GPU Network Model

01 / Silicon Flywheel

1. Users burn GPU + pay an ETH fee to deploy virtual nodes (Wafer Fabs or Cleanrooms).
2. Every mint burns 70% of the GPU cost to the burn wallet, routing the other 30% into the treasury wallet. Compound upgrades and crates split their GPU cost 50/30/20 burn / reserve / treasury.
3. A halving emission curve (E₀ = 20.7 GPU/sec, halves every 14 days) distributes GPU from the 50M reserve. The rate is derived from the reserve rather than fixed, so the schedule spends it exactly and can never promise GPU the protocol does not hold. Each user earns a share proportional to their grow-power, capped at 30% per user to prevent lottery-in-thin-network wins.
4. Under v2 accrual, both Wafer Fabs and Cleanrooms accrue $GPU per second out of that reserve. Progression is wallet-wide: warehouse upgrades (GPU + 0.00001 ETH, 12h cooldown) raise your Warehouse Level, unlocking more node slots, more daily crates, and higher rarity pools. Cleanrooms add bonus node slots at L5+.
5. Protocol ETH revenue (ERC-20 transfer tax (2%) + DEX LP fees (2%)) flows to the treasury ops budget — it funds infrastructure, not user rewards. User accrual is GPU-only from the halving reserve.

02 / Production-Line Economics

Loading…

03 / Network Routing Costs

04 / Yield Routing Diagram

Launch: 1B GPU minted by Flap to the bonding curve
        (fixed supply — the token contract has no mint function)
                │
       ┌────────┴────────┐
       ▼                 ▼
 950M public float       50M acquired at genesis
 (95%, trades freely)    (5%, funds all rewards)
                                 │
  ┌─────────────────────────────┐│
  │   GPU Emission Reserve      │◀┘
  │   50M GPU                   │◀── reserve split on in-game
  └──────────────┬──────────────┘     spends tops the pool back up
                 │  halving curve E(t) = 20.7 × 0.5^(t/14d)
                 ▼
    Each user's per-second rate:
    share = min(userGP / totalGP, 30%) × welcomeBoost(elapsed)
                 │
                 ▼
  ┌─────────────────────────┐       ┌──────────────────────────┐
  │  Wafer Fab claims         │       │  Cleanroom claims     │
  │  pay GPU                │       │  pay GPU (compoundable)  │
  └─────────────────────────┘       └──────────────────────────┘

  Separately: Protocol ETH revenue (ERC-20 tax 2% + LP 2%) → treasury ops
  (XOMX/CVXX swap path is a legacy ops-side flow, not a user-rewards path)

05 / GPU Emission Clock

Global GPU emission follows a Bitcoin-style halving curve. Starting at E₀ = 20.7 GPU/sec at genesis, the rate halves every 14 days until the reserve is fully paid out.

E(t) = E₀ × 0.5 ^ (t / 14 days)

Day   0 :   20.7 GPU/sec  (1.8M/day)
Day  14 :   10.3 GPU/sec  (892,857/day, 50% of lifetime emitted)
Day  28 :    5.2 GPU/sec  (446,429/day, 75% of lifetime emitted)
Day  42 :    2.6 GPU/sec  (223,214/day, 88% of lifetime emitted)
Day 93 : ~0            (99% emitted — tail is negligible)

Lifetime emission = 50M GPU — exactly the 5% of the 1B supply held in the Emission Reserve. The remaining 950M (95%) is public float that trades on the Flap curve. Supply is fixed at launch — the token contract has no mint function, so no new GPU can ever be created.

Each user earns a proportional share of each second’s emission:

user_rate = min(user_gp / total_network_gp, 30%) × E(t) × welcome_boost

user_gp = Σ componentMult for each node (Formula D: base × Π rarityBoost)

A 30% per-user share cap prevents lucky-in-thin-network lottery wins, keeping the experience fair across network sizes.

05B / New-Fab Accelerator

New users receive an 8× share multiplier that linearly decays to 1× over their first 72 hours. This is critical for latecomers joining mid-cycle when the halving curve has already decayed.

boost(elapsed) = 1 + 7 × max(0, 1 − elapsed / 72h)

Hour 0  : 8.00×
Hour 24 : 5.67×
Hour 48 : 3.33×
Hour 72 : 1.00× (boost expired)

The boost applies from your first mint. Separately, the one-time welcome stipend unlocks only after reaching Warehouse L4 — that gate is what stops $100 × 10 alt-wallets from draining the welcome allocation.

05C / Reserve Safety Interlock

Orthogonal to the halving, a runway-based throttle factor f ∈ [0, 1] protects against pathological drain on legacy flat-rate families (not used for GPU under the halving model, but retained for any future secondary-asset families):

runwayDays = reserveBalance / unconstrainedDailyBurn

runway ≥ 45d  →  f = 1.0      (healthy)
14d → 45d     →  f 0.7 → 1.0  (gentle)
7d  → 14d     →  f 0.3 → 0.7  (notable)
3d  → 7d      →  f 0   → 0.3  (emergency)
≤ 3d          →  f = 0        (halted)

Under the halving model for GPU, the emission reserve is pre-minted and cannot deplete beyond lifetime emission, so f = 1.0 effectively always. When paused by admin, f is forced to 0 across all families.

06 / Campus Expansion Tiers

Progression is wallet-wide. Each warehouse level unlocks more node slots per family, a higher daily crate limit, and pricier crates. Upgrades cost GPU (split 50/30/20 burn / reserve / treasury) + 0.00001 ETH, on a 12h cooldown (0.005 ETH expedite skips it). Cleanrooms get bonus facility slots on top: +2 at L5, +3 at L7, +4 at L9. Rarity pools unlock by level too — Legendary at L4, Mythic at L6, Divine at L8.

07 / Process Aura Spectrum

Each node’s emissive material color shifts with its level, making progression visible at a glance in the 3D scene.

08 / Live Network Telemetry

Loading live state…

See Treasury Core for the full event feed and Chip Exchange for aggregated metrics.

09 / Protocol Source Matrix

All constants on this page are imported from @gpu-fab/protocol and @gpu-fab/silicon-engine, and the live family config comes from GET /api/nodes/families, so this page can never drift from what the backend actually enforces. Admin config changes appear here on the next reload.