Institutional Platform

PARTICIPANT CATEGORIES

Find the category closest to your requirements. Each shows typical reservation sizes and terms; reservations use standard capacity reservation agreements, with deposits held in escrow. Contact the Mazipan OS team at info@mazipanos.com to get started.

290 GW
Global data-center power demand by 2030*
7
Demand categories
3
Very High bundle fit
10
Target geographies

* Gartner forecast, up from 104 GW in 2025 and 132 GW in 2026 (press release, June 10, 2026).

Demand ▼
AI Model Training · 50–500 MW Sovereigns · 100–1,000 MW Global Bundle · 100–2,000 MW Industrial · Cloud-Adjacent · Enterprise
Filter by MW
AI Model Training
Very High
50 – 500 MW
1–5 year term
Frontier AI labs · GPU cloud operators · Model developers
Deterministic long-term compute; isolation; cost stability. Delivery windows, not fixed dates. Reservations can be made transferable.
IRLN eligibleTransferableDelivery windows
Sovereign AI Programs
Very High
100 – 1,000 MW
5–10 year term
National AI programs · Sovereign funds · Public research bodies
AI sovereignty; national security compute; cloud independence. Multi-site redundancy and credit support can be structured by agreement.
DFI co-financing eligibleCredit support by agreement
Industrial Compute
High
20 – 200 MW
3–10 year term
Energy · Pharma · Materials · Chemicals
Seismic modelling, drug discovery, molecular simulation. Stable power + compute + fiber for global collaboration.
IRLN eligibleLong-duration pricing
Cloud-Adjacent Buyers
High
10 – 200 MW
1–3 year term
GPU cloud providers · Capacity resellers · GPU startups
Cheaper than hyperscaler GPU pricing; dedicated capacity to resell globally via Global Bundle fiber routing.
Transferable CRAsShorter terms
Enterprises (Fortune 500)
Medium
5 – 50 MW
3–7 year term
Automotive · Finance · Retail · Manufacturing
Internal LLMs, R&D compute, data isolation, predictable cost. Dedicated compute with long-term power pricing.
Predictable pricingData isolation
Hedge Funds / Quant
Medium
10 – 100 MW
1–5 year term
HFT firms · Quant hedge funds · Risk-modelling firms
Simulation, risk engines, model training. Global fiber routing via subsea layer. Transferable CRAs.
Transferable CRAsGlobal fiber routing