By Sam Goldfarb
The threat that artificial intelligence tools will fuel a software apocalypse has rattled stocks, triggered record withdrawals from private-debt funds and stirred fears of a new type of credit crisis.
But a key market is sending a more mixed message: Not all software companies are equally endangered. While prices of software-company loans have fallen sharply on average since late January, a Wall Street Journal analysis of more than 100 loans showed wide variations in price moves, with parts of the sector hit much harder than others.
The Journal's analysis focused on first-lien software-company loans in the Morningstar LSTA US Leveraged Loan Index, which tracks loans that are originated by banks and broadly distributed to investors. Software makes up about 13% of that index and an even larger share of loans in private-credit portfolios, the result of a wave of private-equity buyouts that crested a few years ago.
Unlike those held by private-credit funds, the loans that were originated by banks are actively traded, offering daily insight into how investors are assessing their risk of default. Here's a look at what they're saying:
A mixed picture
To observe the hit from AI fears, it can be helpful to split the sector into four main categories:
-- "Vertical" software that serves specific industries -- "Horizontal" software that serves a range of businesses -- Cybersecurity and identity-verification software -- Software-engineering software
Of these groups, vertical software performed the best through Thursday, with 40 loans falling an average of 4.2 cents on the dollar since Jan. 20, according to the Journal's analysis of S&P Global Price Viewer data.
Next up was cybersecurity software, with 19 loans falling an average of 5.3 cents. Lagging behind were horizontal software, with 33 loans falling an average of 8.8 cents, and software-engineering software, with six loans falling an average of 16.3 cents on the dollar.
A loan's starting point -- an indication of whether there were pre-existing concerns about a company -- also mattered. Loans that traded above 95 cents on the dollar on Jan. 20 have since fallen an average of 4.5 cents, while loans that traded below that threshold have dropped an average of 10.5 cents on the dollar.
Vertical software
AI has made it much easier to make software, potentially lowering barriers to entry into the industry. As a result, investors are now seeking "defensive moats," asking which companies can withstand competition from AI giants like Anthropic, new AI-native startups and even customers who could use AI tools to make their own tailored software.
Though hardly invincible, vertical software companies carry some inherent advantages, having carved out niches in narrow slices of the economy. Some also help customers, such as banks or law firms, handle sensitive data that would cause major problems if compromised, making those businesses loath to adopt new systems.
Examples of vertical software companies include CCC Intelligent Solutions, a publicly traded company that is deeply embedded in the auto-insurance claims process; Relativity, a maker of legal data software; and RealPage, which serves operators of multifamily rental units.
Horizontal software
Horizontal software companies don't serve a particular industry. They can, however, still handle sensitive data or provide an unusual service.
In general, investors are looking for software products that not only "inform workflow, but control that workflow," said Ali Bendarkawi, senior technology analyst at AllianceBernstein. Another key, he said, is that failure of the software "would create real business disruption."
So far, investors have looked relatively favorably on tax compliance software makers like Avalara; companies like SUSE that run large-scale computing infrastructure; and UKG, which makes human-capital management software that handles tasks like payroll processing.
They have been much more concerned about businesses like Qlik and Qualtrics, which help companies analyze and visualize data, tasks that some investors think could become commodified. Document-generation software -- made by the likes of Conga, which automates sales-related paperwork -- has also been deemed vulnerable.
Cybersecurity software
The fate of cybersecurity software is a source of debate among investors. Anthropic's recent announcement that it was giving only limited access to its new AI model, Mythos, because of concerns it could be used to facilitate cyberattacks has added to fears that existing security software could become obsolete.
Still, many investors are optimistic that more attacks will only increase demand for existing cybersecurity products, which can be enhanced with new AI technology.
As a result, loans from companies such as Proofpoint, which specializes in email security, and Gen Digital, the maker of Norton 360, have posted only modest declines so far.
In cybersecurity, "you have a subindustry within software that has a very significant tailwind at its back," said Scott Caraher, head of senior loans at Nuveen.
Software-engineering software
Given the ability of new AI tools to write and analyze code, it's no surprise that software for software engineers has been particularly hard hit by the recent selloff.
Loans from Idera, SmartBear and Perforce were declining even before late January. They then fell sharply afterward to below 80 cents on the dollar, signifying major concerns about the risk of default.
Still, some investors caution that they are only making educated guesses at this point -- and that their thinking could shift quickly.
The threat posed by AI is still somewhat "vague in everyone's mind," said Joe Lynch, head of noninvestment grade credit at Neuberger Berman. There's uncertainty "over what AI can actually do and what it means for a lot of different businesses."
Write to Sam Goldfarb at sam.goldfarb@wsj.com
(END) Dow Jones Newswires
April 26, 2026 05:30 ET (09:30 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments