USD.AI targets AI infrastructure credit gap as neocloud lending risk rises

Traders Union
All News
Crypto
USD.AI targets AI credit gap

Highlights

USD.AI has originated $578.1 million in GPU-backed loans at an 11.5% average borrowing rate, generating $29.35 million in yield since April.
Investors face uncertainty over whether GPU collateral and sUSDai risk buffers can withstand correlated defaults or mass liquidations during a data center credit downturn.
A governance proposal aims to allow pre-maturity loan sales via GLX, but stressed market conditions could lead to withdrawal queues if exit liquidity dries up.

GPU-backed lending model gains attention

As reported by Weiss Ratings, tightening bank credit for independent AI infrastructure operators is increasing interest in alternative lenders that can underwrite loans against physical hardware.

USD.AI is positioning itself in that gap by originating and structuring loans backed by AI GPUs, using on-chain tokens to represent title to the machines. The protocol says depositors earn through its yield-bearing token sUSDai, with yield currently floored at 7% by its CHIP Hurdle Rate and supported by a reserve when loan income falls short.

Borrowers are paying an average rate of 11.5% on those loans, leaving a margin for operating costs and risk buffers. The protocol reports $578.1 million in total value locked, largely spread across 16 loans and two escrowed allocations, and says it has generated $29.35 million in cumulative yield since April.

The business case is tied to the rise of neocloud operators, which buy and install high-end AI chips and rent computing capacity to startups and research labs. Those companies face heavy capital needs and thin balance sheets, making them more vulnerable as banks pull back from lending tied to data center expansion and power-related execution risk.

Stress test concerns remain for investors

The main question for investors is whether GPU collateral alone can protect sUSDai holders if multiple borrowers fail at the same time. While the loan book is described as clean so far, the protocol has not yet faced a large physical liquidation during a broad data center credit crunch.

A sector-wide downturn could expose correlation risk if distressed chips hit the secondary market at once, pushing resale values lower just as recoveries are needed. Insurance and first-loss token buffers may soften losses, but long claims timelines, carve-outs and potential weakness in the CHIP token during a downturn could limit their effectiveness.

A governance proposal to let the protocol sell loans before maturity through GLX could add another exit route, but buyers for those loans may be scarce under stressed market conditions. That raises the possibility of withdrawal queues for sUSDai holders if several backstops come under pressure simultaneously.

The article argues that USD.AI is addressing a real financing gap where demand from AI operators meets investor appetite for yield tied to productive assets. Still, it says rapid growth does not prove resilience, and that an 11.5% average borrowing rate may be too narrow a cushion if neocloud defaults arrive in clusters rather than individually.

In our earlier coverage of the U.S. AI data center buildout, we noted that rising financing costs, project delays and power constraints were starting to undermine confidence in ambitious expansion plans. The piece highlighted that only a fraction of planned projects looked credible, with examples of cost overruns and capital shortfalls showing how quickly sentiment can turn when construction timelines slip and funding tightens.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.