MW A $400 billion bonanza? As AI spending estimates move higher, so do the stakes for investors.

By Britney Nguyen and Emily Bary

Wall Street expects capital expenditures from four big tech companies to exceed $400 billion next year, but those numbers might be conservative, further fueling the debate over AI returns

Wall Street's AI capex expectations for next year are "conservative," Roth analysts said.

Analysts have been ratcheting up their estimates for Big Tech's artificial-intelligence spending, and these sky-high projections may actually be understating the potential for capital-expenditure increases in the near future. That adds fuel to the debate over whether AI will deliver a meaningful return on investments.

Analysts at Roth Capital Partners said in a note on Monday that AI capex expectations for Amazon.com Inc. (AMZN), Meta Platforms Inc. (META), Alphabet Inc.'s $(GOOGL)$ $(GOOG)$ Google and Microsoft Corp. $(MSFT)$ are "conservative" for 2026, regardless of how and when returns on that spending start to appear.

Wall Street is modeling that those megacap companies will spend $350 billion this year and $402 billion next year, representing about $50 billion in incremental spending, the analysts said. The good news, in their view, is that the "intensity" of capex for these companies, measured as a percentage of revenue, could peak in 2026. Ditto for the reinvestment rate, which captures the portion of operating cash flows that goes toward capital spending.

But there's reason for caution as well. Cumulative capex for those four companies rose by $81 billion in 2024 and is expected to jump $115 billion in 2025, the Roth team noted. This suggests that Wall Street's capex estimates for next year might be understating both the amount of AI computing power that will be required as demand grows and the base case for AI inference, or the process of running AI models after training, the analysts said.

Based on Wall Street estimates, Microsoft will see the smallest incremental dollar increase in capex in 2026, at $7 billion, the analysts said, while Google and Amazon will see a $9 billion increase and Meta's spending could jump by $25 billion.

But as operating expenses and capex grow, most hyperscalers are seeing a decline in free cash flow. Amazon, for example, saw negative free cash flow when combining the last two quarters, while Alphabet and Meta have seen declines when looking cumulatively at that period. With the companies ramping capex, trends like those could persist.

Read more: This is the critical detail that could unravel the AI trade: Nobody is paying for it.

But despite the risk there, some on Wall Street are staying optimistic on the spending, bumping up forward price-to-earnings ratios to show that they see future gains from all the spending.

Meta in catch-up mode

The analysts are also concerned that Wall Street is understating Meta's potential depreciation expenses for next year. The company has the "youngest assets" of the megacap companies, in Roth's view, "and is likely playing catch-up in AI build-out." Compared with its Big Tech peers, the company's accumulated depreciation as a percentage of its gross property, plant and equipment assets has not increased, while its spending on servers has seen a strong increase over the past six quarters.

After raising its full-year capex guidance range to between $64 billion and $72 billion earlier this year, Meta again raised the range to between $66 billion and $72 billion in its latest earnings report in July.

Meta's chief financial officer, Susan Li, said capex spending is expected to grow in 2026 "as we continue aggressively pursuing opportunities to bring additional capacity online" for the company's AI efforts and other operations.

What about the returns?

Spending is one part of the equation as companies jockey for positioning in the AI era, but revenue is the other critical component. Large technology companies are pouring money into AI because they foresee big financial returns down the line, but some on Wall Street are skeptical that AI will deliver on its promise in this regard.

"Few of this generation's founders or CEOs will risk a future in which their company fades away by focusing on margins and capital returns as they are passed by new technologies and business models," Citi Research analysts wrote earlier this week. "While both public- and private-market investors have embraced this strategy, they continue to ask from where and in what form these returns will come."

So far, AI has mainly delivered in terms of generating cost savings, according to the Citi team. Companies have found ways to reduce their expenses by employing AI for customer service, coding assistance and "knowledge retrieval" - and those three use cases could together result in $275 billion of annual cost savings, based on conservative estimate inputs, the Citi analysts noted.

What about revenue? The Citi analysts project that AI will contribute to $780 billion in sales by 2030, up from an estimated $43 billion this year. That would make for an almost 80% compound annual growth rate. UBS analysts recently pegged AI's annual revenue potential at $1.5 billion at an unspecified future point, and they said the number compared favorably to the estimated $780 billion that could get spent globally on AI capex from 2022 to the end of 2025.

The Citi team acknowledged some factors that could limit AI's adoption. For one, the stakes are lower when AI is used for customer service or advertising than when it's employed for air-traffic control or medical diagnostics, they noted. So it's possible that AI applications won't be as wide-ranging as some hope.

In addition, there's the economic aspect. "Clearly the most important, the cost to deliver AI is in many cases outstripping the value that companies are getting from that AI," the Citi team noted. "That dynamic is manifesting itself in the tens of billions of dollars in annual losses generated by AI startups."

The analysts added that losses there "are currently being borne by corporate cash flows and venture capital, subsidizing the adoption of AI and masking its true economics." Venture investments certainly aren't slowing down, with the Citi team citing 180% annualized growth in AI investments so far this year.

But overall, they think the tradeoffs could prove worthwhile. "Put simply, we believe AI's impact on innovation and efficiency will yield returns proportionate to the scale of investment," the analysts wrote.

Christine Ji contributed.

-Britney Nguyen -Emily Bary

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

August 27, 2025 12:45 ET (16:45 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.