The AI boom still has fundamental support, but market pricing is already ahead of macro realization

Overview

This report compares the current AI boom with the late-1990s tech bubble. Goldman Sachs believes that the AI capex boom is still advancing, and near-term earnings remain strong, so the typical macro imbalances have not yet clearly emerged. However, the value assigned to AI-related assets by the market has risen further, and the tension between macro fundamentals and elevated valuations continues to widen.

Core views

The report's core view is that the AI investment cycle could still last longer, and related companies' earnings may continue to outweigh valuation concerns in the near term; however, the market has already pulled a large amount of future AI gains forward into prices. The current risk is less likely to take the form of a pure valuation bubble like in 1999-2000, and more likely to show up as the market overestimating the unusually high earnings persistence of AI infrastructure suppliers and related winners.

Analysis framework

The report uses two approaches: historical analogy and macro cross-check. On the one hand, it compares the current AI cycle with the late-1990s tech cycle in terms of investment intensity, profit margins, corporate financing needs, the current account, and market volatility; on the other hand, it contrasts the future present value of capital income that AI productivity gains could generate with the market-cap increase in AI-related stocks to judge whether market pricing has exceeded what the macro economy can ultimately deliver.

Methodology notes

• Historical cycle comparisonComparison with the 1990s tech bubble

  Macro imbalance signals

  The report focuses on four types of signals: whether investment stays abnormally high for an extended period, whether macro profit margins decline, whether corporate financing needs and leverage rise, and whether the current-account deficit widens. At present, only the scale of investment is clearly near 1990s levels, while the other imbalances are not yet prominent.

• Valuation constraintComparison of AI earnings present value and market-cap gains

  Present Discounted Value

  The report uses the present value of future capital income that AI productivity gains could generate as a macro constraint on the increase in market capitalization of AI-related companies, avoiding the aggregation fallacy in which individual company valuations may look reasonable but the total implied gains exceed the economy's overall ability to deliver.

• Market structure analysisProfit share and capital-biased technological change

  Can AI winners capture excess profits

  The most persuasive bullish scenario, in the report's view, is that AI-related companies can capture a profit share above the economy-wide average over the long run; however, competition, investment expansion, and subsequent innovation may erode those excess profits.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• AI-related stocks

  Core beneficiary asset

  Strengths

  The capex cycle remains strong, earnings expectations continue to be revised upward, and market structure and technical characteristics may support a higher share of profits.

  Weaknesses

  Market value has already priced in a large amount of future earnings, and multiple optimistic assumptions are needed to match the present value of macro earnings.

  Comparison

  Compared with the late 1990s, the recent gains have been driven more by earnings than by multiple expansion, but valuations remain near historical highs.

  Risks

  Earnings durability is overestimated, the AI investment cycle peaks, adoption slows more than expected, and competition erodes profits.

• Semiconductor indices and SOX

  High-beta beneficiary of AI infrastructure investment

  Strengths

  The past few years' gains have already come close to the performance of the Nasdaq in the late 1990s, directly benefiting from AI capex.

  Weaknesses

  It relies more heavily on AI infrastructure investment intensity, and cyclical as well as earnings pullback risks are higher.

  Comparison

  Broad U.S. equities have risen more moderately, while the narrow semiconductor index is already closer to the late stages of a historical bubble.

  Risks

  A slowdown in capex growth, overinvestment in the supply chain, and a dual reset in valuation and earnings.

• U.S. equity market

  The AI boom supports the overall index through earnings and wealth effects

  Strengths

  Corporate profit margins remain high, the overall corporate financial balance has not yet deteriorated materially, and forward P/E has not expanded materially alongside the share-price rally.

  Weaknesses

  Backward-looking valuation metrics remain at very high levels, and non-AI economic momentum is weaker than in the late 1990s.

  Comparison

  Compared with 1999-2000, there are fewer macro overheating signals; however, the market's dependence on the AI earnings story is more concentrated.

  Risks

  Equity volatility continues to rise, negative AI news hits the index, and non-AI economic fragility amplifies macro shocks.

• Credit markets

  A supplementary indicator for AI financing pressure and systemic risk

  Strengths

  Credit spreads are mostly still tight, and leverage is below 1998-2000 levels.

  Weaknesses

  AI-related issuance is rising, and hyperscale cloud providers' free cash flow has already declined materially.

  Comparison

  The kind of combination seen in the late 1990s, where credit stress and rising stock prices coexisted, has not yet emerged.

  Risks

  If financing constraints intensify, they could weaken AI capex and earnings expectations.

• Equity options and volatility

  Risk management tool

  Strengths

  Rising single-stock volatility and declining correlations make structural protection more attractive in portfolio allocation.

  Weaknesses

  Protection costs may increase as volatility rises.

  Comparison

  The late 1990s showed that equity volatility may still have room to rise further.

  Risks

  The bull market enters a more fragile phase, and single-stock dispersion and index volatility may rise together.

Key data

• AI-related market-cap increaseabout $27 trillionSince the end of November 2022, the market-cap increase of AI-related companies has risen further from about $19 trillion in November 2025 to about $27 trillion.
• Base-case increase in AI capital valueabout $9 trillionIn Goldman Sachs' base estimate, the additional capital value created for the U.S. economy by AI productivity gains is about $9 trillion, below most estimates of AI-related market-cap gains.
• 2026 capex expectations for hyperscale cloud providersup nearly 80% from six months agoThis shows that the AI investment cycle is still accelerating.
• Growth in U.S. real disposable incomeannualized at about 1% over the past 2 yearsThis is well below the roughly 5%-6% level seen in the late 1990s, indicating that the macro backdrop outside AI is more fragile.
• Conservative AI-attributed market-cap estimateabout $14 trillion to $17 trillionIf only the more purely AI-benefiting companies are counted, or if only part of the gains in certain hyperscalers and other AI-related companies are attributed to AI, the estimate is still above the base-case macro present value of gains.

Impact & implications

For investors, the AI theme still has support from earnings and capital expenditures, but asset prices are more sensitive to negative news. If AI adoption slows, financing constraints emerge, cost limits spread, or technological innovation reduces capital-expenditure intensity, the earnings expectations for current winners may be repriced. The report leans toward staying involved, but with greater emphasis on downside protection and volatility management.

Risks

• The market overestimates the durability of unusually high earnings at AI-related companies.
• The AI investment cycle peaks or capex growth slows.
• AI gains are realized more slowly than expected, or deployment costs limit adoption.
• Competition, investment expansion, and subsequent innovation erode the profit margins of current winners.
• The macro backdrop outside AI is relatively fragile, making the economy more vulnerable to supply-side or demand-side shocks.
• Equity volatility rises further, increasing the market's sensitivity to negative news.

What to watch

• Changes in capex expectations and free cash flow at hyperscale cloud providers.
• Whether earnings revisions for AI-related companies can continue to support stock-price gains.
• Overall corporate financial balances, credit spreads, and financing conditions for AI-related issuance.
• Whether U.S. profit margins, unit labor costs, and wage growth begin to show late-1990s-style pressure.
• Single-stock volatility, implied correlation, index volatility, and changes in option skew.
• AI adoption speed, evidence of productivity gains, and whether non-AI industries begin to capture AI benefits.