Tether CEO Paolo Ardoino questions AI compute subsidy model

Tether CEO Paolo Ardoino warned that Big Tech’s AI infrastructure race may rest on weak economics, arguing that subsidized AI computing, delayed profit realization, short hardware lifespans and rising open-source competition have created four structural mismatches in the market. He said some companies appear to be pricing AI services below their true cost to gain users, a strategy that can support headline growth while leaving margins under pressure or risking weaker demand if prices rise later. Ardoino also pointed to the gap between massive upfront spending on data centers, GPUs and power capacity and the slower timetable for commercial returns, as well as the risk that AI chips can become outdated within 3 to 5 years even as the debt and equity financing those investments assume longer payback periods. His comments come as investors debate whether AI revenue can justify the scale of current spending, with JPMorgan projecting global AI-related spending could reach $5.5 trillion by 2030, Alphabet, Amazon, Meta and Microsoft expected to spend up to $720 billion this year, and Morgan Stanley estimating nearly $3 trillion in AI infrastructure investment could flow through the economy by 2028. The warning also lands amid broader market concern, with Chinese hedge funds Wealspring Asset and Shanghai Banxia Investment Management Center describing AI stocks as bubble-like and the Bank of England warning in October 2025 that AI-related valuations were nearing dot-com-era levels and that infrastructure buildout may require trillions of dollars, some financed by debt.