AI Capex Fuels Half of S&P 500 Earnings Growth, Goldman Says

Earnings concentration raises questions about index resilience

Artificial intelligence capital expenditure now accounts for roughly half of all S&P 500 earnings growth, according to Goldman Sachs chief U.S. equity strategist Ben Snider, who discussed the figure during a recent CNBC appearance. While the statistic is frequently cited as evidence of a durable bull market, it also highlights a narrow foundation: a small cohort of buyers and suppliers is responsible for keeping index profit expansion on track.

NVIDIA, Broadcom, and Microsoft sit at the center of this cycle. NVIDIA reported second-quarter fiscal 2027 revenue of $96.22 billion, up 105.8% year over year, with data center revenue reaching $89.02 billion. Broadcom's AI semiconductor revenue hit $16.70 billion in its third quarter, a 221% annual increase, with fourth-quarter guidance of $21.7 billion. Microsoft is projecting roughly $175 billion in capital expenditure for fiscal 2027, up from $115.95 billion in the prior year.

Why it matters

When half of an index's earnings growth depends on sustained spending by three or four companies, any pullback in that capital cycle translates directly into weaker profit expansion across the benchmark. The setup creates asymmetric risk: if hyperscalers slow infrastructure investment, the earnings tailwind that has supported the S&P 500's advance disappears quickly. For portfolio managers and corporate strategists, understanding this dependency is essential to evaluating both equity positioning and the durability of current AI-driven valuations.

Earnings growth, not multiple expansion, drove the rally

Snider noted that the S&P 500 traded at 22 times earnings at the start of the year and now sits at 19 times, even as the index remains near record highs. The SPDR S&P 500 ETF Trust closed at $757.94 on September 10, up 11.13% year to date. The compression in the price-to-earnings multiple while the index rose signals that earnings grew faster than stock prices—a healthier dynamic than multiple expansion alone.

The ten-year Treasury yield now stands at 4.83%, in the 99.6th percentile of the past year, compared to roughly 4% a year ago when the S&P 500 traded around 6,600. Snider argued that yield volatility matters more than the absolute level, and that the earnings backdrop remains strong as long as profit growth continues at its current pace.

Circularity and capital mobility

The interdependence within the AI infrastructure ecosystem is pronounced. NVIDIA is extending take-or-pay commitments and credit enhancements to neoclouds and frontier labs, while partnering with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to mobilize over $500 billion in third-party capital for AI infrastructure. This structure accelerates deployment but also concentrates risk.

Capital spending is discretionary. If two or three hyperscalers decelerate investment, the earnings growth Snider described thins rapidly, and the index math shifts. Suppliers that serve multiple buyers—power, cooling, networking—offer more diversified exposure to the buildout.

Enterprise adoption and positioning

Snider said enterprise AI spending has accelerated beyond infrastructure over recent months, after four years focused primarily on data center buildout. Microsoft reported that 365 Copilot reached over 30 million paid seats, and commercial remaining performance obligations grew 84% to $678 billion. Most S&P 500 companies have not yet reflected AI-driven earnings accretion in reported figures, so the broadening remains largely narrative.

Goldman's positioning indicator, which aggregates hedge fund, mutual fund, and other investor exposure, is at its lowest level since March, even with the market near a high. Light positioning reduces the risk of forced selling during a drawdown. September seasonality is typically weak, and the VIX has risen to 16.46.

Implications for index investors

The case for the S&P 500 depends on earnings continuing to grow at a pace that justifies current prices. But with half that growth tied to AI capex, any slowdown in hyperscaler spending would materially alter the index's profit trajectory. The ten-year yield at 4.83%, thin positioning, and September seasonality suggest a balanced rather than compelling setup for new capital deployment.

If enterprise AI spending begins to appear in reported earnings from a broader set of companies, concentration risk diminishes and the index foundation strengthens. Until then, a small group of buyers and suppliers is carrying the benchmark, and their quarterly spending plans warrant close attention.

These details were first reported by 24/7 Wall St.