Loading
Loading
Transparent pricing built from real unit economics — not marketing numbers. Every rate is derived from total cost of ownership modelled at multiple utilisation scenarios, validated against operational costs and published only when capacity is confirmed.
NOVACORE Compute pricing is built on a foundation of honest unit economics. We model the total cost of ownership for every GPU configuration — acquisition, delivery, financing, rack infrastructure, networking, storage, software licensing, power, cooling, facility costs, support staffing, insurance and maintenance. We then calculate gross revenue across four utilisation scenarios and derive a contribution margin that ensures the business is sustainable at every level of demand. Published prices reflect real costs, not aspirational marketing rates that assume 100% utilisation. When capacity is confirmed and operational readiness is achieved, rates are published transparently by offering type. Until then, we share our pricing framework and methodology openly and invite prospective customers to shape reserved-capacity terms through early engagement.
Every cost line is accounted for: GPU and server acquisition, delivery and import duties, financing costs, rack space and power distribution, InfiniBand and Ethernet networking, NVMe storage and parallel filesystem licences, GPU software and driver licensing, power draw and cooling overhead, facility operations and security, 24/7 support staffing, equipment insurance, and scheduled maintenance and component replacement. Nothing is hidden or amortised into a fuzzy 'overhead' line.
Available GPU hours multiplied by expected utilisation rate multiplied by the price per GPU hour. We model gross revenue across multiple utilisation bands — not a single optimistic assumption — to understand the revenue range and identify the break-even point for each configuration. Revenue projections are validated against customer commitments and market benchmarks before capital is deployed.
Gross revenue minus direct operating costs — power, cooling, support staffing, licensing and maintenance — yields the contribution margin per GPU configuration. This figure tells us whether each unit of capacity generates positive cash flow and how sensitive that cash flow is to utilisation changes. Configurations that cannot sustain positive contribution margin at conservative utilisation assumptions are not deployed.
Every configuration is stress-tested across four utilisation scenarios — 30%, 50%, 70% and 85%. The 30% scenario represents a conservative baseline that must still cover fixed costs. The 85% scenario represents near-full utilisation and informs pricing discounts for reserved capacity. Sensitivity analysis identifies which cost drivers have the greatest impact on profitability and where operational efficiency investments yield the highest return.
| Offering | Structure | Note |
|---|---|---|
| Dedicated server | Monthly fee per server | Published once operational |
| Reserved capacity | Committed monthly, quarterly or annual term with volume discount | Improves financing predictability for both parties |
| Virtual instances | Hourly or daily, pay-as-you-go | Available post general availability of multi-tenant platform |
| Managed clusters | Platform fee plus compute, tiered by management scope | Available post general availability |
Prices are published only when capacity, power and operational support are confirmed. We do not publish aspirational rates that assume 100% utilisation or ignore the cost of capital. Enquiries are welcome now to inform the pricing framework and reserved-capacity terms — early partners receive preferential rates and direct input into how we structure pricing for each offering type.
Secure AI and high-performance computing for enterprises, governments and research.