Goldman Sachs: The AI boom has reached the same scale as the dot-com boom, but is following a different trajectory

Goldman Sachs Research believes that, in terms of the scale of investment, the AI boom has already matched the tech boom of the late 1990s, but has not yet followed the same pattern. Back then, stock prices rose even as companies’ profitability declined. Now, however, the rise in stock prices is accompanied by increased profits and higher profit forecasts. However, the market may be overestimating how long these profits will last—especially for AI infrastructure providers, whose revenues depend on record spending by tech giants.

Analysts at Goldman Sachs Research note that the tech boom of the late 1990s was accompanied by four warning signs: investment remained high for a long time, corporate profitability was declining, their need for external financing and debt burden were rising, and the gap between U.S. external expenditures and revenues was widening.

In the current market, only one of these four signals has materialized—investment in technology has reached a record share of the economy. By 2026, spending on software and computing equipment had approached 5% of U.S. GDP; at the peak of the dot-com boom, investment in the sector stood at about 4.5%. At the same time, the investment boom continues to accelerate: over the past six months, the largest cloud and computing companies have increased their planned spending for 2026 by nearly 50%.

Other signs characteristic of a bubble have not yet emerged, according to Goldman Sachs analysts. Profits at U.S. companies remain high: in 2026, pre-tax profits are projected to be about 13.8% of GDP, compared with approximately 10.5% at the peak of the dot-com boom. Profit growth is currently allowing companies to increase investments and maintain a stable financial position.

At the same time, outside of AI-related industries, the U.S. economy appears weaker than it did in the late 1990s. Consumer spending is growing moderately, real disposable income is growing much more slowly, and investment outside the technology sector remains weak. Analysts suggest that the AI boom is currently propping up the economy and offsetting the weakness in other sectors.

Despite the rally in AI-related stocks, their forward P/E ratio—the ratio of stock price to expected earnings—has barely risen this year, as earnings forecasts have risen at the same time.

According to Goldman Sachs Research’s base-case scenario, the future incremental profits that AI could generate for U.S. companies are estimated at approximately $9 trillion in today’s dollars. At the same time, since November 2022, the combined market capitalization of AI-related companies has already grown by $27 trillion; seven months ago, this increase stood at $19 trillion. Analysts note that not all of this growth is directly related to AI: the largest technology companies also have other business lines. Nevertheless, current stock prices can only be justified if AI is adopted more quickly, generates more profits for companies than the baseline scenario suggests, and they are able to sustain those profits for many years to come.

“There is a risk that the market is overestimating the sustainability of these profit streams over a horizon of more than two to three years, especially for companies that benefit directly from the capital expenditure boom. Forecasts for the coming years may look very promising, but it is much harder to predict what earnings will look like once the phase of rapid investment growth comes to an end,” the analysts write.

Even if the AI market continues to grow, current market leaders may not be able to maintain high profitability. According to Goldman Sachs, these margins could be eroded by intensifying competition, an influx of new investment, and further technological advancements. At the same time, it remains unclear how high the barriers to entry are and whether they will be able to protect existing players from new competitors.

Goldman Sachs believes that the risk of a classic valuation bubble is lower now than it was in the late 1990s and early 2000s. However, another imbalance—an “earnings bubble”—may be forming: stock prices are based on the assumption that current high profitability will persist for a long time. If the investment boom ends sooner than the market expects, the profits of AI infrastructure providers could decline. In that case, current valuations would prove to be overvalued, and the “profit bubble” would begin to deflate.

This article was AI-translated and verified by a human editor