By Ellen Hazen, CFA®, Chief Market Strategist
- Artificial intelligence (AI) announcements triggered significant declines across various software and services stocks as markets repriced future revenue durability
- Software valuations had already been under pressure since late 2025; AI-enabled insourcing threatens Software-as-a-Service (SaaS) pricing power and volume assumptions
- Private credit markets are feeling second-order effects, particularly in direct-lending portfolios with high exposure to software companies
- Blue Owl Capital’s halt of quarterly redemptions and loan sales intensified scrutiny of liquidity structures
- Despite these stresses, systemic risks appear contained: today’s environment lacks the leverage, opacity, and bank linkages seen in 2007
- Market leadership is shifting toward tangible-asset sectors – energy, industrials, materials – as software valuations compress and investors rotate exposures
This month we examine how rapidly evolving AI developments are reshaping market expectations, particularly in areas where business models rely heavily on intangible assets such as software, data, and subscription services. As AI announcements accelerate, investors have begun to reprice uncertainty across sectors, prompting sharp reactions in software equities and raising broader questions about valuation durability, credit exposure, and the shifting balance between intangible and tangible assets. To frame the landscape, we highlight three linkages: the immediate market response to major AI releases, the downstream effects on private credit tied to software-driven business models, and the emerging rotation toward sectors grounded in physical assets.
Market Response to AI Releases
Over the past month, announcements of rapidly accelerating AI capability proliferated, leading markets to reflect increased uncertainty about the prospects of companies in a variety of industries. For example, software companies that serve industries such as finance, legal, and marketing saw their stocks significantly decline after Anthropic released Claude Cowork, a coding software with industry-specific plug-ins. Thomson Reuters declined by 15%, S&P Global declined by 11%, and LegalZoom declined by nearly 20%.
This was quickly followed by another Claude capability, the ability to manage several AI agents. A couple of weeks later, Anthropic published a blog post describing how its AI agents can modernize and update programs and systems written in COBOL, a programming language that was dominant in the 1980s and that still operates many workhorse systems that require high reliability and high-transaction volumes, like ATMs, insurance claims processing, and airline reservations. IBM is one of a handful of companies specializing in COBOL programming and maintenance. After this blog was published, IBM stock fell by 13%.
Software stocks had been weakening since September 2025, as the market had been anxious about the growth prospects and revenue durability of many software companies. Over the past two decades, many software companies have evolved from a license model, where a customer buys a version of the software as a one-time purchase and must separately purchase newer versions, to a “Software-as-a-Service” model, where customers instead pay a monthly subscription fee and updates are included with the subscription. If AI agents like Claude Cowork can enable corporate clients to self-develop some capabilities, this recurring revenue model in the software industry becomes at risk, from both a volume and pricing perspective. This fear was compounded by technology company CEOs recently stating publicly that they expect to need fewer coders going forward: Salesforce’s Marc Benioff said in December that he anticipates hiring no new engineers due to AI productivity improvements, while Block’s Jack Dorsey announced at the end of February that the company will lay off 40% of its workforce for the same reason. Maybe companies can create in-house solutions, and maybe they can’t, but the claim that they will try is a powerful bargaining chip with software companies.
Downstream Effects on Private Credit Tied to Software-Driven Business Models
More recently, this uncertainty has bled into the private credit realm. Fifteen-to-twenty percent of private credit loans are to software companies. Five years ago, a time of near-record private credit issuance coincided with high software multiples, resulting in private equity takeouts with pre-2022 (i.e., high) leverage. Consequently, the market has become more concerned about the credit quality of those private credit loans. If private equity or private credit companies mark down ownership stakes in software companies, what happens to overall private equity or private credit values and liquidity? Many private markets, including private credit, have grown in recent years in part by offering semi-liquid vehicles like interval funds to the public. Historically, investors in private assets agreed to lock up their capital for a number of years, which meant that any valuation markdowns triggered by more volatile public market valuations could be avoided or at least delayed. More recent private credit vehicles designed for a broader investor base provide periodic (often quarterly) liquidity. One can see how increased investor demand for redemptions can cause general partners to sell those loans that can be sold, driving valuation markdowns across the industry.
In mid-February, a large private credit issuer, Blue Owl Capital, announced that it would end quarterly redemptions for one fund while selling a third of that fund’s portfolio at a small haircut (99.7 cents on the dollar) and distributing 30% of the fund’s value back to investors. This was quickly followed by an activist tendering for Blue Owl assets at a larger discount.
Private credit is a large sub asset class with many variants. The Blue Owl fund discussed above is involved in direct lending, which is one of the larger areas within private credit. However, many other areas of private credit are not materially impacted by these software revaluations, such as asset-backed lending and litigation finance. We have generally avoided the direct lending sub asset class for clients, preferring instead other areas within private credit.
Some investors worry that cracks in the private credit sub asset class may foreshadow broader credit problems; however, today we are not overly concerned that the Blue Owl fund serves as a similar warning. Recall that in 2007, two Bear Stearns mortgage hedge funds lost most of their value and, in hindsight, served as an early warning sign that credit markets were overextended. The environment today is not characterized by the high leverage and opacity present in 2007, nor are these private credit funds as intimately tied to the banking system as the Bear Stearns hedge funds were. In addition, banks have much higher capital than in 2007 and are regularly subject to stress tests by regulators. Thus, contagion from any weakness in the direct lending private credit arena is less likely.
Emerging Rotation Toward Sectors Grounded in Physical Assets
Part of the reason investors are concerned about AI disruption is that starting valuations across several asset classes were high; for example, the S&P 500 started 2026 nearly two standard deviations above its 30-year median multiple of two-year forward earnings estimates. Moreover, investors remain uncertain about how artificial intelligence will develop. In late February, one firm published a hypothetical scenario describing how artificial intelligence could cause economic weakness in upcoming years by reducing employment, which in turn would weaken consumer spending. This report may have contributed to a 7.5% decline in software stocks that day.
More broadly, the market is reflecting fear that anything intangible may be digitized and then commodified, penalizing those very stocks that had outperformed for many years. Year-to-date, the best-performing sectors in US equities are in decidedly tangible industries: energy, materials, industrials, and consumer staples. The market is rotating; while the S&P 500 is flat year-to-date, under the surface, energy stocks (as measured by the S&P 500 Energy Index) are up 28% while software stocks (S&P 500 Software Index) are down by 20%. We have previously written about the narrow market, where only 30% or fewer companies in the S&P 500 Index outperformed the benchmark in each of the past three years.
Conclusion
Taken together, these developments illustrate how AI has become a catalyst for repricing risk, testing the resilience of business models built on recurring revenue. While software-related revaluations have introduced stress in pockets of private credit, the broader system today lacks the leverage and opacity that characterized earlier credit crises, reducing the likelihood of widespread contagion. At the same time, market leadership has shifted toward sectors with more tangible assets, reflecting investors’ desire for clearer cash-flow visibility as AI-related uncertainty persists. As always, diversification remains essential, and we continue to monitor key indicators (employment trends, credit spreads, corporate AI adoption, among others) to assess whether these moves represent temporary dislocations or the early stages of a more durable market regime. Ultimately, a broader market is a healthier market, and we see opportunities beyond the narrow artificial-intelligence equities that have led the market in recent years.