Key Takeaways
- Mid-market neoclouds, often former crypto miners, bought GPU fleets using 70% to 80% loan-to-value debt packages while committing zero cash equity of their own.
- Down payments were funded almost entirely by demanding huge upfront prepayments from inference startups, rerouting venture capital equity straight into debt deposits.
- Private credit lenders are pulling back from 80% loan terms down to conservative 50% LTV terms, cutting off clouds valued under $1 billion to focus only on $20 billion AI labs.
- To prevent insolvencies, mid-market operators are scrambling to raise HoldCo equity and build secondary re-let markets modeled on commercial aircraft leasing.
How Mid-Market Neoclouds Financed Fleets on Zero Down
Over the past two years, dozens of former crypto mining facilities rebranded as specialized cloud providers. They promised fast access to clusters of high-end GPUs without the multi-month waitlists of major hyperscalers. Behind that rapid expansion sat an aggressive financial trick.
These operators purchased hundreds of millions of dollars in hardware by taking on 70% to 80% loan-to-value debt from private credit funds. Under normal corporate finance rules, the borrower supplies the remaining 20% to 30% as cash equity. Mid-market neoclouds bypassed that requirement by squeezing it directly out of their customers.
Stanley Lee explains what actually happened beneath the surface: “Not only is it $300 worth of equity, a lot of these mid-market neoclouds are not even putting in their own equity. They're using customer prepayments that they're charging the large inference providers for.” (play from 9:35)
In practice, early-stage AI startups took venture capital checks and wired them upfront to mid-tier clouds. The clouds then handed those prepayments to lenders as down payments for hardware. The operators took minimal operational risk with their own balance sheets while expanding their fleets on pure debt and customer cash.
Why Private Credit Is Squeezing Mid-Market Clouds
That financing model worked only while venture capital flowed freely and lenders accepted paper promises. Today, private credit lenders are changing the rules. They are pulling loan-to-value ratios down from 80% LTV to 50% LTV, doubling the cash down payment required to buy compute. At the same time, they are refusing to underwrite unproven mid-market operators.
Lee details the flight to quality among institutional lenders: “Lenders in the private credit space are doing the exact same thing. They're asking for lower LTV structures and higher upfront payments, and they're saying, 'Okay, we don't want to finance this company that maybe is worth $500 million or $1 billion. Let's keep it to the list of the five or six inference companies that are worth $20 billion.'” (play from 10:24)
This sudden shift puts smaller neoclouds in a vice. Inference startups cannot keep funding massive upfront checks because venture funds have tightened their belts. Without prepayments, operators must supply real cash equity to close hardware loans, forcing them into expensive corporate fundraising.
Lee summarizes the new operating reality: “The best way that I could say it is access to debt is going to equal access to equity in the middle market.” (play from 14:32)
What Hardware Finance Must Borrow From Aviation
Compute hardware cannot survive on ad-hoc credit tricks forever. Lee argues that compute infrastructure needs the same institutional plumbing that stabilized commercial aviation after past market shocks.
Airlines do not rely on fragile prepayments to finance Boeing and Airbus jets. The aircraft leasing industry relies on liquid secondary markets, independent equipment appraisals, and put hedges on residual value. If an airline defaults, the lessor repossesses the plane and re-leases it within days. Mid-market GPU hosts currently lack that secondary liquidity. Until standardized repossession and re-let contracts exist, smaller operators that overextended on high-LTV debt will struggle to survive.
What to Do With This
Audit your compute vendor contracts before wiring next quarter's compute deposit. Ask your provider's sales engineering team what debt package financed their cluster, and refuse to pay six-figure upfront prepayments unless the agreement includes an escrow structure or a non-disturbance clause protecting your workloads if their hardware lender seizes the machines.