Will generative AI kill software? The topic has gained relevance with the fall in the shares of companies in the sector this year. The iShares Expanded Tech-Software ETF (IGV), a proxy for software companies, has accumulated a 26.5% drop this year.
But the fears are not limited to the stock market. Optimism is also no longer the same in the private credit market, which has accumulated high exposure to the sector in recent years.
With many of these loans originated in the early 2010s nearing their maturity dates, large private credit managers are reportedly underestimating their true exposure to software companies in their official filings, according to a report in The Wall Street Journal (WSJ).
The analysis was based on the credit portfolios of Blackstone, Blue Owl Credit Income, Ares Capital, and Apollo Debt Solutions — firms that, together, manage more than US$2 trillion.
According to the WSJ, the discrepancy is close to 25% compared to the published figures. The biggest difference is in the values reported by Blue Owl Credit Income, which says that 11.6% of its portfolio is exposed to software companies. The analysis indicates that this figure is almost double.
Blackstone, on the other hand, has the highest concentration, holding approximately 33% of the sector's debt, 7.5% more than the company disclosed.
For the analysis, the WSJ used industry classifications from data provider PitchBook and its own analysis to identify software companies within private credit portfolios.
Sources indicated that the way asset managers classify each company by sector is not uniform. A software company that provides services to the healthcare sector, for example, could be placed in that category due to its dependence on the industry.
The high exposure to software companies has been accompanied by strong growth in credit funds , which has also drawn attention to the quality of the debt originated.
Since 2021, the large credit funds of these asset managers have jumped from approximately US$20 billion in 2021 to nearly US$175 billion by the end of 2025.
A report by Standard & Poor's estimates that 20% of the sector's debt is due in the next two years and that more "fragile" or "easily" disruptive companies will have difficulty refinancing their debt.
For companies with loans maturing in four years or more, which represent about 40% of the sector's total debt, the rating agency also does not project an easy scenario.
"The software sector landscape may be significantly different by then, which could make refinancing difficult for traditional companies if they don't make relevant changes to their service offerings," says S&P.
According to S&P, this concern has been present in the market since last year, but it has been highlighted by the launch of AI tools with the potential to threaten software companies.
One of them is Claude Cowork, from Anthropic, which is capable of automating various stages of intellectual tasks directly on users' computers. On the day of its launch last week, the iShares Expanded Tech-Software ETF fell 4.3%. For the week, the drop was 7.3%.
Despite growing concern, S&P believes that AI disruption in the industry will be "complex and nuanced."
For less regulated companies, the agency estimates that AI tools can be implemented more easily, and legacy software that supports these businesses tends to face a greater risk of replacement.
In regulated sectors, which handle sensitive data and are subject to legal frameworks, S&P expects AI adoption to be more cautious and slower.
According to the credit rating agency, software companies continue to show healthy growth and are receiving more upgrades than downgrades in their credit ratings — which eliminates the risk of immediate collapse.
“However, it is important to highlight that our credit estimates are based on historical performance. We recognize that the accelerated pace of artificial intelligence development will likely reshape the competitive dynamics in the direct credit market focused on the software sector,” says S&P.