Sharon AI, an Australian high-performance computing company, said Thursday it had secured up to $500 million in financing from blockchain-based lender USD.AI to scale its GPU infrastructure across the Asia-Pacific region.
The funds will support the deployment of compute systems used to train and run large artificial intelligence models, including an initial $65 million rollout expected to begin this quarter, the company said.
The agreement, per a press release shared with CoinDesk, enables Sharon AI to access capital through a non-recourse credit facility, meaning loans are backed by physical GPU hardware rather than the company’s corporate assets.
USD.AI’s onchain lending system turns verified GPU deployments into tokenized collateral, allowing lenders to track performance directly without the need for traditional credit checks, according to the release.
The structure is designed to help AI infrastructure providers grow faster while sidestepping slower bank and private market financing, and shows how tokenization has been powering the private credit market, the release said.
Maple Finance CEO Sidney Powell told CoinDesk earlier this week that private credit may be the breakout use case for tokenization, the process of representing real-world assets on a blockchain.
The private credit market has limited liquidity and often little transparency, making it ideal for a blockchain-based solution. Powell said, putting real-world loans on a blockchain could make the industry safer and easier to access by improving price discovery and reporting.
Powell, who will speak at CoinDesk’s Consensus Hong Kong next month, said he believes defaults in onchain credit markets will eventually prove the strength of blockchain rails. Transparency around repayments and asset quality could reduce fraud and make it easier for traditional investors to enter the market, particularly as crypto-backed loans begin to get rated by major credit agencies, he added.
USD.AI has now approved more than $1.2 billion in similar GPU-backed facilities for other AI infrastructure firms, including QumulusAI and Quantum Solutions.