The AI infrastructure buildout is in full swing—and it is extraordinarily expensive. Analyst estimates compiled by Bloomberg project that Amazon, Alphabet, Microsoft, and Meta will collectively spend more than $720 billion in capital expenditures this year, with most of it going to AI infrastructure. Speaking at the Fortune Most Powerful Women Summit this week, Alphabet President and Chief Investment Officer Ruth Porat confirmed that Alphabet intends to spend "significantly more" in 2027.

Alphabet has said it expects capital expenditures of $195 billion to $205 billion this year—more than double the $91.4 billion it spent in 2025. Asked whether industry spending on the buildout has peaked, Porat said the investment cycle remains in its early phases relative to the eventual economic value Alphabet believes AI can generate. "We do see this as an opportunity that extends for quite some time," she said.

Debt, too, is playing a growing role in financing the buildout. credit funds such as Blue Owl, Apollo, and KKR have become an increasingly important funding source, with outstanding private loans to AI-related companies climbing above $200 billion by some estimates, according to the Basel-based Bank for International Settlements. Much of that lending flows to special purpose vehicles—standalone financing entities—and to smaller "neocloud" operators, the specialist firms that rent out computing capacity for AI workloads, rather than to deep-pocketed Big Tech companies. Even Alphabet posted negative free cash flow in the second quarter for the first time since its 2004 initial public offering to help fund its AI push, and its long-term debt stood at roughly $98 billion at the end of June, more than four times the level of a year earlier.

Alphabet can absorb such costly long-term bets—and arguably faces a bigger risk in not making them. But many less well-capitalized players are vulnerable, and recent shifts in the competitive landscape may be making them more so.

One of the most significant shifts: open-weight models—whose underlying parameters are published so that anyone can download, run, and adapt them—particularly those developed in China, are increasingly taking business and usage away from closed models such as Anthropic, OpenAI, and Google's Gemini. New financial figures released by Anthropic this week also suggest the company's growth may have flattened.

That trend led Jefferies' global head of equity strategy, Chris Wood, to declare that the most likely long-term outcome of the AI boom is massive capital destruction in the United States, according to Bloomberg. Wood argued that cheaper Chinese open-source models threaten to commoditize large language models, eroding the economics behind America's AI spending, and that China "has become a technological peer to the U.S. in AI."

The global reach of Chinese open-weight models was underscored in a recent conversation with Moonshot AI, one of China's leading developers in the field: 70% of Moonshot's usage comes from outside China.

Asked directly about Wood's thesis that much of the money pouring into the U.S. AI race could be destroyed, Porat pushed back.

"It's illogical to say we're in a capacity-constrained position because there's such strong demand for the insight in the ability to improve business and government, and yet you can't drive a return on it," she said. On the contest between open-weight and closed models, Porat added that AI companies that add differentiated value will find customers paying for their services regardless.

Porat has had a front-row seat to previous bubbles, first as head of Morgan Stanley's technology banking team during the dot-com era, and later during the 2008 financial crisis, when she helped advise then-U.S. Treasury Secretary Hank Paulson. She went on to become Morgan Stanley's chief financial officer in 2010.

Her takeaway from those periods is that companies using AI to build truly differentiated experiences will be the ones that win.

"One of the key lessons from the financial crisis was that you should identify your greatest source of vulnerability early and fortify against it, because you can't address it in the moment. For banks, it was liquidity. Without liquidity, you couldn't run a trading operation. Without liquidity, you shut down," Porat said.

"I've been asked that question frequently since: What is the greatest source of vulnerability today? I think for all of us across industry, one of the most important elements is innovation in your business. If anyone is connecting with your customers in ways you are not, because they are using AI and you are not, you're at a competitive disadvantage."

The column acknowledged that Porat was, in effect, advocating her own company's position, and that the "just add more AI" view does not erase the potential systemic economic risk of overbuilding. It also noted that AI is not going away, and that the pressure on companies to adapt will only increase.

At Alphabet, Porat indicated the company will keep investing heavily in AI out of conviction in the long-term economic upside.

"You must invest for long term growth," she said. "By the time you need it, it's too late to say, OK, now we're going to start."

Porat's remarks came in an essay published in the Oct. 17, 2026 edition of Fortune's Fortune 500 Digest newsletter, which rounds up the week's most important business headlines and coverage of Fortune 500 companies. Fortune's full conversation with Porat is also available, with video on YouTube.

This article was originally published on Fortune.com.