# Stability Mechanisms

USDAI’s $1 compute peg is maintained through three interlocking stability mechanisms, ensuring reliability even under volatile conditions.

#### 1. Dynamic Pricing Oracles

* **Role**: These oracles provide **real-time feeds** from GPU.NET’s marketplace, tracking the cost of GPU compute across providers. They incorporate factors like GPU performance, energy costs, and regional availability.
* **Implementation**: Oracles are decentralized, pulling data from multiple sources (e.g., Dapp.gpu.net, external compute markets) and aggregating it via a median-based consensus to prevent manipulation.
* **Impact**: If compute costs rise to $1.10/hour, oracles signal smart contracts to adjust USDAI’s backing, maintaining the peg by recalibrating the collateral ratio.

#### 2. Algorithmic Supply Adjustments

* **Role**: Smart contracts **automatically mint or burn USDAI** to balance supply with demand, preventing deviations from the peg.
* **Process**:
  * If USDAI trades above $1 (e.g., $1.05), excess demand triggers minting to increase supply, pushing the price down.
  * If USDAI falls below $1 (e.g., $0.95), excess supply triggers burning to reduce circulation, lifting the price back to $1.
* **Example**: During an AI boom, demand for USDAI surges. The system mints 10,000 additional USDAI, backed by newly contributed computecredits, stabilizing the market.

#### 3. Proof-of-Compute (PoC)

* **Role**: Validators on Grid.gpu.net use **PoC** to confirm compute availability, ensuring USDAI is never over-issued or under-collateralized.
* **Mechanism**: Providers submit cryptographic proofs of their GPU capacity (e.g., benchmark results, uptime logs), which validators verify before credits are tokenized. This prevents fraudulent collateral claims.
* **Impact**: PoC ties USDAI’s supply to real, usable compute power, reinforcing trust in its backing and preventing systemic risks.