In an opinion piece shared on X today, Tether CEO Paolo Ardoino warned that Big Tech’s efforts to build AI data centers rely on subsidized computing and hardware that will lose value within three to five years. He believes four structural misalignments put the sector at risk.
The warning comes as hyperscalers are investing record amounts in infrastructure without seeing a clear return on investment.
4 problems to solve
Ardoino said AI companies are subsidizing computing to attract more users and investing heavily in infrastructure that only lasts three to five years, he said.
He cited four main issues.
- Token price does not match cost.
- Profitability timelines don’t match investments.
- The maturity of capital does not correspond to the useful life of the asset.
- Finally, open source AI can reduce returns.
Spending is huge and still growing. In the interim outlook announced on June 24th, JP Morgan raises its estimate for global AI-related capital spending by 2030 from $5.1 trillion to $5.5 trillion, and projects AI-related debt financing to reach $4.1 trillion.
The bank projects hyperscaler capital spending to reach $650 billion this year and exceed $1.1 trillion by 2027. Microsoft alone plans to spend about $190 billion in 2026, a 61% increase from the previous year.
Goldman Sachs estimates that Meta, Microsoft, Amazon, and Alphabet will spend a combined $5.3 trillion on capital spending between 2025 and 2030. This year, these four companies will spend $725 billion, 77% more than last year’s $410 billion.
According to reports, Alphabet also raised $84.75 billion for AI infrastructure, which is said to be the largest U.S. equity capital raise in history.
These large investments have not yet yielded returns.
Ardoino’s concerns about profitability reflect growing uncertainty about whether this spending will actually pay for itself. The average company will spend $11.5 million on AI this year, but most companies can’t demonstrate a clear return on investment. data from Bureau of Economic Analysis It also shows that the growth rate of the information sector slowed from 3.2% in Q3 2025 to 1.5% in Q1 2026.
His warning about open source AI taking a bigger share of revenue aligns with a growing trend for many months. Companies that previously encouraged employees to maximize their use of AI (also known as “tokenmaxxing”) are now scaling back as CFOs question rising API costs.
Amazon eliminated internal leaderboards that tracked employee AI usage, Uber burned through its 2026 AI coding budget in just four months and set a cap of $1,500 per employee per month, and Meta issued a warning to nearly 6,000 employees about rapidly increasing costs.
IDC predicts that by 2028, 70% of major AI adopters will be using multiple models rather than relying on a single vendor, potentially erupting in price competition.
Regulators are also concerned. The Bank for International Settlements warned in a report. annual report He expressed the view that a sharp decline in AI investment could have a more negative impact on global stock markets than past recessions.
The bank cited AI as one of three major risks to the economy. Zhang Tao“The speed of adjustment could be much faster than in previous episodes of banking crisis,” said BIS Asia-Pacific chief representative.
Not everyone is so pessimistic. Wedbush analyst Dan Ives said the buildup is an “arms race” and no major company can afford to back out. He believes the sector will start generating profits within the next six to 12 months.
JPMorgan also expects profits to remain strong, with operating cash flow expected to exceed $900 billion by 2027.
Thomas Hayes, chairman of Great Hill Capital, took a more balanced view, saying one or more of the major companies could announce capital spending cuts in their next earnings report. For now, the upcoming earnings season will be important. If any of the big spenders cut spending, as Mr. Hayes predicted, that would be the first real test of the problems Mr. Ardoino identified.
