Credit markets are emerging as the next major arena for artificial intelligence disruption, with UBS credit strategy head Matthew Mish warning that tens of billions of dollars in leveraged loans and private credit could default over the next year.

This is because software, data services, and other AI-vulnerable companies—particularly those owned by private equity—face intensifying margin compression and revenue erosion.

In a detailed research note released Wednesday and in a subsequent CNBC interview, Mish and his UBS team laid out a baseline scenario projecting $75 billion to $120 billion in additional defaults by the end of 2026 across leveraged loan and private credit markets, which they estimate at roughly $1.5 trillion and $2 trillion in size, respectively.

The forecast assumes default rates rising by up to 2.5% for leveraged loans and up to 4% for private credit—meaningful increases from current levels that are already showing signs of stress. Mish described the acceleration of AI disruption as the primary driver behind the revised outlook.

“The market has been slow to react because they didn’t really think it was going to happen this fast,” he said. “People are having to recalibrate the whole way that they look at evaluating credit for this disruption risk, because it’s not a ’27 or ’28 issue.”

The shift in perception was catalyzed by recent model releases from Anthropic and OpenAI that demonstrated advanced reasoning, tool integration, and task automation capabilities—directly threatening routine knowledge work, data processing, and professional services that underpin many leveraged software and data firms.

Mish categorized companies into three broad groups in the AI landscape:

1. Foundational model creators — Startups like Anthropic and OpenAI (soon potentially large public companies) that develop frontier large language models and stand to capture significant value at the top of the stack.
2. Investment-grade software incumbents — Companies like Salesforce and Adobe with strong balance sheets, established customer bases, and the ability to rapidly integrate AI to defend their moats.
3. Private equity–owned software and data services firms — Highly leveraged companies carrying substantial debt, often acquired in large buyouts during the low-rate era. These firms, according to Mish, are the least likely to emerge as long-term winners in a rapid, disruptive AI transition.

“The winners of this entire transformation—if it really becomes, as we’re increasingly believing, a rapid and very disruptive or severe [change]—the winners are least likely to come from that third bucket,” Mish said.

He also outlined a more severe “tail risk” scenario in which defaults could double the baseline estimates, triggering a “credit crunch” in loan markets, broad repricing of leveraged credit, and systemic shocks. While UBS is not yet calling for this tail scenario, Mish noted the firm is “moving in that direction” as AI model capabilities advance faster than anticipated.

The warning follows a rolling series of selloffs that began with software stocks earlier this month and spread to finance, real estate, trucking, and other sectors perceived as vulnerable to AI automation. The rapid pace of disruption—accelerated by Anthropic’s Claude plug-ins and OpenAI’s tool integrations—has forced investors to reprice credit risk far sooner than the previously expected 2027–2028 timeline.

Private equity–owned software and data firms are particularly exposed. Many were acquired at peak valuations during the low-rate environment of 2020–2022, loaded with leverage, and reliant on recurring revenue from maintenance contracts, legacy system support, and routine analytics—precisely the areas most susceptible to AI automation.

As AI agents handle contract reviews, compliance checks, data extraction, and report generation, pricing power erodes, and customer churn accelerates. Mish emphasized that timing remains the key uncertainty: the pace of large corporate AI adoption, the rate of model improvement, and the ability of incumbents to adapt will determine how quickly credit risk materializes.

“We’re pricing in part of what we call a rapid, aggressive disruption scenario,” The UBS noted, suggesting that the direction is clear.

The UBS note adds to a growing chorus of warnings from credit strategists and analysts. JPMorgan and Morgan Stanley have also flagged rising default risks in private credit and leveraged loan markets, particularly among software and professional services borrowers. Moody’s and S&P Global Ratings have placed numerous private equity–backed software companies on negative watch lists in recent weeks, citing AI disruption as a key risk factor alongside elevated interest rates and slowing organic growth.

The broader leveraged loan and private credit markets—estimated at $3.5 trillion combined—are already showing stress. Secondary loan prices have declined steadily since mid-2025, and spreads have widened significantly for lower-rated issuers. Private credit funds, which stepped in to fill gaps left by retreating banks, now face their own maturity walls and refinancing challenges as portfolio companies struggle to grow in an AI-disrupted environment.

For now, investment-grade software firms with strong balance sheets and clear AI integration strategies (e.g., Salesforce, Adobe, ServiceNow) are seen as more resilient. However, the middle and lower tiers, especially private equity portfolio companies, face the highest risk of default and restructuring.

As the market recalibrates for faster AI disruption, credit investors are increasingly differentiating between AI winners and losers. Companies that can demonstrate defensible moats, rapid AI adoption, and strong balance sheets are likely to see credit spreads tighten, while those slow to adapt or burdened by leverage could face sharp repricing and distress.

Analysts expect some changes in the coming quarters. If large enterprises accelerate AI deployment and begin replacing traditional software and services contracts, default rates could rise quickly. If adoption lags or incumbents successfully integrate AI to defend their positions, the credit impact may be more contained.

Meta-backed artificial intelligence training firm Scale AI is suing the U.S. Department of Defense in a case that could involve classified information, according to court filings.

The complaint, filed on January 30 in the U.S. Court of Federal Claims, is largely under seal. One of the few publicly available documents indicates that materials in the case are expected to include information classified at the “secret/no foreign” level. The United States is the only named defendant.

Another AI company, Enabled Intelligence, has joined the case as an intervenor defendant, a third party that voluntarily enters litigation to protect its interests.

Scale declined to comment directly on the litigation, saying only that it “relates to a recent procurement decision.”

“Scale AI stands firmly with Secretary Hegseth and the Department of War in their mission to get frontier AI capabilities into the hands of warfighters. We are committed to ensuring the procurement process reflects the high standards required for our nation’s most critical AI initiatives,” a spokesperson said.

Attorneys for Enabled Intelligence and the Defense Department did not respond to requests for comment.

Disputed $708 Million Contract

The lawsuit follows Scale’s unsuccessful bid for a contract worth up to $708 million from the National Geospatial-Intelligence Agency, a DoD component. The contract, awarded to Enabled Intelligence, could span up to seven years and represents the agency’s largest data-training agreement to date.

The award includes work supporting Maven, the Pentagon’s flagship AI initiative aimed at improving analysis of imagery and geospatial intelligence.

Scale filed a bid protest with the Government Accountability Office in late December, challenging the procurement decision. The GAO dismissed the protest in late January, two days before Scale filed its lawsuit in federal court. The GAO typically does not publish detailed information about routine dismissals.

In 2024, Scale had secured a separate $24 million, one-year contract from the same agency for data labeling services tied to Maven. The loss of the larger follow-on contract appears to have triggered the dispute.

Because most filings remain sealed, it is unclear whether Scale is seeking to overturn the contract award, obtain monetary damages, or compel a new procurement review.

Scale’s Expanding Defense Footprint

Scale has signed multiple multimillion-dollar agreements with the Defense Department since 2020, positioning itself as a key supplier of AI training data and related services.

In March, the company announced a collaboration with defense technology firm Anduril Industries and Microsoft to deploy AI agents within the U.S. military under a DoD initiative known as “Thunderforge.” In August, Scale disclosed a $99 million Army contract to develop AI tools.

The company built its reputation through data labeling services that train large language models and computer vision systems, supporting technology firms such as Google and Meta Platforms.

In June, Meta invested $14.3 billion in Scale in exchange for a 49% stake, marking one of the largest single investments in an AI infrastructure provider.

Scale’s former chief executive, Alexandr Wang, wrote an open letter to President Donald Trump after his second inauguration outlining policy recommendations to accelerate U.S. AI development. In the letter, Wang urged greater federal spending on data and computing infrastructure and highlighted Scale’s defense work.

Wang later left Scale to join Meta’s Superintelligence Labs as chief AI officer and attended the president’s AI-focused dinner at the White House in September.

The lawsuit places Scale in a delicate position. The company is both a contractor seeking expanded defense work and a litigant challenging a procurement decision tied to a highly sensitive national security program.

The dispute comes at a time of transition for Scale. Since Meta’s investment, the company has laid off approximately 200 employees, or about 14% of its workforce. It has also lost major clients, including Google and xAI, while facing intensified competition from newer AI data providers seeking to capture market share.

The outcome of the case could carry implications beyond the companies directly involved. Defense-related AI contracts are increasingly lucrative and strategically important as the Pentagon accelerates deployment of machine learning systems in intelligence analysis, logistics, and battlefield operations.

A court ruling that scrutinizes the procurement process for high-value AI contracts could shape how future awards are evaluated, particularly in programs involving classified capabilities.

For now, key details remain sealed, leaving unanswered questions about the legal arguments at the center of the dispute and the potential impact on one of the Pentagon’s largest AI data initiatives.