GPU compute is becoming the core infrastructure layer of AI, but financing that infrastructure is still broken. Traditional lenders can’t efficiently underwrite small-to-mid GPU loans and the ABS market can’t move fast enough for hardware that depreciates in 3–5 years.In this episode of the DePINed Podcast, host Tom Trowbridge, Co-Founder & CEO of Fluence, sits down with Conor Moore, сo-Founder of USD.ai, a lending protocol building asset-backed GPU loans and turning them into a yield-bearing on-chain token.- Why GPU lending barely exists in TradFi and why private credit doesn’t serve the “missing middle”

- How on-chain rails make GPU credit “tradable day one,” bypassing slow securitization cycles

- What USD.ai’s loan terms look like: ~3-year duration, 70–80% LTV, ~7–15% rates depending on offtake quality

- Why non-recourse, asset-localized collateral (bankruptcy-remote SPVs + title) matters for compute operators🎧 Listen and watch the episode on YouTube, Spotify and Apple Podcasts — and follow DePIN Space for more materials with founders building real-world decentralized infrastructure.