Author: Claude, Deep Tide TechFlow

Deep Tide Insight: The latest chart from Torsten Slok, Chief Economist at Apollo Global Management, shows that the forward price-to-earnings ratio of the S&P 500 Information Technology sector has compressed from about 40 times at the peak of the AI frenzy to about 20 times, returning to the level before the AI boom began. The intertwining pressures of sector rotation triggered by the Middle East conflict, doubts about returns on AI capital expenditure, and slowing earnings growth mean that tech giants are experiencing the most painful valuation repricing since 2022.

The valuation bubble in the S&P 500 Information Technology sector is being rapidly squeezed.

According to a Daily Spark research report published by Apollo Global Management on April 11, the company’s partner and chief economist Torsten Slok revealed the current situation of tech stocks in a chart: the forward price-to-earnings ratio of the S&P 500 Information Technology sector has compressed from about 40 times during the AI frenzy to about 20 times, returning to pre-AI boom valuation levels.

This chart covers the ten largest components of this index: Nvidia, Apple, Microsoft, Broadcom, Oracle, Micron, Palantir, AMD, Cisco, and Applied Materials. In other words, the core group of winners in the AI era has collectively given back more than two years of valuation premium.

Intertwined pressures lead to valuation repricing for tech giants

The valuation compression is not driven by a single factor, but rather the result of multiple negative pressures.

The Middle East conflict is the most direct catalyst. Since the outbreak of the Iran war, the energy sector surged over 34% in the first quarter, and ExxonMobil's gains have approached 42% since the beginning of the year. Large amounts of capital have flowed from tech stocks to energy and defensive sectors, with tech stocks becoming the largest outflow source. The S&P 500 formed a "death cross" (50-day moving average crossing below the 200-day moving average) at the end of March, and as of early April, the index was struggling around 6582 points, less than 100 points from the "adjustment zone" threshold of 6300 points.

Concerns over the return on AI capital expenditures are the second layer of pressure. According to FactSet data, the expected earnings growth for the S&P 500 in the first quarter is 12.6%, with a forward price-to-earnings ratio of about 20.4 times. Tech giants have invested massive capital expenditures over the past two years (Amazon plans to spend $200 billion by 2026, while Microsoft, Meta, and others have plans for similar levels of investment), but AI-related revenues are still far below the scale of investment. According to estimates cited in a CEIBS report, AI capital expenditures of about $400 billion by 2025 will need to generate annual AI-related revenues of $160 billion to break even, while actual revenues at that time are expected to be only about $15 billion to $20 billion.

Slowing earnings growth constitutes the third layer of pressure. According to Bloomberg Intelligence data, the expected earnings growth for the "seven giants" in 2026 is around 18%, the lowest level since 2022, and the gap with the expected growth rate of 13% for the remaining 493 companies in the S&P 500 is narrowing. David Lefkowitz, head of U.S. equity at UBS Global Wealth Management, stated in January that the trend of spreading earnings growth is occurring, and technology is no longer the sole protagonist.

Nvidia at 21 times, Microsoft plummets 23%: Divergence among giant stocks intensifies

The valuation compression is even more pronounced at the individual stock level.

According to Zacks analysis, Nvidia's forward price-to-earnings ratio has fallen to about 21.4 times, well below its median of 45.3 times over the past decade, even though the annualized growth rate of earnings over the next three to five years is still expected to reach 39.1%. Microsoft has dropped about 23% since the beginning of the year, with its market capitalization falling below $3 trillion after breaking the $4 trillion mark last October. Apple is relatively stable among the "seven giants," partly because its AI capital expenditures are far lower than its peers; it repurchased $24.7 billion in stock within a quarter, and its capital discipline has earned a premium in a market that punishes those who make large expenditures.

Insider actions may explain the situation more clearly. According to Motley Fool citing SEC Form 4 data, during the past two years up to April 2, insiders at Nvidia, Apple, Alphabet, Microsoft, and Amazon have net sold stocks totaling about $16.1 billion. Although most of the selling is related to tax-related compensation handling, the absence of a buying signal from insiders amid such significant net selling still leaves the market uneasy.

Debates about the AI bubble are heating up, but there are essential differences from the 2000 internet bubble

Does the return of tech stock valuations to pre-AI levels mean that the AI bubble has burst?

There are clear differences among institutions. BlackRock pointed out in its report on the technology sector that the forward price-to-earnings ratio of the S&P 500 Information Technology Index was about 30 times in October 2025. While this is at a historical high, it is far lower than the approximately 60 times level of the Nasdaq 100 at the peak of the internet bubble. BlackRock emphasized that the current valuations reflect real revenues, validated business models, and accelerating AI adoption, which is completely different from 2000.

Goldman Sachs' previous research report also pointed out that while the long-term dividend growth implied by current stock prices is unreasonably high, it is still below the extreme levels during the internet bubble and the "Nifty Fifty" period of the 1960s.

However, warning signs are also evident. According to Globe and Mail, the S&P 500, when measured by the Shiller price-to-earnings ratio, entered the second highest valuation range in 155 years at the beginning of the year. Historically, whenever the Shiller price-to-earnings ratio has exceeded 40 times (during the internet bubble and January 2022), the S&P 500 subsequently declined by 49% and 25%, respectively.

Zacks analysts are more pragmatic in their assessment: while stock prices are falling, earnings expectations are being raised, leading to a passive compression of valuation multiples. The risk-reward ratio for some individual stocks is improving. Nvidia is considered to have the highest growth-to-valuation match at the current valuation, while Microsoft is seen as having potential for a "catch-up rebound."

For investors, the key question is not whether AI is valuable, but whether the massive capital expenditures can convert into profit returns that match the valuations within a reasonable time frame. If 2026 marks the cyclical peak of capital expenditures from large-scale customers, then even if the technology itself continues to evolve, the investment return cycle surrounding AI infrastructure development may far exceed the market's patience.