This article was published in the Fortune 500 Digest newsletter on October 17, 2026. The newsletter summarizes the most important business news of the week and headlines about Fortune 500 companies. Readers can subscribe to receive it every Saturday morning.
AI Infrastructure construction is being carried out on a comprehensive scale, and the costs are extremely high.
According to analyst estimates compiled by Bloomberg, Amazon, Alphabet, Microsoft, and Meta are expected to invest a total of over $720 billion in capital expenditures this year, with the majority of that going towards AI infrastructure. Alphabet's President and Chief Investment Officer, Ruth Porat, confirmed to me this week at the Fortune Most Influential Women Summit that Alphabet intends to "significantly increase" spending in 2027.
Alphabet stated this year that capital expenditures are expected to range between $195 billion and $205 billion. This figure is more than twice the $91.4 billion spent in 2025. When asked whether construction expenditures across the industry have peaked, Porat indicated that, compared to what Alphabet believes AI will ultimately create in economic value, this phase is still in its early stages.
She said, "We believe this is an opportunity that will last for a considerable amount of time."
An increasing amount of debt is also providing funding for the construction of AI. Private credit funds such as Blue Owl, Apollo, and KKR are becoming a growing source of capital: according to some estimates, the outstanding private loans related to companies in AI have exceeded $200 billion. Typically, these loan recipients are special purpose vehicles and smaller-scale neocloud operators, rather than large technology companies with substantial funds. Even Alphabet experienced negative free cash flow for the first time since IPO in 2004 during the second quarter, in order to help fund AI. By the end of June, its long-term debt amounted to about $98 billion, more than quadrupling from the previous year.
Alphabet has the capability to undertake such expensive long-term AI bets – in a sense, for it, the greater risk might be not to make these investments at all. However, many participants with weaker capital are vulnerable, and recent changes may make them even more fragile.
One of the biggest changes is that open-weight models, especially those from China, are increasingly taking business and usage shares away from closed models such as Anthropic, OpenAI, and Google Gemini. The new financial data released by Anthropic this week also shows that its growth seems to have slowed down.
This trend has prompted Chris Wood, the head of global equity strategy at JPMorgan Chase, to assert, according to Bloomberg, that the most likely long-term outcome of the AI craze is massive capital destruction in the United States. Wood believes that cheaper Chinese open-source models are threatening the commercialization of large language models, eroding the economic efficiency of AI expenditures in the US, and claims that China has become a technological peer on a par with the US in the field of AI.
I recently learned from one of China's leading open-weight model developers, Moonshot AI, that 70% of its usage comes from regions outside of China.
I directly asked Porat about Wood's view, that is, a large amount of funds flooding into the AI competition in the United States might be ruined.
Porat replied, "It is illogical to say that we are in a state of limited capacity, given the strong market demand for insight and for the enhancement of business and government capabilities, yet you claim that there is no return to be made from it." She also pointed out that when discussing the trends of open-weight models versus closed models, she believes that companies like AI which can add differentiated value will continue to be willing to be paid by customers, regardless of how the market changes.
Porat has personally experienced previous bubbles: during the Internet bubble, she served as the head of Morgan Stanley's technology investment banking division; during the 2008 financial crisis, she provided advice to then-U.S. Treasury Secretary Hank Paulson. She later became Morgan Stanley's Chief Financial Officer in 2010.
The conclusion she drew from those periods was that companies that use AI to create truly differentiated experiences will be the winners.
A key lesson from financial crises is that you should identify your greatest sources of vulnerability as early as possible and strengthen them, because you won't be able to address them once the crisis occurs. For banks, that is liquidity. Without liquidity, you cannot carry out trading operations. Without liquidity, you will come to a standstill," said Porat.
She said, "Since then, I've often been asked the same question: What is the greatest source of vulnerability today? I believe that one of the most important factors for all industries is business innovation. If someone is connecting with customers in a way that you are not, because they are using AI while you are not, then you are at a disadvantage in the competition."
That's right, she is endorsing her own views, and the notion that "just add more AI and it will be fine" cannot deny the systemic economic risks that over-construction may bring. But what is equally true is that AI will not disappear, and the demand for enterprises to adapt to it will only grow stronger.
At Alphabet, Porat indicates that the company will continue to make substantial investments in AI because it firmly believes that such investments will yield long-term economic returns.
"You must invest for long-term growth," said Porat, "by the time you really need it, it will be too late to say 'Okay, let's start now'."
For more information, you can view the full conversation between me and Google's CIO, who also serves as President Ruth Porat.
