Software Sell-Off on AI Disruption Is Indiscriminate, Before Defensive Business Models Emerge; Levfin Borrower Management AI Commentary Centers Around Opportunities Than Threats But Limited Quantifiable Impact Presented

Since the late 2022 release of ChatGPT, large language models and other advances in AI have upended some business models, vastly reduced the cost of producing code, and reshaped certain jobs and processes that proved most vulnerable to the innovation. The gargantuan scale of investment, bullish guidance from the AI tech leaders, and accelerating pace of progress has grabbed the imagination of investors in the leveraged finance space, leading to a reassessment of near-term and long-term risks to their exposures.

This has led to substantial drops in debt prices, often hurting long established business models across industries which had little in common previously. This selloff has accelerated, with equity and debt markets falling precipitously two weeks ago, triggered by a new legal tool from Anthropic’s Claude large language model.
 

The price action has been reflecting a growing fear and evidence, with the narrative that AI could replace certain products, reduce moats of others, and change pricing models. Meanwhile, management teams are hard at work figuring out how and where to implement AI to drive new business, reduce costs and defend from competition.

The leveraged finance universe is particularly in the spotlight, as the software sector has been popular among private equity investors, fetching large purchase multiples underpinned by fast growth, wide margins and a belief in the defensive nature of the recurring revenue. On the syndicated side, information technology represents around 17% of the U.S. leveraged loan index – heavily weighted towards the low B-rated, highly leveraged buyout debt. These buyouts were mostly funded with high leverage, reliant earnings expansion to rightsize the capital structure. However, amid recent disruption, heavy debt loads present constraints on ability to invest and pivot business models, with limited room to maneuver. Meanwhile sponsors that invested into the businesses often at high teen multiples will have to evaluate their approach to the portfolio companies as valuations in the public markets have plummeted.

The impact of AI goes well beyond software companies, impacting services business and other industries. At Octus we have analyzed more than 1,000 private and public transcript calls and flagged the ones where “AI” related keywords were mentioned. Figures, charts and commentary shared below represent companies where “AI” related keywords were mentioned and where the commentary relates to the third-quarter 2025 calendar period, unless otherwise stated.

Note: Approximately 68% of the companies covered in this report are U.S. based, 18% have operations both in the U.S. and Europe and the remaining 14% are European companies. The distribution of companies sector-wise was as follows: 25% are in the industrials sector, 25% – IT, 14% – Consumer Discretionary, 11% – Healthcare, 10% – Financials and the remaining 15% spread out across the remaining sectors. There were around 240 unique companies in the sample size used for this report.
 

• The recent selloff has been concentrated in issuers whose business models rely on labor-intensive, rules-based, or intermediary workflows across both services and software. Names spanning business process outsourcing (BPO), document capture, legal, HR, tax, financial, DevOps, and workflow management have been repriced as AI increasingly substitutes human execution with limited regulatory, legal, or financial downside.

• While some sectors such as cybersecurity and financial software retain partial structural protections, the market is increasingly differentiating between platforms that own proprietary data, regulated risk, or embedded mission-critical workflows and those whose value is derived primarily from scalable, rules-based execution, content generation, or information intermediation that can be replicated or automated by AI.

• Based on company management commentary, at the end of the third quarter of 2025, the majority of AI use cases still fell under the “General/Strategic” category, which were characterized by abstract ideas about potential future value rather than immediate, evidence-based results.

• Management teams have been vocal about the potential and uses of AI, though only roughly 28% of companies analyzed in this report could provide a tangible use case paired with a quantifiable impact on business metrics, growth or margins.
   
• Operations and customer support were the leading areas for practical AI application, specifically in automated code writing, document review, and the deployment of AI chatbots and agents.

Artificial intelligence was one of the key focal points for investors in 2025, garnering buzz as the year progressed. According to FactSet, in the third quarter of 2025, 306 companies out of the S&P 500 had mentioned the word “AI” in their earnings transcripts – the highest number over the last 10 years. However, so far it has been questionable, or, on a broader scale at the very least, not well documented what is the actual impact that AI is having on businesses.

Despite limited visibility into near-term financial impacts, market pricing increasingly reflects AI-related disruption risk. Across our sponsored coverage where we have done deep dive analysis, loan pricing continues to reflect differentiated but persistent AI-related risk.

Internet Brands, the largest issuer in the cohort with strong liquidity and dominant vertical positions, has its first-lien loan due 2031 trading in the low- to mid-80s, indicating a persistent discount tied to long-term structural and AI-driven uncertainty. Foundever, the third-largest global CX BPO provider, trades in the 30s despite adequate liquidity and no near-term maturities, reflecting market concern around labor displacement, pricing compression, and the rising likelihood of future balance-sheet intervention. More recently, Cornerstone OnDemand and Tungsten Automation (Kofax) have each seen their first-lien loans trade down 8–10 points during the latest software-led selloff, as investors price in execution risk from AI adoption, product commoditization, and limited financial flexibility versus better-capitalized peers.

Finastra, a global financial software provider serving retail and corporate banks, has seen its first-lien term loan due 2032 trade in the low 90s following a downbeat earnings release, amid a market environment that is actively pricing in an AI-related risk premium. ION Platform, the recently formed combination of ION Analytics, ION Markets, and ION Corporates, is a global provider of mission-critical capital markets, trading, treasury, and analytics software, with its loans trading down from around 98 to the mid 80s during the latest software-led sell-off. Notably, both companies carry meaningfully higher leverage and weak growth momentum, with Finastra previously struggling to refinance in the syndicated loan market and turning to private credit, while ION’s capital structure reflects significant reliance on underwritten cost synergies. We also note that certain offerings, including document generation and compliance solutions, may be more exposed to AI-driven disruption. In addition, there are concerns that both companies maintain legacy solutions within their product portfolios.

Request our PCA for more details. See our special situation watchlist for profiles on the deals.

Price changes across a select group of names across impacted verticals looked as follows:
 

 

However, the recent selloff has been much broader based with U.S. leveraged loan decliners of over 10% since Jan. 23 as follows, with a long tail of deals falling less than 10%.
 

Looking specifically at software loan deals, the top U.S. decliners are below:
 

The negative sentiment in public markets towards software from AI-driven disruption has grown. Shares of large software companies such as Salesforce, Adobe, and ServiceNow are down more than 30% since early last year, while an index of small and mid-cap software stocks has declined by over 20%. The selloff has further intensified over the last few weeks, driven by renewed concerns around AI-driven disintermediation and weakening growth visibility, as recent advances in AI development and workflow tools lower the cost of replicating core functions across software and services.

At the same time, investor skepticism has grown that large and accelerating AI investments may not translate into incremental revenue or sustainable growth, heightening concerns around pricing power, demand durability, and long-term relevance in AI-exposed business models. The S&P 500 Software & Services Index (.SPLRCIS) has declined nearly 13% over six consecutive sessions and is now down approximately 26% from its October peak, underscoring the severity and speed of the sectorwide repricing.

The share price pullback should lead investors to reevaluate loan-to-value ratios of sponsored investments, with the buyouts having been mostly done at double digits, reflecting fast growth, robust margins, and defensive terminal value, all of which are now in question. Below is an overview of software buyout multiples, which have mostly been done in low single-B rating categories, carrying significant leverage. For individual company details, click HERE.
 

The recent announcement in the first week of February 2026 by Anthropic further exacerbated market declines, and in particular software companies, as the company announced new upgrades to their Claude Cowork product, introducing 11 new plug-ins. The media’s narrative focused almost exclusively on the legal upgrade and the shockwaves it sent amongst public companies that offer legal services and products, such as Thomson Reuters and LegalZoom, which saw double-digit share price declines in a single day after Anthropic’s announcement.

However, legal was only one of the 11 plugins that Anthropic unveiled. Each of the plugins targeted a particular niche – sales, finance, data, marketing, customer support, productivity, product management and biology research amongst a few others. While it’s always hard to know exactly why share prices decline in the short term, it seems that companies’ share prices in these other fields were also affected. In the finance and data analytics space, S&P Global and FactSet stock prices dropped by double-digits. In the sales, marketing and customer support fields, Salesforce, HubSpot and ServiceNow experienced sharp declines. In the productivity and product management front, Atlassian and Monday.com were amongst the ones that got hit.

Almost every other company, executive and management team have been throwing out buzzwords and claiming the benefits that their companies are reaping from AI, but they are providing very little actual tangible evidence. An analysis done earlier in the year by the Financial Times newspaper analyzing transcripts of the S&P 500 companies has shown that large public company management teams were eager to mention the potential opportunities of AI, but very few were able to provide tangible examples of how AI was benefiting their business in particular. AI capabilities and related investment have largely become table stakes, serving more as a defensive necessity than a source of sustained competitive advantage.
 

Based on transcript commentary and observed market behavior, AI-related impacts currently appear to be skewed more toward potential disruption and competitive rebalancing than toward realized operational or financial upside. Recent weakness in public software markets has drawn increased interest from private equity buyers. In an interview with the Financial Times, Thoma Bravo co-founder Orlando Bravo described the recent sell-off that was driven by investor concerns around AI-related disruption as a “huge buying opportunity,” arguing that market fears have outpaced evidence of fundamental deterioration. Bravo emphasized that software value is driven by deep domain expertise and specialization in mission-critical processes, such as payroll and cybersecurity, which he views as less susceptible to displacement by AI, while acknowledging that more generic and undifferentiated software businesses remain vulnerable.

Given the rapidly changing AI landscape and also leveraging the significantly larger public and private company database at Octus, rather than just looking at the largest public companies, our goal with this report was to look at the third quarter transcripts of sponsored sub-investment grade deals and provide more clarity on the impact of AI on companies business models, financial implications, tangible use cases and which business and/or companies are being disrupted by AI.

As of the third quarter 2025, the “general/strategic” use cases of AI were the most prevalent based on the eight AI use case categories we grouped companies into based on the transcript analysis. The “general/strategic” category reflects abstract and mostly visionary ideas that majority of the companies have when it comes to AI implementation. For example, companies were stating that artificial intelligence is a “huge opportunity” that will help boost growth and profitability in the future, by creating more value for clients and improving efficiency internally, though in most cases there wasn’t any supporting evidence provided by the management team to support such claims.

Many AI-related investments, particularly in data infrastructure, cloud computing, and product development, require sustained upfront spending with uncertain near-term returns. Well-capitalized public companies may be better positioned to absorb these costs and iterate rapidly, potentially widening competitive gaps over time. By contrast, highly leveraged private companies may face greater constraints in funding AI initiatives, increasing execution risk if AI capabilities become a competitive necessity rather than a differentiator.

Operations were the second most common category where companies aimed to leverage AI. These ranged from company specific use cases to broader, more generally applicable scenarios. For instance, Barracuda Networks launched an AI-powered cybersecurity platform, BarracudaONE, designed to maximize threat protection and cyber resilience; Fortra has embedded AI across its cybersecurity and automation suite; Ivanti rolled out Ivanti Neurons, an AI-driven IT and security platform; DigiCert introduced DigiCert ONE for automated digital trust management; Cision deployed AI tools across PR Newswire and Brandwatch; Internet Brands embedded AI into healthcare and legal workflows; and Newfold Digital launched AI-powered website and domain tools across its flagship brands.

Some education technology, or EdTech companies have used AI internally and also enabled their teachers to create, personalize, and/or improve curriculum content. For example, Houghton Mifflin Harcourt has embedded AI into curriculum and assessment tools to surface real-time learning insights, personalize instruction, and support teachers’ in-class decision-making based on student performance data.

On a more general basis, a few companies have been using AI to perform document review, highlighting the significantly faster turnaround time and improved scalability, which could minimize human input and ultimately positively reflect on margins. The recent selloff in the legal space highlights how quickly AI is shifting from augmentation to substitution. New AI-native legal Cowork tools by Anthropic have intensified concerns that document review, discovery, and compliance preparation can now be executed with significantly less human input, accelerating pressure on pricing and the long-term relevance of traditional legal software and services.

Finally, assistance in code writing is one of the most popular AI use cases across companies, as engineering teams across the board are using AI in some shape or form to help with coding tasks.
 

The table below summarizes the main AI use cases by category based on the findings from the third quarter 2025 private and public company transcripts. However, when it came to providing quantifiable evidence regarding the impact of AI, only around every fourth company’s management team provided a tangible use case paired with a quantifiable impact on a particular business metric (for instance, one company reduced customer calls by 17% by leveraging AI agents and chatbots), growth and/or margins.
 

 

Different from the market reaction, and based on the sentiment regarding AI, only a handful of companies “admitted” facing challenges and potential disruptions, as artificial intelligence gets more and more integrated globally across a multitude of industries. However, even in such instances management teams were quick to highlight the headwinds being faced by the industry as a whole in which the company is operating, trying to steer away from internal difficulties.
 

So far, AI has been mostly disrupting white-collar industries and digital jobs. When benchmarking the AI-disrupted digital business categories from the “Software & Services Industry Areas Disrupted by AI” table presented above against the broader software and services industry, for the most part, revenue and EBITDA growth, and EBITDA margins have been lagging over the last few years of the disrupted companies (additional benchmarking figures can be found in the Market Data file on the Fundamentals landing page). In terms of top-line and EBITDA growth, marketing related categories and website builder companies struggled the most, as quarterly revenue and EBITDA has mostly declined for these firms over the last two years.
 

The field of marketing has arguably been one of the most impacted fields by AI. For instance, traditional web traffic generated by search engine optimization (SEO) is now being cannibalized by generative (GEO) or answer (AEO) engine optimization, as traditional search is shifting towards LLMs. Red Ventures’ SEO-dependent businesses are facing meaningful traffic pressure from AI search summaries, contributing to revenue declines and weaker EBITDA despite ongoing cost actions.

Furthermore, services offered by advertising and creative agencies can now be done significantly quicker and at a fraction of the cost by using AI tools, with minimal barriers to entry of accessing such tools, as majority of it revolves around natural language prompting. Nevertheless, it’s important to note that the output quality of AI varies and also depends on the complexity of the task. Consequently, it is still one of the drawbacks and limitations impacting the scalability of the use of AI at this point in time. As improvements of AI and LLMs rapidly grow and evolve over time though, companies that will actually leverage the technology to its full potential and apply it in practice will be able to meaningfully improve their growth and margins.
 

Margin-wise, the median AI disrupted customer support and business process outsourcing (BPO) companies have mostly posted compressing EBITDA margins over the last few years. One of the most popular AI use cases as of the writing of this report is AI agent and chatbot usage to deliver customer support related services, such as answering customer questions, dealing with complaints, order tracking, etc. Due to the relatively low complexity nature of such tasks, companies that have been using AI have been able to successfully automate a big portion of these tasks, resulting in the reduced need of customer support staff, ultimately alleviating some of the pressure from operational costs. At the same time, companies that have yet to leverage AI in customer support services have seen their margins contract. BPO companies failing to implement AI into their workflows will face further headwinds, as AI-enabled competitors will be able to essentially offer the same services for a fraction of the cost and eat into competitor market share.