Capital investment spending by U.S. tech giants is growing much faster than their revenue. Their stock values, long favored by investors, are beginning to suffer.
And for 200 billion dollars more: this is the amount that Amazon intends to invest in AI in 2026, the latest announcement in the series that constitutes the spending madness of American big tech around this technology. The online commerce and cloud giant is thus (no doubt momentarily) ahead of its colleagues, who are also not very careful about spending.
“They are the pickaxe sellers of the Gold Rush. As during the cloud turning point, they are financially crushing all competition. They alone invest more than the entire American energy sector”, deciphers Jules Brochard, consultant and researcher at the consulting firm Square Management.
Google, for its part, plans to spend a little less than $200 billion this year (more than double last year, already a record year), while Meta is counting on a sum between $100 and $150 billion. Microsoft and Oracle close the loop with expenditures estimated above $100 billion and around $50 billion respectively. Apple, lagging behind in AI, plans to invest “only” 14 billion.
700 billion dollars of investment in 2026
The sums are not only enormous: they continue to increase. Thus, the spending of the five American AI giants (Amazon, Microsoft, Google, Meta and Oracle) in 2026 was estimated at 600 billion at the beginning of January. A sum today increased to nearly 700 just for the first four (without Oracle, therefore). A dizzying 60% increase compared to the previous year.
These record investments are mainly devoted to the continued increase in computer computing power, which involves the construction of giant data centers and the purchase of hundreds of thousands of graphics cards acquired from Nvidia and AMD. All in order to train and then run the large language models, pillars of generative AI today marketed in businesses under the umbrella of agentic AI.
A risk of the bubble bursting
On the one hand, these all-out investments have a reassuring aspect for those who worry about the formation of an AI bubble, insofar as they signal great confidence on the part of the tech giants, who clearly have faith in this technology and are convinced that it will keep its promises.
But on the other hand, by straining their finances, they also constitute a risk for these companies if the gains from AI do not materialize quickly. The stock market reacted badly to Amazon’s announcement. “-14% the next day: obviously, Wall Street does not like it, but we must separate the stock market from the real economy. These investments are above all there to secure real activity, but at a lower margin.”, estimates Jules Brochard.
However, the capital investment expenditure of these companies is now growing much faster than their turnover. “The recent publication of Oracle’s results concretely illustrates the financing risk within the artificial intelligence ecosystem. The company is sharply accelerating its investment spending and its debt, while the growth in its turnover is disappointing. This configuration underlines that the scale of investments presupposes rapid and sustainable monetization of AI uses. In the absence of this monetization, the risk materializes first and foremost on access to financing,” analyzes Laurent Chaudeurge, member of the investment committee of BDL Capital Management, an independent investment house, in a recent note. The reasoning is based on Oracle, but it applies to all the giants racing headlong into the AI race. The risk is to see the debt rise, with greater difficulty in borrowing.
“Officially, Oracle displays a net debt to EBITDA ratio close to 3x. But by integrating rental leases and other economic commitments, this ratio exceeds 13x. This development is already reflected in the credit market, with a marked tension in the CDS. The company’s ability to sustainably finance its investments thus becomes more dependent on the validation of cash generation hypotheses.”
However, launched into an arms race, the AI ecosystem needs continuous funding to go the distance. Building the necessary infrastructure and maintaining it requires stable and predictable capital flows. If the AI market does not accelerate quickly enough to justify the investments made, access to financing risks being restricted, potentially generating a chain reaction that could burst the famous bubble.
Big tech faces difficult financial choices
If these companies have long been comfortable cash machines, the unprecedented amount of these investments will undoubtedly force their managers to choose between reducing returns for shareholders, drawing on their cash reserves or relying more than expected on the bond and stock markets. It is undoubtedly this anticipation on the part of the financial markets which explains their recent setbacks on the stock market.
Amazon has thus signaled that it could soon seek to raise new capital through debt or shares. Its share price closed the day down 5.6% following the announcement. Its planned capital spending of $200 billion this year is expected to exceed its operating cash flow, estimated at $180 billion, according to S&P Capital IQ estimates. The company has also slashed its workforce, laying off 30,000 people, or 9% of its staff.
To obtain liquidity, Meta for its part issued $30 billion in bonds at the end of 2025, while Google announced the issue of 100-year bonds on the British markets, a very rare occurrence in the world of tech. The Californian online search and cloud giant also plans to raise $15 billion in the United States.
Added to all this is a final risk highlighted by American investor Michael Burry: underestimated depreciation. An analysis shared by Laurent Chaudeurge in his note. “Amazon recently highlighted technology risk. The company reduced the depreciation period of certain data center assets by one year to reflect potentially faster obsolescence, linked to the acceleration of technological innovation. This decision directly affects the anticipated economic life of AI infrastructure.
Faster obsolescence reduces visibility into future cash flows, shortens the profitability horizon of assets and increases the need for reinvestment. In an environment where competition on AI chips is intensifying and where Nvidia is no longer the only competitive solution, the economic profitability of the infrastructures deployed today becomes more uncertain.
Markets that have become very dependent on AI
Add to this the scheduled expiration of a US law that allows companies to immediately deduct all eligible investment expenses related to their activities, and we obtain a significant cocktail of risks for the years to come.
However, the American economy has become extremely dependent on AI. The American equity markets are heavily concentrated on this theme. The ten main stocks directly exposed to AI constitute almost 40% of the capitalization of the American market, which alone represents around 70% of the MSCI World. In the United States, spending on artificial intelligence has also contributed to around 60% of recent real GDP growth. A bursting of the bubble would therefore have repercussions well beyond American big tech.
