Investors riding the AI trade’s historic momentum face a stark new arithmetic: Goldman Sachs says an AI slowdown—if hyperscaler capital spending reverts toward pre-boom levels—could compress S&P 500 valuation multiples by 15–20%, translating into a market drawdown as steep as 20% when enthusiasm cools. The bank pegs roughly $1 trillion of expected S&P sales growth to AI, warning that a capex reversion could slice about 30% from that tally. Timing risk builds into late 2025. [1]

In separate guidance this year, Goldman cut its S&P 500 year‑end target to 6200 from 6500, lowered 2025 EPS growth expectations to 7% from 9%, and projected $262 in profits per share, underscoring how softer GDP and potential AI capex declines could weigh on multiples. Those macro and AI-linked headwinds narrow upside while raising downside sensitivity. [5]

Probability math is deteriorating too. Goldman’s equity-asymmetry framework shows a greater than 10% chance of a >10% correction in three months and greater than 20% within a year, with stretched tech valuations and possible AI capex reversion among the catalysts raising correction odds. [4]

Key Takeaways

– Shows 15–20% S&P 500 multiple compression risk if an AI slowdown hits, aligning with Goldman’s warning of a market drawdown up to 20%. [1] – Reveals a $1 trillion S&P sales-growth case tied to AI could face a 30% haircut under a capex reversion scenario toward 2022 levels. [1] – Demonstrates hyperscaler capex reached about $368bn YTD, with deceleration risks flagged for late‑2025 into 2026 as spending normalizes. [1][3] – Indicates a >10% probability of a >10% drawdown in three months and >20% within a year amid elevated tech valuations. [4] – Suggests Goldman’s 6200 S&P target and 7% EPS growth forecast reflect sensitivity to slower GDP and potential AI capex declines. [5]

Why Goldman sees an AI slowdown as inevitable

Goldman’s base case doesn’t assume AI demand disappears; it assumes the current pace of investment is unsustainable. Hyperscaler spending surged, with capex reaching about $368 billion year-to-date, but management guides and historical cycles imply eventual normalization as capacity catches up with demand. A reversion toward 2022 spending levels would mark a cooling after an extraordinary buildout, leaving less fuel for AI suppliers and the broader equity narrative. [1]

The bank’s analysts argue that a slowdown is “inevitable” when early infrastructure phases mature, utilization rates stabilize, and returns are forced to justify incremental outlays. That pivot, if it arrives into late 2025, risks colliding with already stretched U.S. equity valuations and a narrow leadership profile dominated by AI beneficiaries. Those conditions heighten the market’s exposure to any deceleration in the AI capex engine. [1][2]

How an AI slowdown could compress S&P 500 multiples

The mechanism runs through both growth expectations and discount rates. Goldman estimates roughly $1 trillion in incremental S&P sales growth embedded in AI adoption cases; cut 30% from that, and medium‑term revenue and margin assumptions fall, reducing fair values. Historically, when structural growth narratives cool or delay, valuation multiples compress faster than earnings adjust. Goldman quantifies this at 15–20% for S&P 500 multiples under a capex reversion scenario. [1]

That compression would layer onto Goldman’s already tempered 2025 outlook—EPS growth of 7% versus prior 9% and a 6200 index target—leaving less cushion for disappointment. If growth slows while risk premia rise, price/earnings ratios reset lower. Even without an earnings recession, a 15–20% multiple contraction can drive double‑digit index declines as the market re-prices future AI cash flows. [5]

The hyperscaler capex swing factor

Spending by hyperscalers is the fulcrum. AI data centers, advanced networking, accelerators, and power infrastructure have pushed capex into the $300–$368 billion range, dwarfing prior cycles. Timing matters: Goldman flags late‑2025 and into 2026 as the window when deceleration could emerge, as capacity utilization improves and boards scrutinize marginal returns. A shift toward steadier, maintenance-like outlays would cascade through semiconductor, compute, and cloud ecosystems. [3]

If capex reverts toward 2022 levels, suppliers most levered to training compute, memory bandwidth, and power delivery could see the sharpest near-term revenue pressure. The broader index impact arises because AI leaders carry large weights, and their multiple compression transmits to benchmarks. Even modest capex moderation can curtail backlog visibility and temper sales-growth trajectories across adjacent sectors. [1][3]

Market context: concentration magnifies the cost of an AI slowdown

Concentration and valuation amplify the asymmetry. Goldman’s strategists highlight how U.S. equities have grown pricier and more concentrated in large-cap tech and AI winners, increasing the market’s sensitivity to any cooling in AI spend. In this “postmodern” market frame, where narratives and capital flows shape returns, uncertainty around hyperscaler capex elevates downside risk to both earnings and index multiples. [2]

Risk distribution has tilted. Goldman’s equity-asymmetry model indicates a greater than 10% chance of a >10% drawdown within three months and greater than 20% within a year, with tariff and inflation risks compounding the vulnerability posed by a potential AI capex normalization. The upshot: the market’s payoff profile has worsened even before any definitive AI slowdown shows up in quarterly prints. [4]

Timelines and triggers for an AI slowdown

Goldman’s team points to late‑2025 into 2026 as the most likely window for deceleration, as the first wave of AI infrastructure saturates and companies pivot from build to optimize. Firms will increasingly demand proof of return on invested capital, while bottlenecks in power, cooling, or supply chains may ease, removing “forced spend” dynamics. Any shift in capex guidance from hyperscalers could ripple through supplier order books within weeks. [1][3]

Macro conditions are a secondary trigger. Goldman’s March 26 research note lowered index and EPS growth assumptions as GDP momentum cooled, underscoring that a softer macro can reduce the willingness to fund long‑dated AI projects at current hurdle rates. In that context, even a small slowdown in AI capital intensity can have outsized market effects because valuations already discount robust growth. [5]

What a 15–20% multiple compression implies for index math

Goldman’s stress test translates narrative risk into numbers. If the S&P 500’s blended forward P/E compressed by 15–20% from today’s levels due to fading AI capex momentum, index pricing could fall by a similar magnitude, assuming earnings remain near current forecasts. The effect compounds if EPS revisions follow lower capex, given Goldman’s own EPS growth cut from 9% to 7% this year. [1][5]

Because the largest AI beneficiaries carry significant benchmark weight, their multiple compression can mechanically pull down the index even if many constituents hold steady. With around $1 trillion in AI-linked sales growth embedded in expectations, a 30% haircut would permeate margin and free‑cash‑flow models, curtailing the premium investors are willing to pay for long-duration AI cash flows. [1]

Portfolio playbook before an AI slowdown

Goldman’s MarketWatch‑cited guidance stresses diversification away from single-factor bets and concentrated exposures, recommending allocations that can withstand uncertainty in hyperscaler capex trajectories. That includes balancing growth with value, leaning into quality balance sheets, and maintaining liquidity to exploit volatility spikes if multiples compress. Elevated valuations and capex uncertainty make index downside more convex to AI newsflow. [2]

On the numbers, watch three dials: hyperscaler capex guidance and purchase commitments; revisions to the S&P’s 2025 EPS outlook (currently 7% growth and $262 per share in Goldman’s model); and market-implied drawdown probabilities. A meaningful downgrade in any of these can shift fair value ranges quickly, especially if concentrated leaders guide cautiously on AI monetization timelines. [5][4]

What could delay or blunt an AI slowdown?

There are offsets. If AI monetization runs ahead of plan—faster revenue per inference, higher enterprise adoption, or improved inference margins—boards may sustain elevated capex, extending the growth runway. Yet Goldman’s warnings, and FT’s reporting, emphasize that even strong adoption can leave equities vulnerable when valuations are rich and growth narratives are priced to perfection. In such regimes, multiple compression can arrive despite healthy fundamentals. [3][2]

Another cushion is macro stabilization. Should GDP growth reaccelerate, risk premia fall, or rates decline, multiples could prove more resilient even as AI capex normalizes. Nonetheless, Goldman’s baseline remains that slower GDP and potential AI capex declines would weigh on S&P upside relative to earlier expectations, keeping the risk/reward less favorable than in 2023–2024’s early AI surge. [5]

The bottom line on an AI slowdown for the S&P 500

The bull case for AI is intact over multi‑year horizons, but the spending cycle will not stay vertical. Goldman’s math is explicit: a capex reversion toward 2022 levels could cut about 30% from a $1 trillion AI sales-growth case and compress S&P 500 valuation multiples by 15–20%, implying a potential index drawdown up to 20% when momentum cools. In a concentrated, richly valued market, that’s a risk investors cannot ignore. [1][2]

Sources:

[1] Fortune (via inkl) – When AI’s ‘inevitable slowdown’ comes it could tank the S&P 500 by up to 20%, Goldman Sachs says: www.inkl.com/news/when-ai-s-inevitable-slowdown-comes-it-could-tank-the-s-p-500-by-up-to-20-goldman-sachs-says” target=”_blank” rel=”nofollow noopener noreferrer”>https://www.inkl.com/news/when-ai-s-inevitable-slowdown-comes-it-could-tank-the-s-p-500-by-up-to-20-goldman-sachs-says

[2] MarketWatch – Global markets are entering a new ‘postmodern’ era. This is how investors can win big, says Goldman Sachs: www.marketwatch.com/story/global-markets-are-entering-a-new-postmodern-era-this-is-how-investors-can-win-big-says-goldman-sachs-ce5f369e” target=”_blank” rel=”nofollow noopener noreferrer”>https://www.marketwatch.com/story/global-markets-are-entering-a-new-postmodern-era-this-is-how-investors-can-win-big-says-goldman-sachs-ce5f369e [3] Financial Times – Does the market need to be concerned about AI adoption?: www.ft.com/content/3c3977cd-e503-4d71-aa3b-7b2f23591d55″ target=”_blank” rel=”nofollow noopener noreferrer”>https://www.ft.com/content/3c3977cd-e503-4d71-aa3b-7b2f23591d55

[4] Business Insider – Goldman Sachs says the risk of stock-market decline has suddenly spiked: www.businessinsider.com/stock-market-outlook-sp500-prediction-correction-tariffs-economy-goldman-sachs-2025-8″ target=”_blank” rel=”nofollow noopener noreferrer”>https://www.businessinsider.com/stock-market-outlook-sp500-prediction-correction-tariffs-economy-goldman-sachs-2025-8 [5] Goldman Sachs (Research note) – The S&P 500 may rise less than expected as GDP growth slows: www.goldmansachs.com/insights/articles/the-sp-500-may-rise-less-than-expected-as-gdp-growth-slows” target=”_blank” rel=”nofollow noopener noreferrer”>https://www.goldmansachs.com/insights/articles/the-sp-500-may-rise-less-than-expected-as-gdp-growth-slows

Image generated by DALL-E 3