Stock Markets February 25, 2026

Goldman Sees Hyperscaler AI Capex Hitting Its High Point Later This Year

Analysts warn a slowdown in infrastructure spending could pressure AI 'picks and shovels' stocks, with Nvidia singled out for vulnerable valuations

By Ajmal Hussain AMZN MSFT GOOG META ORCL
Goldman Sees Hyperscaler AI Capex Hitting Its High Point Later This Year
AMZN MSFT GOOG META ORCL

Goldman Sachs analysts say colossal capital spending by hyperscaler firms on AI infrastructure is set to crest later this year, a shift that may strain stocks tied to AI hardware and services. The bank projects cumulative hyperscaler AI capex of $1.5 trillion between 2023 and this year, and forecasts a 2026 spend of $667 billion. Goldman expects quarterly capex growth to slow in late 2026, creating revenue and valuation risks for suppliers of the AI ecosystem.

Key Points

  • Goldman Sachs estimates hyperscaler capital spending on AI reached about $1.5 trillion between 2023 and this year, versus roughly $600 billion before 2022.
  • Hyperscaler capex is forecast to reach $667 billion in 2026 - $127 billion higher than at the start of the fourth-quarter earnings season and 62% above 2025.
  • Analysts warn that a deceleration in AI infrastructure capex growth is likely in late 2026, creating revenue and valuation risk for suppliers of AI hardware and services - notably those seen as providing the industrys "picks and shovels".

Goldman Sachs analysts are forecasting a near-term peak in spending by the largest technology platform companies as they expand AI capabilities, and they warn the expected deceleration could be a headwind for companies that supply the underlying hardware and services.

In a client note, the team led by analysts including Ryan Hammond and Ben Snider estimated that so-called hyperscaler firms - Amazon, Microsoft, Alphabet, Meta Platforms and Oracle - will have invested roughly $1.5 trillion into AI-related capital expenditures between 2023 and this year. By comparison, the analysts said, total investment prior to 2022 was about $600 billion.

Goldman projects hyperscaler capex in 2026 alone to reach $667 billion. That figure is presented as $127 billion higher than at the start of the fourth-quarter earnings season and represents an increase of 62% over the amount spent in 2025, according to the note.

The brokerage warned that the intensity of the current spending cycle is exceptional. The analysts wrote that capex is on pace this year to consume more than 90% of cash flows for these companies, a share they say exceeds the level seen during the dotcom boom in the late 1990s and early 2000s.


Despite the scale of investment, Goldman flagged a likely slowdown ahead. The analysts predicted that the rapid quarterly growth rate in AI infrastructure capital expenditures should decelerate in late 2026. That slowdown, they said, would make revenue growth and market valuations for certain suppliers - often characterized as the industrys "picks and shovels" - vulnerable.

The note highlights a recent divergence between Nvidias share price and its results as an example of how difficult it can be to sustain outsized returns. Nvidia, which produces processors that serve as the computational core for many advanced AI systems, has reported blockbuster revenue expansion in recent years. Yet Goldman observed signs of cooling growth and noted that Nvidias share performance this year has been muted.

Analysts also emphasized the competitive environment facing Nvidia. They noted that Nvidia will report its latest quarterly results after the close of U.S. markets on Wednesday, and that the company faces tough year-over-year comparisons as well as intensifying pressures from rivals such as Advanced Micro Devices.

Goldmans strategists argued that a dominant position in the near term does not remove the risk that investors will fear over-earning if competition increases and uncertainties about sustained demand emerge. In their view, those dynamics underpin the vulnerability they see among stocks whose prospects have been buoyed by the AI capex wave.


The analysts did not suggest precise timing beyond the late-2026 slowdown or single out every company by name as uniformly exposed. However, their analysis implies that both the hyperscaler firms that are directing large sums into AI infrastructure and the hardware and services vendors that depend on that spending are parties to a cycle that may have reached or be nearing its peak in intensity.

Markets that could be affected include semiconductors and related AI infrastructure equipment, as well as software and services firms whose growth trajectories are tied to hyperscaler capex. The Goldman note frames the risk as a function of steep near-term investment followed by moderation in growth rates for that investment, with attendant effects on revenue growth and valuations of suppliers.

Investors and market participants will be watching upcoming earnings reports and subsequent capex guidance closely, as those updates may provide clearer signals on how sustainable current AI investment trends will be.

Risks

  • A slowdown in hyperscaler AI infrastructure capital expenditures in late 2026 could weaken revenue growth and valuations for hardware and services vendors tied to that spending - impacting semiconductor and AI infrastructure sectors.
  • A recent disconnect between Nvidias price and results highlights the challenge of maintaining persistently strong returns, suggesting equity risk for companies whose rallies have been driven primarily by earnings expectations.
  • Intensifying competition, exemplified by pressure from rivals such as Advanced Micro Devices, increases uncertainty about the sustainability of current demand levels and could erode near-term dominant market positions.

More from Stock Markets

Tel Aviv stocks slip as Real Estate, Oil & Gas and Communications lead losses Feb 25, 2026 Buyout Interest in MarineMax Grows After Donerail Proposal Feb 25, 2026 Citigroup Assembles Dedicated AI Infrastructure Banking Team Feb 25, 2026 Athens bourse closes higher as banking, construction and travel stocks lead gains Feb 25, 2026 UBS Moves IBM to Neutral After Sharp Drop, Cites Improved Risk-Reward on Valuation Reset Feb 25, 2026