For years, investors in the biggest technology companies could count on one reliable outcome: the cash would come back to them. Profits piled up, balance sheets swelled, and buyback programs grew year after year, quietly supporting stock prices even when markets turned volatile.

That dynamic is changing. A research note published May 7 by Goldman Sachs lays out exactly what is replacing it, and the implications stretch well beyond any single company or sector.

AI spending is crowding out buybacks

Goldman Sachs released a research note on May 7 identifying a major change in how the largest technology companies are allocating their cash.

The bank’s analysts found that artificial intelligence infrastructure spending has become so large that it is directly squeezing the share repurchase programs that investors have long relied on.

More Wall Street

The five largest AI hyperscalers, Amazon, Alphabet, Meta, Microsoft, and Oracle, are expected to spend a combined $755 billion on capital expenditures in 2026, according to consensus estimates cited in the report.

That figure represents an 83% increase year over year, according to Goldman Sachs.

The scale of that commitment is notable for one specific reason. Goldman Sachs estimated that hyperscaler capital expenditures will amount to roughly 100% of their cash flows from operations this year. That leaves almost no room for shareholder returns unless companies take on significantly more debt or cut their investment plans, Goldman Sachs noted.

Buybacks are already feeling the pressure

The evidence is already in the data. Goldman Sachs noted that buybacks among the five hyperscalers fell 64% year over year during the first quarter as companies directed more cash toward data centers, chips, and related infrastructure, according to Prism News.

The group now allocates about 15% of total cash spending to buybacks, down from an average of 27% between 2017 and 2022, before AI capital spending began accelerating at its current pace.

For the broader S&P 500, Goldman forecasts gross buyback growth of just 3% in 2026, compared with a 33% surge in capital expenditures across the index. The bank described the dynamic as an ongoing “rotation from buybacks to capex and R&D,” according to Goldman Sachs May 7 note.

Goldman Sachs just identified a quiet shift in how big tech is spending its cash that every equity investor should understand

Abbas/Getty Images

Why the cut in share buybacks matters

Corporate buybacks have served as one of the most consistent sources of demand for U.S. equities over the past decade.

=When companies repurchase shares, they reduce the number of shares outstanding, which can lift earnings per share and support stock prices even when broader market conditions are uncertain.

Goldman Sachs warned that this structural support could weaken. The bank expects gross S&P 500 buybacks to exceed $1 trillion in 2026, but noted that the net buyback yield, which measures corporate demand relative to total market capitalization, is likely to decline to its lowest level since 2020 as market caps continue expanding.

Goldman also found that buybacks are historically the most sensitive type of corporate spending to uncertainty. The report indicated that a 100-point increase in the Economic Policy Uncertainty Index has historically been associated with a roughly 20% decline in buybacks, compared with a 10% decline in total cash spending. With policy uncertainty still elevated following tariff-related disruptions, that creates an additional headwind, according to the Economic Policy Uncertainty Index.

Key findings from Goldman’s May 7 report:

  • S&P 500 capital expenditures forecast to grow 33% in 2026 and 22% in 2027, with the five AI hyperscalers accounting for 32% of S&P 500 capex in 2025, according to Goldman Sachs.
  • Microsoft indicated it expects to spend roughly $190 billion in capital expenditures in calendar 2026, partly reflecting higher component costs, including memory, CNBC noted.
  • Meta raised its 2026 capex guidance by $10 billion for similar reasons, according to Meta IR.
  • S&P 500 R&D spending forecast to grow 18% in 2026 and 10% in 2027, with the top 10 spenders expected to grow R&D by 27% in 2026, Goldman Sachs May 7 note revealed. 
  • If hyperscaler capex growth in 2027 matched the 2026 pace, total spending would approach $1.4 trillion, far exceeding projected cash flows from operations of $980 billion, according to Goldman Sachs

Semiconductors and banks could offset some of the decline

Goldman Sachs acknowledged that the pullback in hyperscaler buybacks will not go entirely uncompensated. Companies that benefit directly from AI infrastructure spending, particularly semiconductor firms, have been returning more cash to shareholders as revenues and profits climb.

Nvidia authorized a $60 billion buyback program in August 2025, according to The Motley Fool.

Broadcom authorized a separate $10 billion program this year. Combined, the two companies repurchased $52 billion in stock in 2025, up from $40 billion in 2024.

Financial sector firms, which represent about one quarter of all S&P 500 buybacks, are also expected to increase repurchase activity as regulatory conditions improve. Goldman noted that financials appear poised for continued buyback growth, which could absorb part of the gap left by the hyperscalers.

Balance sheets remain healthy for now

Despite the enormous scale of current spending plans, Goldman Sachs said corporate balance sheets are not yet under serious stress.

Hyperscaler net debt has risen from $69 billion at the start of 2025 to $206 billion, but net leverage relative to EBITDA stands at just 0.3x, well within manageable territory, according to Goldman Sachs.

For the broader S&P 500, interest coverage remains healthy, and much of the existing debt is fixed-rate and does not mature until after 2028. Goldman suggested that companies have the financial flexibility to continue investing aggressively if they believe AI demand remains strong.

Even so, the bank cautioned that persistently high policy uncertainty, supply chain disruptions, and rising component costs remain meaningful risks.

The broader message is that corporate America has entered a phase where AI investment takes priority over financial engineering, and investors may need to adjust their expectations about buyback-driven market support accordingly.

Related: JPMorgan resets S&P 500 price target for the rest of 2026