GWRTs: One Trick Unlocks a Trillion Dollar Market

At USD.AI, one of our main innovations was deceptively simple: we changed a single word.

That change unlocked a trillion-dollar commodities market for a brand new asset class.

So what’s the word? We turned a datacenter into a warehouse.

From warehouses of grain to warehouses of compute

Datacenters may look like ordinary buildings filled with racks of GPUs and servers, but at their core they function like warehouses for high-end computing hardware.

And that legal distinction matters.

In U.S. law, warehouses carry special privileges. They enable global trade finance by issuing receipts for stored goods, oil, steel, grain, that can be pledged, financed, and traded at scale.

The Uniform Commercial Code (UCC) governs these transactions across all 50 states and territories, ensuring consistency and enforceability even across jurisdictions.

Under Article 7 of the UCC governs bailments, called Documents of Title, which make commodities financeable. A bailment is just a IOU for whatever its issued against.

In 2023, UCC Article 12 was amended to explicitly include electronic documents of title, with NFTs recognized as legally valid receipts. For the first time, digital tokens could directly represent ownership rights in physical assets with full legal certainty.

And so here was our breakthrough.... although datacenters have existed for decades, no one had ever classified them as warehouses under UCC Article 7 or had they ever issued NFT warehouse receipts for GPU's.

By doing so, we’ve unlocked a framework that makes GPUs financeable like oil or grain, opening an entirely new trillion-dollar asset class for on-chain markets.

The mechanics

When a business installs GPUs in a datacenter, those GPUs can be converted into GPU Warehouse Receipts (GWRT's) through bailment using USD.AI. Whoever holds the receipt holds legal title to the hardware.

Borrowers use these GWRT's to draw liquidity directly from the protocol, taking stablecoins as a loan.

Once the loan is paid back, they can then remove their GWRT from USD.AI and reclaim their hardware. It's that easy.

Learning from DeFi

The model we're using draws direct parallels other onchain lending markets like Aave which allow borrowers to access securitized lending against a generalized lending pool.

This design is powerful because it:

• Abstracts counterparty risk
• Enables fungibility and liquidity
• Scales efficiently

By issuing GWRT's, GPUs can effectively be treated as commodities. The digital ticket, the NFT, stands in for the physical GPU, just as aTokens stand in for deposited ETH in Aave.

Once a GPU owner deposits their GWRT into the protocol, they no longer have free access to the hardware. The GWRT is held by the protocol, locked programmatically until the loan is paid off.

This ensures that the collateral is secured. The borrower can continue to use those GPUs to generate revenue, but they cannot remove or repurpose the hardware elsewhere until their debt is repaid.

That lock-in is what gives USD.AI its security, the same way Aave ensures lenders always have a claim on ETH or BTC, USD.AI ensures lenders always have a claim on GPUs to be repaid. Our model uses the best of DeFi to create a brand new securitized lending model for a nascent, expontially fast growing industry.

Wrapping up

The UCC was written to make oil, copper, and grain financeable at global scale. With the 2023 update recognizing electronic documents of title, that same legal foundation now applies to digital receipts, including NFTs.

USD.AI takes this framework and applies it to the most critical resource of our era: compute.

By treating datacenters as warehouses and GPUs as warehoused goods, we are turning GPUs into a new commodity class. One that can be financed, traded, and liquidated with the same legal certainty as steel or wheat.

The vision is simple but powerful: GPUs are the backbone of our AI future.

We don’t care what model is being trained, or whether the compute is centralized or decentralized. We want to serve everyone, from operators running $500,000 of GPUs to institutions deploying $500 million.

The only way to do that is to commoditize GPUs themselves as GWRT's, creating a liquid, financeable market for compute.

This is the one shift, datacenter to warehouse, hardware to commodity, that transforms GPUs into the oil of the 21st century.

USD.AI unlocks the capital flows needed to power the next generation of AI.