Pegging Mechanism

USDAI’s value is explicitly tied to GPU compute power, with a peg defined as:

• 1 USDAI = $1 USD of GPU compute: This peg reflects the real-time market value of GPU compute resources, as determined by GPU.NET’s decentralized marketplace (Dapp.gpu.net). For example, if renting a high-end GPU for one hour costs $2 USD, 2 USDAI would cover that reservation.

How the Peg is Established

• Real-Time Market Rates: GPU.NET’s pricing oracles aggregate data from the network’s compute marketplace, tracking the cost of GPU resources across providers (e.g., data centers, independent GPU owners). These oracles pull live metrics such as GPU type (e.g., NVIDIA A100, RTX 4090), availability, and demand, ensuring the peg reflects current economic conditions.

• Tokenized Compute Credits: The collateral backing USDAI consists of compute credits, which are tokenized as real-world assets (RWAs) on Sol.gpu.net (GPU.NET’s Solana-based tokenization layer). Each credit represents a reserved unit of GPU capacity—measured in GPU-hours or equivalent metrics—that users can redeem USDAI for on the network.

Practical Example

Imagine a developer needs to train an AI model requiring 10 hours of GPU compute, valued at $1/hour in the marketplace. They would use 10 USDAI to reserve this capacity directly through Dapp.gpu.net. The peg ensures that the cost remains stable, regardless of fluctuations in crypto markets, because it’s tied to the underlying compute value.