filings and analyst forecasts. Alone, Amazon’s AI-driven capex for Amazon Web Services (AWS) is expected to top $200 billion, with Alphabet and Microsoft allocating $185 billion and $80 billion respectively.

The sheer size of these investments surpasses key financial benchmarks that contextualize its magnitude. For instance, AI capex is poised to exceed U.S. corporate income tax receipts for fiscal year 2025, which the Congressional Budget Office confirmed at approximately $453 billion. It also dwarfs U.S. customs revenues by over threefold and exceeds the combined defense budgets of major G7 nations, excluding the United States.

But it’s not just the scale—it’s the concentration. Unlike government programs or household and small-business lending facilitated by banks, the decision-making power behind this unprecedented capital wave rests with a few high-tech titans. “This is a shift in macroeconomic power,” Slok noted, underscoring its potential ripple effects.

These ripple effects are already evident in global supply chains, particularly in energy demand and semiconductor manufacturing. Data centers alone accounted for 4% of total U.S. electricity use in 2024, according to Pew Research, and are projected to double their energy consumption by 2030. To meet AI infrastructure’s requirements, grid expansion will be crucial, with an estimated 100 gigawatts of additional power generation necessary worldwide, as detailed by J.P. Morgan Asset Management.

Meanwhile, the semiconductor supply chain remains strained, exacerbated by limited capacity for advanced GPU production and reliance on rare earth metals. "Of the 125 gigawatts of data centers globally, only about 20 gigawatts can currently handle AI workloads," noted Stephanie Aliaga of J.P. Morgan. "That gap is extraordinarily wide and requires both higher spending and faster innovation."

The financial markets are also absorbing these changes. Technology-sector bond issuance surpassed $200 billion in 2025 and could contribute another $300 billion in 2026, according to the Dallas Federal Reserve. Analysts warn that such large-scale borrowing could drive interest rates higher and strain corporate credit markets. Vanguard economist Shaan Raithatha has flagged the “hidden risks” tied to this debt-driven capex cycle, which could ripple through equity valuation models that depend on extended earnings projections.

While advocates of these AI investments tout their transformative potential for industries ranging from healthcare to logistics, skeptics wonder whether the revenue and productivity gains will measurably align with this level of capital deployment. Harvard economist Jason Furman pointed out that, while AI-related investments constituted 39% of GDP growth in the first nine months of 2025, GDP excluding those contributions grew at just 0.1% during the same period. Such figures echo the "Solow Paradox" of the 1980s, when the impact of information technology on productivity remained elusive despite substantial investment.

One significant concern is whether AI’s benefits will diffuse widely or remain in the hands of a small number of stakeholders. Bridgewater Associates highlighted that while the investment cycle boosts GDP in the short term, it risks crowding out labor-intensive industries by increasing capital costs. A surge in AI infrastructure builds may also amplify burdens on supply chains and energy markets without delivering proportionate job growth. “The labor required to build or operate data centers is quite small relative to the capex spend,” Bridgewater's research noted in an analysis earlier this year, pointing out the sector’s limited capacity to directly support broader employment gains.

Looking ahead, systemic uncertainties persist, from potential supply constraints in advanced semiconductors to the sustainability of enterprise AI demand beyond this initial investment phase. Long-term utilization rates for hyperscaler infrastructure, as well as the alignment of realized revenues with sunk costs, remain open questions.

The implications of capital expenditures on this scale are clear: AI spending, once viewed as a niche element of the tech sector, is now a macroeconomic driver. Whether the infrastructure will yield the intended returns remains to be seen, but its effects on supply chains, credit markets, and policymaking are undeniable. As capital keeps flowing, the question is no longer just how AI will transform the world technologically—but how its funding architecture will reshape the global economy.