Light capital dividends are not what they used to be, US tech giant AI gambling is being tortured by multiple pressures

The Zhitong Finance App learned that in the past 20 years, the success story of large technology companies has been quite simple and extremely successful: create disruptive innovation, achieve amazing growth rates, and control spending. A few giant companies, such as Alphabet Inc. (GOOGL.US), Amazon (AMZN.US), Meta (META.US), and Microsoft (MSFT.US), use this formula to seize market share from traditional companies and push US stocks to new highs over and over again. But a critical part of this successful model, the relatively small amount of capital required to generate huge profits, is increasingly threatened by the AI R&D competition.

Jim Morrow, CEO of Callodine Capital Management, which manages $1.2 billion in assets, said, “These companies have one of the best business models the market has ever seen,” “but now, as capital intensity has surged, they are now the most capital-intensive sector in the market. It's simply a fundamental change.”

Together, the four companies are expected to invest more than $380 billion in capital expenses in the current fiscal year, most of which will be spent on chips, servers, and other data center-related expenses. That's a jump of over 1,300% from ten years ago. Moreover, all of these companies have promised to significantly increase spending in the next fiscal year.

According to data, Microsoft's capital expenditure now accounts for 25% of its revenue, which is more than three times what it was ten years ago. The software and cloud computing giant's spending sales ratio is in the top 20% of the S&P 500 index's constituent stocks, as are Alphabet and Amazon, far higher than companies in traditional capital-intensive industries such as oil and gas exploration and telecommunications.

Despite uncertainty about future returns, investors have so far trusted the tech giants' AI plans. The stock prices of almost all large spenders have risen this year, and their valuations are high. For example, Microsoft's stock price rose 15% in 2025, and its stock's price-earnings ratio based on expected profit over the next 12 months is more than 28 times, higher than its ten-year average of about 27 times, and 22 times higher than the S&P 500 price-earnings ratio.

But signs of doubt are quietly surfacing. Meta, which owns Facebook and Instagram, was penalized by the market after its third-quarter earnings report was announced, mainly because CEO Mark Zuckerberg failed to chart a clear path to greater profits from rising AI spending. On October 30, Meta's stock price experienced its worst trading day in three years. It plummeted 11% the day after the earnings report was released, and fell further 3.8% thereafter. After surging 25% in the first three quarters, the stock is up 9.5% so far this year, underperforming the S&P 500.

One point of contention is the rise in depreciation costs brought about by AI chips and servers. Hedge fund manager Michael Berry, famous for “big shorts,” suggested that depreciation of such devices should be accelerated, which would seriously weaken these companies' profit growth.

These expenses also put pressure on free cash flow and may limit the extent to which capital can be returned to shareholders through share repurchases and dividends. For example, Alphabet is expected to generate $63 billion in free cash flow this year, down from $73 billion last year and $69 billion in 2023. According to the data, Meta and Microsoft's free cash flow is expected to be negative after considering shareholder returns, while Alphabet is expected to be roughly flat.

At the same time, many companies are increasingly turning to debt and off-balance sheet financing instruments to fund their expenses, which poses a risk to themselves. For example, Meta recently issued $30 billion in bonds for the largest public offering of high-rated corporate bonds this year, in addition to arranging about $30 billion in private financing packages.

Fulton Breakefield Broenniman research director Michael Bailey said that a shift from a capital-light model to a capital-intensive business model may result in lower valuations.

“A more capital-intensive business is likely to experience a more pronounced boom-bust cycle,” he said. “Investors usually pay a lower valuation for this.”

Since the tech giants account for about one-third of the market capitalization-weighted S&P 500 index, a reduction in the valuation multiplier will almost certainly put significant pressure on the index. All of this highlights that investors are in uncharted territory when it comes to AI spending. Never before has the world's largest and most successful company decided to invest so much cash into a promising yet unproven technology.

Callodine's Morrow said, “Historically, these companies didn't really need to compete with each other. In fairly oligopolistic or monopolistic market niches, they each have their own territory and reaped huge profits in low-capital-intensive businesses. Now, they are in some kind of competition with their different capital-intensive AI business models.” “Facing uncertain results under very high valuation multiples, I think this is a risk that the market must deal with.”