Two of the biggest names in collateralized loan obligations are now treating software companies like risky bets.
Blackstone and Guggenheim are actively trimming software sector exposure from new CLO deals. The concern: artificial intelligence is about to upend the business models of companies they’ve been lending to for years.
Software and services make up roughly 15% of collateral in US syndicated CLOs, with software loans alone accounting for about 12%. In dollar terms, that’s approximately $235 billion in software loans, or 16% of the $1.5 trillion leveraged loan index.
Blackstone President Jon Gray points to an AI risk “traffic light” system-a framework requiring AI risk analysis in every investment decision, with faster exit options on vulnerable assets.
Guggenheim’s Rob Zable says most managers are simply not equipped to handle the speed at which AI is reshaping software.
The de-risking started ahead of a March 17, 2026 report, with managers executing secondary sales and becoming far more selective in primary markets.
Pricing data shows certain software loans and bonds sold at 89 to 98 cents on the dollar in February and March 2026-down from earlier premiums.
Global CLO loan supply is projected to drop about 25% in 2026, to around $150 billion, per JPMorgan.
Firms that develop credible AI risk assessment frameworks, like Blackstone’s traffic light system, could gain an edge marketing new CLOs to institutional investors.