Announced capital spending across major U.S. technology companies now points to more than $1 trillion in desired annual AI investment, a figure that has risen dramatically in just a few years. Yet real U.S. GDP growth has averaged only roughly 2% to 2.5% since AI investment began accelerating in early 2025, a pace that looks nothing like an economy experiencing a historic investment boom. In a new PIMCO Macro Signposts analysis, economist Tiffany Wilding puts that contradiction into focus with a direct diagnosis: "AI may be reshaping the economy more than expanding it."
Crowding Out Revisited for a Private-Sector Boom
Wilding grounds her analysis in a classical economic framework. Crowding out, a term central to the Keynesian-monetarist debates of the 1960s and 1970s, describes what happens when a surge in one category of spending displaces another. A 1975 Federal Reserve Bank of St. Louis research paper concluded that the effect is a matter of degree, not an all-or-nothing outcome; that interest rates are only one of several possible channels; and that crowding out reflects constraints on both financing a boom and producing the goods, services, infrastructure, and labor it requires. Wilding applies that same structural logic to the current AI moment. The surge is private rather than government-driven, but when desired investment exceeds the saving and productive resources immediately available in an economy already operating near or above potential, adjustments must follow. Wilding identifies four pathways through which an economy can absorb the pressure: generating more domestic saving, importing capital goods and foreign saving, increasing the productivity of existing capital and labor, or reallocating resources away from other domestic uses.
Rate Insensitivity at the Top, Rate Sensitivity Below
Higher interest rates have been the central adjustment mechanism, but their effect is deeply asymmetric. The major hyperscalers driving AI buildout are unusually insulated from borrowing costs. They carry strong balance sheets, generate substantial internal cash flows, and concentrate spending in shorter-lived assets, where depreciation rather than interest rates dominates the economic cost of capital. What Wilding describes as a race to capture potentially large future returns is not a function of the current cost of capital. As a result, "Rate-sensitive sectors are bearing much of the cost," Wilding states, with housing, smaller businesses, leveraged companies, and conventional corporate investment absorbing the monetary tightening that the AI complex largely sidesteps. Outside the technology sector, private fixed investment has weakened materially. U.S. equity market dispersion has reached historically elevated levels, with an unusually high share of S&P 500 companies now negatively correlated to the index even as the index itself has advanced roughly 12% year-to-date.
Three Release Valves, Each With Limits
Wilding identifies three additional forces shaping the pace and severity of the adjustment. First, AI investment is heavily import-intensive: roughly 90% of equipment used by U.S. high technology sectors originates abroad, giving the U.S. access to foreign production capacity and foreign saving as a partial release valve, though higher prices for chips and related components signal that global supply capacity is itself becoming constrained. Second, physical bottlenecks, including data-center permitting, power generation and grid interconnection, component lead times, and specialized labor shortages, are spreading AI spending across a longer horizon and moderating near-term inflationary pressure, even as they raise the cost of individual projects. Third, credit markets are tightening conditions for riskier borrowers: spreads for lower-quality issuers have widened, high yield issuance slowed sharply in October, and markets are growing more selective even within the AI supply chain itself.
Labor Is Already Absorbing Part of the Adjustment
Wilding's analysis extends to the labor market in a way that carries direct implications for consumption. Recent data show a disproportionate increase in job separations among workers who had previously earned above-median wages, contributing to softer wage and aggregate income growth even as equipment spending accelerates. Real private labor income and real equipment shipments are moving in opposite directions, consistent with the view that labor is already taking on part of the adjustment.
The Key Question Is Orderliness, Not Continuation
Crowding out need not be permanent. If AI delivers its anticipated productivity gains, potential output rises and the economy can eventually sustain higher levels of investment and consumption with less pressure on inflation and rates. But Wilding is precise about the uncertainty that precedes that outcome. The central macro question, she argues, is "who makes room for it" and "whether that adjustment remains orderly." AI, in her framing, looks less like a net addition to an otherwise unchanged economy and more like a wholesale reorganization of it.
5 Key Takeaways for Advisors and Investors
- AI is restructuring the economy's capital map, not expanding it. Moderate aggregate GDP growth alongside a historic AI investment surge signals sectoral reallocation rather than broad-based expansion. Portfolios built for uniform growth exposure need to account for this divergence.
- Rate sensitivity is not evenly distributed. The largest technology companies driving AI capex are largely immune to higher borrowing costs. The adjustment burden falls disproportionately on housing, small business credit, leveraged issuers, and non-AI corporate investment, where rate pressure is most direct and most consequential.
- Import dependence is a buffer with a ceiling. Access to foreign equipment and foreign saving has limited domestic overheating, but rising prices for semiconductors and related components suggest that global production capacity is itself becoming a constraint. That buffer is narrowing.
- Widening credit spreads and slowing high yield issuance are signals, not noise. Credit markets are already repricing the distinction between AI beneficiaries and rate-exposed businesses. Growing selectivity within the AI supply chain itself adds another layer of discrimination to monitor.
- Declining real labor income for above-median earners is a structural trend to track. Disproportionate job separations in higher-wage categories are compressing aggregate wage and consumption growth. Advisors should treat this as a durable signal with implications for consumer-facing sectors and household spending capacity.
Footnote:
1 Wilding, Tiffany. "Is AI Crowding Out the Rest of the Economy?" PIMCO Macro Signposts, 28 Oct. 2026, https://www.pimco.com/gbl/en/insights/is-ai-crowding-out-the-rest-of-the-economy.
