Can hyperscalers justify huge AI capital investments?

This report is from this week’s The Tech Download newsletter. Is it what you see? You can subscribe here.

In earnings season, tech companies are pledging endless spending as they rush to ramp up data center capacity amid the AI ​​boom.

Hyperscalers (including) Amazon, microsoft, meta and alphabetannounced that AI-related capital spending could reach $700 billion this year, more than the GDP of countries such as the United Arab Emirates, Singapore, and Israel.

Investors are feeling anxious. Stocks tumbled last week as more than $1 trillion was wiped from the market capitalization of Big Tech companies amid concerns about the scale of AI spending and questions about profits.

Although there has been some recovery this week, the uncertainty surrounding astronomical AI spending continues to make people nervous.

“2026 will see a total 60% year-over-year increase in committed capital spending,” said Michael Field, chief equity strategist at Morningstar.

“At some point, the bet becomes dichotomous: either demand and monetization will follow and you recoup your spending, or else your business will fail,” he added. “Investors were comfortable when it was a side bet, but they become much less comfortable when the entire business is at risk.”

Hyperscalers’ capital spending levels this year will consume nearly 100% of their operating cash flow, compared with a 10-year average of 40%, according to a note released by UBS this week.

Bob Savage, head of market macro strategy at BNY, said capital spending is less of a concern than where the money is coming from.

“If this increases the net borrowing of mega-cap stocks, it will take away their holdings,” he said. “As Oracle has shown, investors are happy to buy bonds, but the problem is that it reduces free cash flow and puts balance sheets at risk for some.”

early this month oracle The company announced plans to raise between $45 billion and $50 billion in the 2026 calendar year, and Alphabet also plans to raise $20 billion from a sale of U.S. dollar debt, people familiar with the matter told CNBC on Monday.

Despite the market turmoil, many analysts remain broadly bullish on hyperscaler stocks.

“The major data center builders (Amazon, Microsoft, Google) pre-sell all of their capacity before building data centers, so revenue is already positive,” Gil Luria, head of technology research at DA Davidson, told me.

He added that further upside is likely in the future. “We expect to see positive returns realized as the use of[AI]increases exponentially and consumers and businesses are willing to pay more for the value created.”

But Field said the timeline for recouping the huge capital investment is “mostly unknown” at this point. “The estimated useful life of much of this spending, including data centers and chips, could be on the order of three to five years, meaning hyperscalers will need to achieve a significant return on investment by 2030, a very tight schedule.”

Field said clear timelines on payback periods and “credible” strategies for monetization are needed for hyperscalers to allay concerns.

Until that happens, investors will continue to be hesitant about further plans to increase capital spending, he added. This could lead to further market turmoil in the coming months.

Alphabet is returning to the debt market to fund AI development after reporting it could spend $185 billion in capital spending this year.

Elon Musk’s xAI has lost two co-founders in two days, with researchers Jimmy Barr and Tony Wu announcing their departure from the company.

Taipei’s top customs and trade negotiator said in an interview on Sunday that a US proposal to move 40% of Taiwan’s semiconductor supply chain to Taiwan is “impossible.”

Apple suffered its worst day on the stock market Thursday since April after reports of Siri delays surfaced and the company’s news app faced regulatory scrutiny.

Anthropic announced Thursday that it has completed a $30 billion funding round at a post-money valuation of $380 billion. This is the second largest amount of funding for a private engineer in history.

Questions over the region’s digital sovereignty, or lack thereof, have resurfaced as geopolitical tensions between Europe and the United States have come to a boil in recent weeks after President Trump refused to eliminate the use of force to acquire Greenland before withdrawing. The data paints a grim picture.