Map customers, size capacity, and underwrite the build — in one tool. Pick your accelerator and installed capacity, deploy solutions from the catalog, check the national demand map, then take the portfolio into bankable economics. The planner accounts every workload against your data-center capacity in real time — committed GPUs, headroom, revenue, and bankability. Per-unit usage and throughput calibrated against real, named deployments (CoreWeave, NVIDIA DeepStream, BMW, BofA, Nuance, Aramco, SDAIA — see docs/18). Whole-enterprise sizing re-validated Jul-2026 against named on-prem deployments: BNY (dozens of DGX H100 ≈ 250–500 kW), Mayo Clinic (128-GPU SuperPOD), SK Telecom (1,040 A100), Aramco Dammam-7 (7,900 GPU), Princeton (300 H100). Enterprise procurement quantum = 32-GPU BasePOD ≈ 42 kW.
Infra revenue and addressable solution opportunity are potential / uncontracted — not booked revenue. Bookable revenue requires signed customer offtake (Economics → Bankability).
1M tokens — loading…
KSA addressable demand map
Two views: one typical customer per segment (the catalog's calibrated per-unit values × typical quantities) and national TAM = typical customer × addressable KSA customers in that segment (counted from the named-target lists). Burst workloads counted at 35% steady-equivalent. Set your capture assumption to size capacity against the market. Scope: domestic enterprise + sovereign demand only — excludes hyperscaler/AI-export offtake (HUMAIN·AWS·Oracle-scale international capacity), which is quote-driven and sits on top of this map.
Demand by segment (steady GPU-eq at typical scale)
Anchor-tenant ranking (bankability view)
Ranked by anchor value = GPU demand × contract quality. Grade reflects offtaker credit + realistic contract tenor — this is what lenders underwrite. Chase A-grade anchors first; fill with B-grade; monetise burst as premium on-demand.
Economics levers
12-year project DCF on your configured capacity (installed MW × accelerator from the Planner). Engine: v2 benchmark-calibrated project finance — S-curve IDC, DSCR-sculpted senior debt (refresh-aware), generation-aware GPU pricing (4-yr contract lock, yr-5 refresh on 60% GPU-backed tranche), KSA SEZ/standard tax. Bankable threshold: min DSCR ≥ 1.35.
Split your installed MW across four models and see the blended deal live. Shell: building + power only, tenant does the rest. Colo: full facility, tenant brings GPUs ($/kW-mo, power passed through). Condo pods: you build, customer buys their pod (upfront $/kW) + pays you to operate it. GPU cloud: you own the GPUs, sell $/GPU-hr (uses the levers above). Colo/shell-anchored mixes borrow cheaper (−40bps per 100% RE share).
0%
65%
10%
25% (remainder)
$180
$80
$5,500
$80
Benchmarks: US wholesale colo >$200/kW-mo (CBRE H2-25, NoVA $215) — KSA rate unverified, get local quotes; stabilized colo trades at 5.5–7% cap rates (JLL) → the “stabilized RE value” card shows the development-margin play. Condo sale price ≈ your facility cost share + margin. Condo/colo tenants own GPU risk; cloud share carries refresh + erosion.
Site & capacity summary
CapEx & financing
Returns (12-yr DCF)
equity cash flow by year (yr 0 = equity invested; refresh in yr 5)
Bankability & scenarios
—
Sensitivity analysis
Equity IRR and min DSCR vs one lever at a time (current build; DSCR < 1.35 in red). Illustrative, from the underwriting model (docs/19).
Profit optimizer
Searches accelerator × gearing × contracted-utilization × price × PUE × power tariff × tax regime × debt cost — 6,048 configurations — for the highest equity IRR that still clears a bankable min DSCR ≥ 1.35. Objective, constraints and lever sensitivities follow project-finance best practice (docs/19).