How Does USD AI Work?

RWA|Risk C|5 mechanisms|4 interactions

USD AI is a novel DeFi lending protocol that converts stablecoin deposits into loans for AI companies, using NVIDIA GPU hardware as collateral through its CALIBER tokenization framework. With $571M in TVL and $13M in Series A funding from Framework Ventures, its C grade reflects the significant novelty risk of using physical hardware as DeFi collateral and the untested legal framework bridging physical and digital asset ownership.

TVL

$288M

Sector

RWA

Risk Grade

C

Value Grade

D+

Core Mechanisms

6.1.2

Novel

GPU hardware-backed lending via CALIBER tokenization

Physical GPU hardware tokenized as NFTs and used as loan collateral; novel asset class for DeFi lending

2.1.2

Interest income from AI infrastructure loans

Standard percentage-based fee on loans

7.3.1

Allo points system for future token airdrop

Standard points-to-token conversion model for pre-TGE incentives

6.4.3

Novel

Custom valuation oracle for GPU hardware assets

No standard oracle exists for GPU hardware pricing; protocol must maintain custom valuation methodology

2.2.1

Stablecoin yield distribution to depositors from loan interest

Standard interest rate pass-through to lenders

How the Pieces Interact

GPU hardware collateral valuationLending liquidation mechanicsHigh

GPU hardware depreciates rapidly and is highly illiquid to liquidate compared to crypto assets, creating risk of bad debt accumulation if borrowers default

Physical asset tokenization (CALIBER)On-chain lending protocolHigh

Legal disputes over physical GPU ownership or data center insurance claims could delay or prevent liquidation of collateral

Points/airdrop incentivesStablecoin deposit yieldsMedium

High yields attract mercenary capital seeking both APY and airdrop rewards; post-TGE withdrawal could create sudden liquidity reduction

AI sector demand for GPU financingProtocol yield sustainabilityMedium

Protocol revenue depends entirely on AI companies needing GPU financing; an AI sector downturn could simultaneously reduce loan demand and increase default rates

What Could Go Wrong

  1. GPU hardware collateral is a novel and illiquid asset class for DeFi lending; liquidation of physical GPUs during market stress could realize significant haircuts compared to crypto collateral.
  2. The CALIBER tokenization framework bridges physical hardware to on-chain NFTs under U.S. commercial law, introducing legal and jurisdictional risk not present in pure crypto protocols.
  3. High yields (up to 20% APY) on stablecoins depend on continued demand from AI companies for GPU financing; a downturn in AI investment could reduce loan demand and yields simultaneously.
  4. Pre-token protocol with points-based airdrop system creates uncertainty around future token distribution and potential sell pressure at TGE.

AI Sector Downturn with GPU Collateral Depreciation

Moderate

Trigger: Major AI sector correction causing GPU demand to drop 40%+ within 6 months, combined with next-generation chip release that obsoletes current hardware

  1. 1.AI investment cycle peaks and demand for GPU financing declines sharply New loan originations drop, reducing protocol revenue and yield for stablecoin depositors
  2. 2.Existing GPU collateral depreciates as new chip generations launch CALIBER-tokenized GPUs backing loans become undercollateralized as hardware values fall 30-50%
  3. 3.AI company borrowers default on loans as their revenue projections fail Protocol attempts to liquidate GPU collateral but faces illiquid secondary market for used hardware
  4. 4.Liquidation proceeds cover only 40-60% of outstanding loan values Bad debt accumulates, reducing effective yield and triggering depositor withdrawals
  5. 5.Depositors withdraw stablecoins as yields drop and solvency concerns emerge Protocol TVL contracts by 50-70%, leaving remaining depositors bearing disproportionate losses

Risk Profile at a Glance

Mechanism Novelty9/15
Interaction Severity6/20
Oracle Surface5/10
Documentation Gaps4/10
Track Record3/15
Scale Exposure7/10
Regulatory Risk6/10
Vitality Risk3/10
C

Overall: C (43/100)

Lower score = safer

More on USD AI

Related RWA Explainers