Messari's Unqualified Opinions How USD.AI Disrupts Private Credit with On-Chain GPU Financing
Feb 17, 2026
Conor Moore, co-founder and COO of USD.AI, builds on-chain financing and tokenized debt for AI GPU infrastructure. He discusses originating GPU-backed loans, focusing on NVIDIA H100/H200 collateral. Topics include why GPUs were chosen, Munich Re-backed value insurance, on-chain execution for tradable debt, and scaling originations toward institutional liquidity.
AI Snips
Chapters
Transcript
Episode notes
Focused GPU Lending Beats Generalized RWA Play
- USD.AI pivoted from generalized RWA lending to GPU-only financing because focused collateral reduces adverse selection and improves product-market fit.
- Conor Moore traced iterations across NFTs, land, watches, then shifted to GPUs after AI CapEx boom and regulatory clarity made tokenizing GPU title viable.
Startup Tested Many RWAs Before Landing On GPUs
- Early USD.AI experiments covered tokenized land, gold bars, watches and NFTs before the team discovered GPUs fit legal and market constraints.
- Conor recalled regulatory headwinds during the Gary Gensler era which made broad RWA offerings suffer from adverse selection.
GPUs Are Better Financed As Project Loans
- Traditional banks avoid financing GPUs and private credit funds demand large deals and complex docs, leaving a gap for streamlined secured GPU loans.
- USD.AI treats GPU buys as project finance: tokenized title, escrowed in high-security data centers, and boilerplate loan terms for speed.