Software Company Loans Plummet Amid AI Concerns

As Bloomberg news reported Saturday (Jan. 31), investors are concerned that artificial intelligence (AI) advances, such as the coding abilities of Anthropic’s Claude model, will make many software offerings redundant. 

“A storm has hit the loan market,” Scott Macklin, head of U.S. leveraged finance at asset manager Obra Capital, told Bloomberg.

“The heaviest calendar in months, largely repricing-driven but still overwhelming, has collided with mounting existential questions around software business models as AI reshapes the sector, which is the single largest in loans. Layer on an unusually heavy flow of BWICs [bids wanted in competition] and you have a full blown ‘loan-ageddon.’”

Software is one of the biggest components of the leveraged loan market, accounting for 12% of the credits in the Bloomberg U.S. Leveraged Loan Index. Software debt in collateralized loan obligations, which are bonds backed by portfolios of leveraged loans, has notched the worst total returns so far this year versus all sectors, according to data compiled by Nomura.

Bloomberg noted that the selloff in the software loan market is in sharp contrast to the remainder of the levered loan space, where sales surged after President Donald Trump halted his tariff threats against Greenland. Leveraged loan sales hit a new weekly record in Europe.

Anthropic last week introduced a new plugin for its AI collaborator Cowork, letting users tailor Claude to specific job functions. For instance, a sales plugin could connect Claude to the user’s customer relationship management (CRM) system and knowledge base, let it learn the user’s sales process, and offer commands for prospect research, sales follow-ups and other tasks.

In other AI news, PYMNTS wrote last week about new research showing that most people interact with conversational AI through a single dominant platform and a decreasing set of mobile assistants shaped by operating system defaults.

Research by PYMNTS Intelligence shows that — among consumers who used a dedicated AI platform for at least one task — 83% used OpenAI’s ChatGPT. That’s compared to 48% who reported using Google’s Gemini, and 30% who said they used Microsoft’s Copilot. 

“The gap matters because it reflects where usage begins,” PYMNTS wrote. “ChatGPT reaches most consumers at the point of first interaction, while Gemini and Copilot reach smaller subsets. Rather than trying several comparable tools, most users appear to start with one platform and stop there.”

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.