Deal Structure Overview

WWT finances GPU racks + cooling for WorldLink Chandragiri co-location. Earns on three streams: EPC margin, financing spread, and future Markichowk supernode purchase rights.

🔄 Three-Party Deal Flow
🏢
WWT
EPC + Financier
Procures & finances
GPU racks + cooling
Kaiju Labs
Operator + SPV
Deploys, manages,
sells GPU compute
🏦
Off-Takers
Revenue Source
F1Soft, IncessantRain,
Banking Consortium
📜 What WWT Provides
🎯 What WWT Gets
📅 Critical Path Timeline
LOI Target: Mid-May 2026

CapEx Scenario Builder

Break down all costs for Phase 0A and reconcile against Nick's target deployment budgets.

📊 CapEx Breakdown
Component Amount % of Total
⚙️ Adjust Parameters
GPU Rack Cost $3.7M
WWT EPC Margin 10%
Contingency 5%
NVAIE $/GPU/yr $4,500
Colo Monthly/Rack $8,000
🍩 Cost Composition

Revenue & DSCR Modeler

Model revenue scenarios and test bankability thresholds. DSCR ≥ 1.35× is the minimum for equipment financing.

💧 5-Year Projection
🔥 DSCR Sensitivity — Price × Utilization
Green = Bankable (≥1.35×)
🎚️ Revenue Parameters
GPU-Hr Price $10.50
Utilization 80%
Ramp-Up (months) 3
Take-or-Pay Floor 70%
Annual Escalator 3%
💳 Financing Parameters
SOFR Rate 4.3%
WWT Spread 4.0%
Loan Term (years) 5
--
Year 2 Committed DSCR
Bankability threshold: ≥ 1.35×

WWT Value Proposition

Why this deal makes financial sense for WWT — triple-earn structure with massive future upside.

💎 Three Earnings Streams
📊 WWT Earnings Breakdown
🏔️ Markichowk Future Purchase Rights (The Real Prize)
Supernodes Committed 2
Supernode Cost $85M
Reseller Margin 8%
🎤 Key Talking Points for Tanuj

Off-Taker MSA Calculator

Model off-taker commitments and test whether contracted revenue supports bankable DSCR for equipment financing.

🏢 Off-Taker Commitments
Name Racks $/GPU-Hr Min Util %
🥧 Capacity Allocation
--
Committed DSCR
From contracted take-or-pay only
💧 Revenue Waterfall
📏 Pricing Guidance
📚 Research Sources — NotebookLM Deep Research (172 Sources)
Validated 2026-04-22
Key research areas: GPU equipment financing structures, NVIDIA Financial Services (NVFS), DSCR requirements for equipment-backed lending, take-or-pay bankability, MPA frameworks, WWT as NVIDIA Elite Partner, VAR/EPC margin benchmarks, SPV off-balance-sheet structures.

🏷️ Token & GPU-Hour Pricing Structure

GB300 NVL72 inference economics — what we charge, what it costs, what the market pays

Research-Validated — 172 Sources NotebookLM Deep Research × 2
🔄 GPU-Hour → Token Price Conversion
Kaiju Labs Pricing
💰 Cost Per Million Tokens — H100 vs GB300/NVL72
Benchmark Data
☁️ API Provider Price Comparison
Self-Hosted Advantage
📊 Batch Inference Economics
Utilization Impact
📐 Methodology, Math & Assumptions
Click to expand

⚡ Hyperscaler & Neocloud Comparison

GPU-hour pricing landscape, ROI analysis, and training economics across the competitive field

Source: APAC GPU-as-a-Service Infrastructure Report Source: Blackwell-Scale AI Infrastructure Report
🟢 H100 Pricing Landscape (Legacy Tier)
Value Tier
🔵 Blackwell GPU Pricing (GB200 / GB300 / B300)
Next-Gen Premium
🎯 ROI & Unit Economics
Investment Returns
🧠 Training Economics — NVL72 vs H100 Clusters
Architecture Advantage
🚫 Egress Cost Analysis — Zero-Egress Advantage
Cost Elimination
Token Velocity & Job Completion Time
Speed Advantage
🚀 Performance Multipliers — Blackwell Optimization
Hardware + Software Gains
📐 Methodology, Math & Assumptions
Click to expand