Two major trends appear to be converging: private lending for AI capital expenditures and the tokenization of real-world assets (RWA). This week, Bullish, an institutional crypto platform listed on the NYSE and the parent company of news outlet Coindesk, announced a $100 million loan facility to USD.AI. The company plans to use these funds to provide loans to businesses involved in artificial intelligence (AI) development and related infrastructure.
Essentially, AI companies are using their extremely valuable computer chips (Graphics Processing Units, GPUs) as collateral. Bullish believes the capital involved in this new form of lending is enormous, noting that it actually surpasses the volumes of many older, more conventional types of credit instruments.
"Bullish is targeting a capital-intensive sector that has quickly become one of the largest in private lending, and its scale surpasses traditional borrowing markets such as auto loans and home equity lines of credit (HELOC)," the company stated.
Bullish Bets on Bringing Real-World Assets onto the Blockchain
Block.one co-founder and CEO Brendan Blumer founded Bullish in Hong Kong in 2020. Following Friday's announcement, the company's shares on the NYSE fell by 2%, but they gained 10.5% over five days and have risen more than 44% over the past month.
Thomas Cowan, Head of Tokenization at Bullish, stated that his company is firmly convinced that real, physical objects should be connected to digital money systems, and this deal reflects that stance.
"Our involvement in the USD.AI project reflects a belief we have held since our first investment in this protocol: reliable, well-structured real-world assets should be on-chain," Cowan noted. "USD.AI's on-chain transparency allowed us to evaluate this mechanism with the same institutional rigor we apply across our platform, and its support is an important step towards bringing tokenized assets to the institutional level."
GPU Collateral Faces a Crucial Stress Test
At the same time, this concept comes with certain risks. GPU prices can fall very rapidly. Furthermore, they quickly become obsolete as newer, more advanced models are released. As of Saturday, August 29, 2026, GPU prices are high, especially for high-memory Nvidia consumer graphics cards, and their cost is increasing again this summer rather than declining. The real test will come if a borrower is unable to repay their loan.
end-content




