USD.AI Secures $10 Million Loan for AI Infrastructure Using B300 GPUs

USD.AI has executed a new $10 million (approximately 135 billion KRW) loan for enterprise AI infrastructure development. The funds will be used to deploy an AI computing platform utilizing 192 NVIDIA B300 GPUs. BlockMedia reported that USD.AI implemented the loan on May 15th (local time) to support high-performance AI infrastructure for enterprise workloads, with a stated interest rate of 13%. Details about the lending platform's company name, loan maturity, and collateral structure remain undisclosed.

GPUs are essential devices for processing large-scale computations required for AI model training and service operations. AI infrastructure encompasses not just GPUs but also servers, data centers, power, and network resources that support computational needs. USD.AI operates a lending structure connecting on-chain liquidity with AI computing demand using GPUs as collateral.

The company offers funding in individual GPU deployment units to support the construction of AI infrastructure. As AI model development and service operations expand, initial capital demands for securing computational resources increase. Data centers and GPU facilities require significant upfront costs with long payback periods, typically involving equity investment, loans, and project financing together.

USD.AI has previously executed 17 loans totaling $279.9 million (approximately 377 billion KRW), including 15 active loans with a principal amount of $180.8 million (approximately 243.5 billion KRW). Two escrowed loans totaling $91.6 million (approximately 123.4 billion KRW) are currently held in escrow. The average interest rate for executed loans is 11.6%. Escrow refers to funds held separately until transaction conditions are met.

Active loans and escrowed loans differ in funding execution stages, making direct comparison difficult. GPU-backed lending involves financing secured by AI infrastructure equipment as collateral. The value of the collateral equipment, actual operational revenue, and loan terms can affect repayment feasibility. USD.AI's GPU-backed lending structure was previously included in Bitwise's tokenized physical asset vault collateral assets.

This loan represents an application of GPU-backed lending for enterprise AI infrastructure. Publicly disclosed information does not include details about the borrowing platform's operations, loan maturity, or collateral terms.