gUSD

Protocol

Architecture · resolved mechanics only

01

Architecture

how the Index reaches the market
  GPU PRICING SOURCES
  provider observations, USD per GPU-hour
              │
              ▼
      ORACLE · INDEX  ───  gUSD DATA
      weighted reference     published series
              │
              ▼
        gUSD HOOKS  ·  Uniswap v4
        permissioned logic on every swap
              │
              ▼
     gUSD MARKETS  ·  4 GPU classes
     market price ⇄ reference = basis

Provider observations feed the Index; the Index publishes per epoch; gUSD hooks carry that reference into Uniswap v4 pools so every GPU market trades against a live Index reference.

02

Markets & pools

one pool per GPU class
PairVenueHookStatus
H100 / gUSDUniswap v4gUSD hookLIVE
H200 / gUSDUniswap v4gUSD hookPLANNED
L40S / gUSDUniswap v4gUSD hookPLANNED
RTX4090 / gUSDUniswap v4gUSD hookPLANNED

Each GPU class is its own pool against gUSD — no basket, no shared curve. The pools execute onchain: H100 is deployed and trading from primary issuance today, and the rest of the catalog deploys with the rollout.

03

The hook

why v4

Uniswap v4 pools run their logic through hooks — code attached to the pool that participates in every swap. gUSD uses that seam to keep each GPU market anchored to the Index rather than drifting as a standalone AMM.

The hook is the difference between a pool that merely prices GPU-hours and a market that quotes them against a published reference. The basis — the gap between the two — is the product's central read.

Hook parameters finalize with the protocol contracts; this page describes the resolved shape, not implementation detail.

04

Principles

standing design decisions
  • Not a standalone AMMEvery market is anchored to the Index through the hook; pricing is a dialogue between flow and reference, never an island.
  • Not CDP-basedgUSD is not minted against over-collateralized debt positions. Issuance and the markets it settles follow the compute economy directly.
  • Issuance follows demandSupply responds to the GPU markets themselves — compute demand sets the shape, not a governance dial.
  • LPs are centralLiquidity providers are the protocol's counterparties and first citizens; the products around earning and minting exist for them.