AI Ecosystem Concentration Risk Mirrors Pre-2008 Mortgage Market, Sona Analysis Warns

The AI buildout has become so interconnected that a single point of failure could ripple across the multitrillion-dollar sector, according to a new analysis from London-based Sona Asset Management highlighted in the Financial Times. The authors compared the current landscape to the mortgage market before the 2008 financial crisis, citing "opaque and concentrated exposures" and counterparties linked in complex ways. Sona mapped an AI ecosystem of 255 public companies with a combined market cap of $50 trillion — more than double five years ago — and nearly $6 trillion in debt. Capital circulates among a handful of names, with smaller firms heavily dependent on a few large customers. CoreWeave draws about 67% of revenue from Microsoft; Applied Digital gets 56% from Oracle and 30% from CoreWeave. The analysis follows S&P's warning on weakening hyperscaler credit quality and arrives as borrowing costs rise across the sector. Sona identified the highest risk "one ring out from the core" — neoclouds and data center platforms with the highest leverage and weakest cash flows.