Increasing Number of Companies Report Quantifiable AI Benefits in Q4’25; Certain Industries Face ‘Existential’ Disruption Following SaaScalypse Price Re-Ratings; Financials Services May Be Next

The Mass Integration Phase of AI Adoption Has Begun

The adoption of artificial intelligence, or at least the claims, amongst companies has been increasing exponentially over the last few quarters. According to FactSet, in the fourth quarter of 2025, 68% (331 out of 485) of S&P 500 companies had mentioned “AI” in their earnings calls – a record over the last 10 years. That’s an almost 60% jump in comparison just to the first quarter of 2025 (210 mentions).

However, AI has been a double-edged sword for companies. On the one hand, companies providing the infrastructure to enable mass adoption of AI, such as semiconductor providers (example Nvidia and AMD), energy companies (example Constellation Energy, NextEra Energy), and cloud infrastructure platforms, are reaping immediate financial rewards. These entities serve as the “picks and shovels” of the modern “AI gold rush”, supplying the computing power, specialized physical real estate and enormous electrical output required to train and run complex models. On the other hand, for many end-user enterprises attempting to integrate these technologies, AI reflects a meaningful cost, both financially and the time it takes to educate employees on the rapidly evolving AI tools, with a still-uncertain return on investment. To take a step further, some industries have already been disrupted by AI. A perfect example of this was the SaaScalypse – a sharp sell-off of legacy software providers that happened earlier in the year after Anthropic’s announcement of the release of new Claude plug-ins across 11 different categories, which was covered in detail in our previous report HERE.

This article is a continuation of our third-quarter report on how management teams have been thinking about and integrating AI in their companies and workflows based on the fourth quarter 2025 earnings call transcripts. We will continue to share another update once the first quarter 2026 earnings conclude.
 

Notes & Methodology Overview: we analyzed over 1,000 private and public earnings call transcripts screening for various artificial-intelligence related keywords. The other criteria was to narrow down the commentary related to the fourth quarter 2025 calendar period. After this screening, we were left with approximately 150 unique companies.

Approximately 71% of the companies covered in this report are U.S. based, 14% have operations both in the U.S. and Europe and the remaining 15% are European companies. The distribution of companies sector-wise was as follows: 22% are in the IT sector, 19% – industrials, 16% – consumer discretionary, 13% – financials, 10% – healthcare and the remaining 20% spread out across the remaining sectors.
 

• AI adopters capitalize on efficiency and growth gains. Issuers successfully integrating AI are demonstrating concrete cost optimization, such as substantial headcount reductions in support functions alongside rapid top-line expansion. For example, one IT services company cited a 6x increase in AI-related revenue in a single year (though important to note starting from a relatively low base).

• AI-disrupted companies posted deteriorating financials. Certain industries and pockets of the market such as legacy software and consulting, for example, were amongst the few that admitted to facing challenges from AI and have experienced dents in financial performance over the last few quarters. In particular, for the median issuer that admitted to facing negative headwinds from AI – EBITDA has declined for the second consecutive quarter, followed by rise in debt, ultimately resulting in higher leverage.

• Credit markets are actively punishing AI laggards. The fundamental deterioration among disrupted issuers has already been priced into the secondary market. Leveraged loan instruments for issuers citing AI headwinds in the fourth quarter of 2025 have experienced accelerated price erosion compared to their peers.

• Flexible enterprises are bypassing rigid software contracts. Rather than renewing expensive, per-seat legacy SaaS agreements, agile companies are starting to utilize AI coding assistants to build custom in-house tools in a matter of days or weeks. This shift allows smaller-scale organizations to escape generic off-the-shelf software solutions and mold systems directly to their proprietary data.

• Financial services is emerging as the next major automation frontier. In a recent keynote speech done by the team at Anthropic, it was stated that financial services is the second fastest-growing automation category after coding. Some structural changes in financial services have already taken place. For example, Block laid off 4,000 of its employees earlier this year (40% of the workforce), and Jack Dorsey, the CEO, said that now 100% of employees use AI daily.

AI Adopters – Issuers Capitalizing on Efficiency and Growth

As AI adoption increases, we are seeing more concrete, quantifiable use-cases shared by management teams during earnings calls. During the third quarter of 2025, approximately 28% of companies had provided a quantifiable impact on their business from the implementation of AI, while in the fourth quarter the figure increased to 33%.
 

The immediate impacts were seen in the efficiency and cost optimization areas. One insurance provider stated that it has initiated a 50% reduction in its finance department, and at the same time are planning to eliminate over 2,000 roles throughout the firm in the next 12 months as it focuses to further adopt AI. A portfolio management company now has over 80% of portfolio planner use cases assisted or fully replaced by AI. An established legacy media firm anticipates approximately $2.5 million in cost savings in 2026 from AI in creative and content production.

However, we are starting to see more and more use cases where AI is also driving top line growth. One IT services and solutions provider mentioned that AI-related project revenue had increased by 6x in a single year. A cybersecurity firm saw a 3x increase in AI revenue during 2025, as cybersecurity is becoming increasingly more important with the advancements of AI models. It’s important to note though, that these AI revenue figures representing growth in orders of magnitude, so far, are generally due to starting from a relatively low base.

Nevertheless, AI integrations do come with a cost. One management team stated that as a company, cumulatively it has spent over $700 million on AI in the last seven to eight years. It seems that the majority of companies understand and are banking on further rapid improvements of artificial intelligence models and cost decreases per token, which would increase scalability even more and result in a higher ROI.
 

AI Vulnerabilities – Companies Facing Disruption and Headwinds

Certain industries, on the other hand, have had their business models disrupted by AI. The industry that has experienced the most pain so far is software. We covered the shake-out of the “SaaScalypse” in greater detail in our earlier report, which you can find HERE. We have also done a deep-dive on almost every single private software credit available in the Octus universe. If you would like to find out more on a particular name, reach out to our PCA team for more details.

In the fourth quarter of 2025, only nine companies admitted to facing headwinds from AI. Interestingly, eight out of the nine companies were European issuers. Part of this could be due to the stricter regulations on new technology, innovation and GDPR rules that limit European companies on how much they can implement AI into their workflows. The three key areas disrupted from AI and reflected in the table below were legacy software, consulting and media. With the rapid advancements in AI models, more and more companies, in particular those of a smaller scale that are more agile, are building custom software in-house and moving away from legacy software subscriptions. For instance, B2B software company Chili Piper recently eliminated 10 legacy SaaS vendors in a single quarter. Instead of renewing expensive and rigid software contracts, their team used AI coding assistants to build internal tools in a matter of days. This shift allowed them to escape restrictive ‘per-seat’ pricing models while creating software molded to their proprietary data, proving that agile companies no longer need to rely on generic off-the-shelf solutions. Similarly, a big portion of consulting tasks can now be done by LLMs at a much faster pace and at a fraction of the cost. For example, synthesizing a year’s worth of multiple competitors’ earnings transcripts to map industry strategy might take a junior consultant north of 40 hours, billing the client upwards of $10,000. An LLM can process that same massive volume of text and generate a structured competitive matrix in minutes, for an operational cost of less than five dollars – drastically shifting the value proposition of traditional consulting.
 

Companies with negative AI sentiment based on the fourth quarter 2025 transcripts experienced more volatility when it came to growth over the last year. While top-line growth remained mostly positive for the median issuer, EBITDA had declined for the last two consecutive quarters. At the same time, net leverage had gradually ticked up in the last few quarters in comparison to flat leverage amongst peers, which was impacted from both EBITDA decline and debt increases. Out of the nine issuers with negative AI sentiment, seven had posted year over year increases in gross debt in the fourth quarter of 2025. Companies with negative AI sentiment also posted one to two percentage point higher capex margins. While it hasn’t been explicitly stated that the increase in debt for most of these issuers over the last year is, at least partly, related to funding the costs of implementing AI, it is in line with the broader consensus in the market that more and more companies, even the hyperscalers with huge cash flows, may struggle to fund AI integration costs and as a result have started issuing debt. The dent in financial performance of the companies that admitted to facing challenges from AI coincided with price declines when looking at lev-loan instruments of these issuers over the last 1.5 years.
 

Software Segment in Focus

Following the SaaScalypse that happened earlier this year, we did a deep-dive on almost every individual software related credit that we cover in the Private Company Analysis (PCA) universe, including such names as BMC, Tungsten Automation, Cornerstone OnDemand, Ivanti, Fortra, Yahoo, Red Ventures, Newfold Digital, RSA and Internet Brands. Overall, enterprise software and cybersecurity issuers are positioning AI as a demand accelerator rather than a disruption risk, using it to drive platform migration, workflow automation, customer stickiness, new product SKUs and internal productivity. Media and traffic-dependent businesses face greater exposure to AI-driven search disruption but several are mitigating that risk through proprietary distribution, vertical data assets and AI-enabled products.

AI is increasing enterprise complexity, making orchestration and control more valuable. BMC framed the proliferation of AI agents and AI-driven workflows as a tailwind, arguing that it creates “incremental complexity and fragmentation” and increases the need for coordination, orchestration and operational control. Public peers echoed the same theme: CDW stated that “the hard part of AI is not the model. It’s the orchestration,” while ServiceNow and Dynatrace similarly emphasized that more agents and agentic AI increase the need for governance, data connectivity, observability and autonomous operations.

Vertical workflow incumbents are using AI to accelerate land-and-expand. Tungsten Automation emphasized that generic LLMs do not scale efficiently or provide sufficient auditability in regulated workflows, highlighting its ability to combine “deterministic and probabilistic capabilities” within governed, document-heavy processes, specifically for highly regulated industries such as financial services, insurance, healthcare and government. Public-company commentary supports this thesis: EXL said its domain-specific AI and orchestration capabilities allow it to embed intelligence “directly into how work gets done, not as an overlay,” reinforcing the value of workflow-specific data, rules, and governance.

AI is becoming a catalyst for legacy-platform migration. Cornerstone said AI features in Galaxy are helping drive migrations from Saba and SumTotal and migrated customers generating double-digit ARR uplift on average. Management noted that the value customers receive from the new platform, “particularly with the AI features,” has become a migration catalyst. SAP framed AI as part of the value proposition for moving customers onto its newer platform architecture, noting at its 2026 Financial Analyst Conference that it now has “the platform,” “the new AI platform,” and “AI migration tools,” where it can “deliver a ton of value given the size of this market.” NICE similarly identified “driving AI-first growth across every customer touch point,” “agentic AI on our platform” and “capitalizing on the CCaaS cloud migration” as core growth catalysts. Calix also linked its migration to a next-generation platform with the agentic AI opportunity, saying its “migration to our third-generation platform in partnership with Google allows us to support the success of our existing customers” while positioning the company to capture “the Agentic AI opportunity ahead.” These examples support the broader point that AI is not just a product feature, it is increasingly being used as a strategic reason for customers to modernize from older platforms to cloud-native, data-rich, AI-enabled architectures.

Cybersecurity vendors see AI as both a threat multiplier and demand accelerator. Ivanti and Fortra both argued that AI expands the attack surface while increasing demand for security infrastructure. Ivanti highlighted AI-enabled vulnerability discovery, exploit generation, and attack scale, supporting its autonomous endpoint management thesis. Fortra described AI as a “tremendous net opportunity,” noting that it creates new risks around shadow AI, phishing, deepfakes and data loss, but is also “the most important tool to protect companies from that risk.” Public cybersecurity issuers such as CyberArk, Qualys and N-able made similar points, tying AI-driven threats to increased demand for identity, detection, remediation, data security and governance.

Media and traffic-dependent businesses face AI-search disruption, but model matters. Red Ventures, Yahoo, Newfold Digital, RSA and Internet Brands/WebMD have varying exposure to search traffic, affiliate monetization, and consumer intent capture. The core risk is that AI answer engines reduce search click-through and shift traffic capture away from traditional publishers. However, companies with proprietary distribution, vertical data assets, direct user relationships and high-intent workflows appear better positioned. Red Ventures noted that SEO is becoming less central as proprietary distribution and nontraffic revenue scale, while Internet Brands argued that its SEO/AIO dependency is now only mid-single digit and that AI is a net positive given its vertical data, omnichannel reach and productized AI offerings.

Financial Services – the Next Industry to Follow the Software Path?

In a recent keynote speech done by the team at Anthropic, it was stated that financial services is the second fastest-growing automation category after coding. Members of Anthropic expect to see exponential growth and improvement in the types and complexity of tasks that can be automated within finance in the next 1-2 years.
 

An announcement from Block, a FinTech company with a focus on providing payment solutions, in February this year announced layoffs of over 4,000 employees, approximately a 40% reduction of the total workforce. The company’s CEO Jack Dorsey in the latest Q1 2026 earnings call said “…AI tools have changed what it means to build and run a company. A significantly smaller team, using the tools we are building, can do more and do it better.” However, the counter-argument on the street is, which is not aimed at Block directly but more broadly, is that some companies have accumulated unnecessary “extra-fat” during the post-pandemic boom due to over-hiring and have been trimming their workforce over the last few years to an extent using the AI narrative as a scapegoat.

Fund administrators such as Citco and Alter Domus, for the most part, have expressed how they are implementing AI into their operations and increasing efficiency. However, fund administration is mostly human-capital intensive business, and if artificial intelligence continues to improve and ultimately be able to accurately perform key billable tasks such as reconciliations and capital call processing amongst others – the pricing pressure from GPs to reduce administration fees may intensify.

Goeasy, one of Canada’s largest non-prime consumer lenders, which serves borrowers who cannot access traditional bank credit, faces potential AI disruption risks, on top of its abysmal Q4 2025 earnings results (the company suffered a credit event that resulted in a C$178 million incremental LendCare charge-offs, a C$160 million goodwill impairment, annualized net charge-off rates of 23.8% (up from 9.2% a year earlier), dividend suspension, withdrawal of its three-year forecast, and a stock that fell over 60% in weeks). While the poor fourth-quarter results were due to a mix of macroeconomic (Canada’s consumer insolvency rate reached 17-year high), operational and M&A (acquired in 2021 LendCare’s indirect dealer channel was structurally riskier than the core business) factors, the company faces disruption that AI-native lenders are beginning to create at the origination level. Platforms like Upstart, which uses AI and non-traditional variables to price non-prime credit more accurately than FICO-based models, are capturing origination volume in the non-prime segment. If AI native lenders systematically take the better quality end of the non-prime borrower pool (those who are underserved by FICO but creditworthy by behavioral/employment data), legacy lenders like goeasy are left with worse selection.