GPU Loans
Questions about USD.AI's GPU financing service, how they work, and how they generate yield.
How many loans are live?
You can check can the status of all the loans on the “Reserves” page of the USD.AI app https://app.usd.ai/reserves
Why aren't there more loans?
The loan process can take as little as a few days or as long as a few months depending on its complexity and in these early stages, the protocol is being more prudent to ensure users and USD.AI is not negatively impacted.
The 2 main stages of a loan are:
Underwriting = evaluating business, assets, legal, etc
Install = receiving hardware, installing, verifying
You can read more about the onboarding/underwriting process at 1. CALIBER [Yield]or 2. FiLo Curator [Scale] and can also view the current status of any upcoming loan on our Reserves page
GPUs depreciate fast. How is that risk handled?
GPU depreciation is addressed through conservative credit structuring, not long-term asset assumptions. GPUs typically depreciate ~15–20% annually, which is mitigated by:
Low loan-to-value (~75%), providing a ~25% equity cushion
Amortizing loan structures (e.g. ~36 months) that reduce exposure over time
A focus on installed, revenue-generating GPUs, not speculative hardware
The existence of an active secondary market that supports recovery in liquidation scenarios
The protocol underwrites GPUs as productive infrastructure with near-term cash flow, not as long-duration assets held for appreciation.
Last updated