The artificial-intelligence build-out is reshaping how data centers get funded, and a growing slice of that money is coming from private credit. But the way those loans are valued means investors may not see problems until they've already been building for months.
Big tech names like Meta, Alphabet, and Amazon can tap public bond markets, but a large and rising share of data-center financing now runs through private credit funds and infrastructure vehicles. These loans don't trade on an exchange, so managers use what's called “mark-to-model” pricing: they estimate the value using cash-flow assumptions and comparisons to similar deals. That's a standard practice, but it can smooth out volatility and understate how quickly conditions are shifting.
Why the numbers matter now
The private credit market has grown large enough that those assumptions carry real weight. The OECD estimated AI-related deals made up 34% of global private credit transactions in 2025. The Bank of England has said roughly $700 billion of data-center spending over 2026-28 could be financed by private credit. And the Financial Stability Board put the entire private credit market at $1.5-2 trillion at the end of 2024.
That scale means a lot of retirement money is now exposed to loans that are valued on paper rather than by live trading. For everyday investors, the key is understanding how this valuation gap works.
When a loan doesn't trade, its price isn't updated every second like a stock. Instead, managers estimate what it would be worth based on assumptions about future cash flows and what similar loans are fetching. That's fine in calm times, but it can mask deterioration. If the underlying borrower's business weakens, the model may not capture it until the next quarterly valuation date.
The collateral problem
There's also a question of what's actually backing these loans. JPMorgan Asset Management has noted that the building itself is only about 30% of an AI data center's cost; the rest is chips and equipment. And the pricing power for that gear can move fast. Silicon Data said the hourly rental rate for Nvidia's H100 chip fell 25%-50% across regions since early 2025.
That kind of decline matters because it affects the value of the collateral behind the loans. Even if borrowers keep making payments, the cushion protecting lenders can shrink. In a downturn, that could mean bigger losses than the models suggest.
Liquidity adds another wrinkle. Many private credit funds are “semi-liquid,” meaning they offer periodic redemptions but can cap withdrawals if too many investors ask for cash at once. That's a safeguard for the fund, but it also means investors can't always get out when they want to.
What it means for your pension
For most people, the clearest exposure comes through retirement accounts. More pension money is being routed into private markets, including private credit. In the UK, the Mansion House Accord targets 10% of default workplace pension funds in private markets by 2030. In the US, target-date funds dominate many 401(k)s — Vanguard said 59% of participants in plans it administers held their entire account in a single target-date fund in 2024.
When underlying loans are valued with quarterly models rather than live prices, pension unit values can look smoother than the real “sell it today” price. That can be comforting, but it can also be misleading. The Financial Stability Board has flagged a “first-mover incentive”: if models still imply close to full value, early redeemers may get paid at that level, and any later write-down lands on those who remain.
So if losses show up, they may arrive late and in steps — on reporting dates or when funds tighten redemption terms — rather than gradually on everyday statements. That's a different experience from holding a public stock, where bad news often shows up in the price within minutes.
For investors, the takeaway isn't that private credit is doomed. It's that the smoothness of quarterly marks can hide real risk. As the AI data-center boom continues, keeping an eye on how these loans are valued — and how quickly marks catch up to reality — will be part of understanding what your retirement balance actually reflects.
Related reading: SAP's AI model strategy shows how even big tech is watching costs, and chipmakers are still betting on AI demand. But the private credit angle is one that pension savers should watch closely.
