Recent client conversations have converged around two questions:
Whether artificial intelligence (AI) represents a speculative bubble
How concerned we should be about private credit
This note outlines our perspective and where we see the most risk.
AI's unprecedented potential for technological advancement is evident in anecdotal cases:
Enterprise adoption is also gaining traction. As of March, over 30% of companies in technology, education, and finance report using AI in business functions, with all sectors expecting increased adoption over the next six months.
This trend is similarly reflected in our own experience. Since ChatGPT’s release in 2022, the Burney Analytical team has seen meaningful productivity gains by using AI tools for data validation and organization, coding, and company and manager due diligence.
In most technological cycles, early leaders are not guaranteed long-term dominance. At this juncture, market confusion reflects a combination of:
Two structural features of the stock market add to the problem:
1. The stock market is not fully rational—especially in the short term.
Sentiment and uncertainty lead to distortions and dislocations, with some AI-related stocks selling off on fear rather than fundamentals.
2. The stock market is inherently forward-looking.
Since Q4 2025, much of the stock market’s anxiety has centered on how to position for AI. Investors are actively searching for “the next winner,” leading to frequent rotations and short-term volatility. Mega Techs are now being evaluated not just on CapEx (capital expenditure), but on visible ROI (return on investment). Following last Wednesday’s earnings release,
In 1913, Henry Ford introduced the modern moving assembly line. Its net effects:
AI is a real and POSITIVE structural shift, but the question financial markets are grappling with is who ultimately captures its economic value. Jensen Huang described AI as a “five-layer cake,” from bottom to top: energy, chips, infrastructure, models, and applications. Winners are—and will continue to—emerge within and across these layers, creating opportunities for investors who are diligent and adaptable.
Cliffwater Direct Lending Index (CDLI) data indicate a healthy private credit market as of the end of 2025, with defaults, non-accruals (pre-default), leverage, and PIK (payment-in-kind, a form of loan modification) all showing no signs of unusual stress or adverse trends.
While headlines draw attention to fund redemptions, actual investor behavior is more measured. In Blue Owl’s case, tender offers at steep discounts attracted less than 1% of shares, suggesting limited desire to exit at distressed prices. Flows into private credit continue, though they have become more selective. In Q1, Goldman Sachs’ private credit BDC saw 7.1% inflows (according to Fitch Ratings). JPMorgan is also planning to launch a new private credit interval fund, underscoring continued opportunity and sustained investor demand.
The media has been beating the drum on payment-in-kind (PIK) risk for several years. In reality, about half of current PIK exposure is present at origination, indicating a significant portion is structural rather than stress-driven. Defaults only became more visible in late 2025, with several high-profile cases, all involving fraud or alleged fraud.
By November 2025, investors began seeking exits, and Blue Owl was unfortunately caught in the shift in sentiment. Private credit loans typically have a 5–7 year maturity, and unlike public bonds, managers generally intend to hold them to maturity. Interval funds provide quarterly liquidity gated at 5%, funded through revolvers, CLO warehouse lines, and portfolio cash flows from loan repayments and interest. Private BDCs are not legally required to provide periodic liquidity, but many adopt a similar schedule as interval funds. In either case, we believe that investors should not treat these vehicles as if they were as liquid as ETFs, and managers should maintain the 5% gate in order to protect remaining investors.
It is important for investors to recognize that, although private credit typically involves unrated or lower-rated companies, direct lending—roughly half of the market—is often structured as senior secured debt, backed by the company’s balance sheet and sits at the top of the capital structure. For losses to occur in private direct lending, financial stress must first erode the equity and junior debt layers.
The majority of private direct lending loans are also "sponsored", meaning a private equity manager is actively involved with the borrower. The private credit sector is also tangled with banks and insurers. U.S. banks hold nearly $300 billion in direct exposure to private credit (per Moody’s), and U.S. life insurers hold more than 31% of their assets in private bonds (most are a form of private credit; data per BlackRock).
Structurally, in our view, it is unlikely for private credit to be systematically underwater while banks, private equity, and the broader economy remain healthy. A recent piece from Morningstar echoes this conclusion from a different angle, noting that, as a lender, private credit may be more stable than banks due to much lower leverage, less asset-liability mismatch, and more diversified holdings.
While AI is a positive force for the broader economy and private credit portfolios are currently healthy, we see elevated risks in two specific AI-related segments.
Historically, software has been an attractive business. The median annualized growth in EBITDA earnings among private credit software borrowers has been 36% over roughly the past two years (per KBRA), leading to software's steadily rising portfolio weights since 2013 and outsized exposure in private credit today.
Now, AI is disrupting enterprise software business models—particularly seat-based pricing. Software companies also have limited time to adapt with a wall of maturities approaches in 2028. Risk looks increasingly binary and outcomes may vary significantly across positions in the value chain. For private credit, all of a sudden, software's asset-light structure—limited collateralizable assets— is no longer advantageous and instead implies lower recovery rates if loans deteriorate.
Private credit is increasingly funding AI data centers. Morgan Stanley estimates that, although nearly half of the $2.9 trillion in funding needed for data centers from 2025–2028 will come from hyperscalers’ capex, private credit could provide roughly $0.8 trillion, making it a significant second source of funding.
Data centers present the opposite profile of software: rather than asset-light, they are asset-heavy, requiring substantial upfront capital while still facing the risk of lower recoveries due to potential technological obsolescence. Lenders to unprofitable or overleveraged projects could incur losses if some AI-driven buildouts prove uneconomic due to: