Private credit just became the preferred term sheet for companies that need a hundred million dollars worth of GPUs yesterday.
The Summary
- Blue Owl structured a $2.8B debt deal for Iren to buy Nvidia GPUs, while Lambda separately raised $1B in private debt for its Nvidia chip collaboration with Microsoft.
- Both deals point to the same trend: AI infrastructure companies are bypassing traditional venture equity and tapping private credit markets to finance GPU purchases at scale.
- When you need billions in compute before your revenue catches up, debt is faster than dilution.
The Signal
Two major AI infrastructure debt deals dropped within 48 hours, and they tell the same story from different angles. Blue Owl led a $2.8B financing for Iren to acquire Nvidia GPUs, the kind of deal size that would have been a Series C equity round two years ago. Meanwhile, Lambda closed $1B in private debt specifically tied to its Nvidia partnership with Microsoft. Neither company is reaching for venture capital. They're reaching for credit facilities.
This marks a structural shift in how AI compute gets financed. GPUs are expensive, predictable, and revenue-generating. That makes them perfect collateral. A H100 cluster doesn't pivot. It doesn't have product-market fit risk. It prints money if you can keep it fed with workloads.
"Private credit in AI infrastructure funding highlights evolving financial strategies amid tech growth."
The Lambda deal is particularly instructive because it's explicitly tied to Microsoft collaboration. That partnership likely de-risks the debt from the lender's perspective. Microsoft isn't going to let a key infrastructure partner run out of compute mid-contract. The revenue visibility is there. The customer concentration risk is offset by the customer being one of the three companies that can actually afford infinite H100s.
Blue Owl's Iren financing is even larger and suggests institutional credit markets are now comfortable writing nine-figure checks for GPU acquisition. These aren't bridge loans. They're not growth debt with warrants attached. They're infrastructure financing deals that treat AI compute the way commercial real estate gets financed: as a hard asset that generates predictable cash flows.
Key financing dynamics at play:
- Debt preserves founder equity while scaling infrastructure
- GPUs as collateral are more tangible than software IP
- Partnership deals (like Lambda-Microsoft) reduce lender risk and improve terms
The Implication
If you're building in the AI infrastructure layer and you need scale fast, watch how these deals are structured. Equity dilution isn't the only path anymore. The credit markets are open, the terms are getting competitive, and lenders understand the asset class now. This matters especially for compute providers, data center operators, and anyone sitting between Nvidia and end customers.
Expect more of this. When the asset you're financing is literally the bottleneck resource for the entire AI economy, and when that asset has a secondary market and predictable depreciation curves, debt makes more sense than equity. The companies that figure this out first will own more of their upside.