Is the market overestimating AI? Goldman Sachs' macro team provides a "simple calculation"

Author: Zhao Ying

Source: Hard AI

In the current AI-driven market frenzy, the question investors are most concerned about is: What is the true value of AI? Has the market overvalued it?

Goldman Sachs' macro strategy team, in their latest report, does not get bogged down in the prospects of individual companies but instead raises a warning from a top-down macroeconomic perspective through a "simple arithmetic calculation": the U.S. stock market may have already overdrawn most of the potential gains brought by artificial intelligence.

Research shows that the market capitalization of AI-related companies has soared by over $19 trillion since the launch of ChatGPT, reaching or even exceeding Goldman Sachs' benchmark estimate of the present value of future capital gains from AI (approximately $8 trillion).

This means that market pricing has significantly outpaced the actual macroeconomic impact, and investors need to be wary of the risk of overvaluation, especially when there is a turning point in the economy or the AI investment boom.

The "Ceiling" of Macro Gains — $8 Trillion

The report first sets aside individual stock valuations and attempts to estimate the total gains that AI may bring to the entire U.S. economy. Based on its previous prediction that AI will boost productivity by 1.5 percentage points over the next decade, Goldman Sachs economists estimate that the present value (PDV) benchmark for future total capital income generated by this will be approximately $8 trillion, with a reasonable range between $5 trillion and $19 trillion.

Goldman Sachs states that if labor is not permanently displaced and the profit share in the economy remains stable, corporate profit levels will also permanently increase by about 15% after the transition period. The growth in stock market value should equal the discounted value of these additional profits, and due to the transition period, this increase will be lower than the long-term growth rate of GDP and profits of 15%.

$19 Trillion — Market Valuation is "Far Ahead"

The problem is that market pricing has long surpassed the realm of macro fundamentals. The report sharply points out that since the release of ChatGPT, the market capitalization growth of just semiconductor and AI model developers has exceeded $8 trillion, exactly equal to Goldman Sachs' benchmark forecast for potential total capital gains from AI. If the scope is expanded to all AI-related listed companies, the market capitalization increase is as high as $19 trillion.

Goldman Sachs states:

These simple calculations suggest that the market may have already pre-digested most of the potential value of AI, and this value is mainly concentrated in companies that are directly involved in or closely adjacent to the AI boom. … Market pricing has far outpaced macro impacts.

Not Yet at Bubble Levels, Stay Alert but Don’t Panic

Goldman Sachs points out that although company valuations are high, they have not yet reached bubble levels. The macro perspective reveals the systemic risk of potential overpricing in the market; even if the valuation of individual companies seems reasonable, the total valuation of all companies may not be reasonable—because even significant productivity gains have an upper limit on their impact on overall economic profits.

The report also warns investors to be wary of two common cognitive fallacies:

One is the "aggregation fallacy," which is the mistaken belief that many companies can all become big winners, leading to expectations of total profits exceeding reasonable bounds;Secondly, the "extrapolation fallacy," which views the excess profit growth in the early stages of innovation as the norm, while ignoring the rule that competition will eventually erode profits.

However, research also points out that even if a large amount of valuation gains have already been accounted for, it does not mean that more cannot continue to be included. Even if the final earnings are at the lower end of the estimated range, this situation may still occur. Past innovation-driven booms—such as those in the 1920s and 1990s—have led to the market paying excessively high prices for future profits, even though the underlying innovations were real.

For investors, this means a need to more cautiously assess the risk-return ratio of AI-related investments, with particular attention to early signals of changes in economic cycles and the sustainability of the AI investment boom. In the current high-valuation environment, once a fundamental turning point occurs, the extent of market adjustments may exceed expectations