Paolo Saw Four Cracks in the AI Bubble — Guess Who’s Still Pouring Billions In 🚀

Tether CEO Paolo Ardoino has identified four "structural mismatches" inside the artificial intelligence infrastructure race pursued by the world's largest technology companies, warning that spending has raced ahead of the underlying economics. In a July 4 post on X, Ardoino wrote: "AI big tech subsidizes compute to increase user count building expensive infrastructure / capex subject to fast decay (3/5 years). – Token price mismatch. – Profitability timeline mismatch. – Cost of capital maturity mismatch. – Open-source AI taking growing chunks of revenues."

According to Ardoino's framing, the first mismatch sits in pricing: AI providers are charging customers less than the real cost of delivering compute, effectively subsidising usage to win user count and making growth appear stronger than the underlying business model. If prices are later raised, demand could soften; if they stay low, margins remain compressed.

The second mismatch is timing. Big Tech is front-loading capital spending on data centres, GPUs and long-dated power contracts, while the revenue payback from those assets may arrive over a far longer horizon. That gap between present capex and future commercial returns places direct pressure on the sector to demonstrate that AI can become a durable, recurring source of income rather than a one-cycle investment theme.

The third mismatch is the depreciation curve. AI chips typically become outdated within three to five years, yet the debt and equity used to finance the underlying infrastructure are often priced against payback assumptions that stretch much longer. If demand slows or compute prices fall before the first generation of hardware is fully paid down, the unit economics of the build-out become harder to defend.

The fourth mismatch is competitive. Open-source AI models are improving quickly, Ardoino said, and could erode the pricing power of commercial AI providers if free or low-cost alternatives reach sufficient quality. Such a shift would complicate the recovery of the hundreds of billions of dollars being committed to AI capacity and weigh on the revenue assumptions that have supported elevated valuations across the sector.