What Wholesale Colocation Is and What It Isn’t
Scale: 100 kW to 5 MW+
Wholesale starts where retail becomes operationally expensive – around 100 kW of committed power. At this threshold the pricing model changes: you’re buying power capacity, not rack space.
Pricing model: $/kW, not $/rack
Pricing is quoted per kilowatt of committed power per month. Typical ranges: $80–$150/kW in secondary markets (Texas, Pacific Northwest, Midwest); $130–$200/kW in primary markets (Ashburn, Silicon Valley, NYC/NJ).
Contract structure: 3–10 year terms, NRC + MRC
Deals include a Non-Recurring Charge (NRC) for fit-out and a Monthly Recurring Charge (MRC) for power, cooling, and services. Expansion options – the right to take additional power at pre-agreed rates – are a critical negotiation point retail deals don’t involve.
It is NOT hyperscale
Hyperscale (10 MW+) is built for AWS, Azure, GCP, and Meta. Wholesale colocation is for enterprises, SaaS platforms, AI operators, and neocloud providers deploying their own infrastructure at significant but not cloud-provider scale.
Typical Wholesale Colocation Specifications:
| Specification | Typical Range |
| Rack count | 20 – 200+ racks |
| Power density | 5 – 80 kW per rack (varies by cooling type) |
| Total committed power | 100 kW – 5 MW+ |
| Connectivity | 10G – 100G |
| Contract terms | 3 – 10 years |
| Pricing model | $/kW/month (MRC) + one-time fit-out (NRC) |
Who Is Actually Buying Wholesale Colocation in 2026
| Organization | Situation | Needs |
| Enterprise IT, consolidating from on-prem | 3–5 aging on-premise data center closures. Moving 200–500 racks to wholesale colo to reduce CapEx. Board wants a 5-year cost model. | Dedicated suite with guaranteed power delivery timeline. 5–7 year term with CPI-capped escalators. Documented expansion path for years 3–5. |
| SaaS platform, cloud repatriation | Cloud bill crossed $1M/month. CFO approved infrastructure CapEx. Need 500 kW to 2 MW of wholesale space. Timeline: deployed in 6 months. | Wholesale provider with available power in primary carrier market. Fit-out in 90 days or less. Cross-connect density for cloud on-ramps. |
| AI / neocloud operator | Building GPU cluster for inference at scale or AI training. 1–5 MW of high-density power required. Liquid cooling mandatory. Private suite required. | Operators with DLC or immersion-capable infrastructure. A+B power at 30–80 kW per rack. Power delivery in 60–120 days. Expandable to 10 MW over 24 months. |
| Content / streaming platform | CDN infrastructure requiring multi-site presence. Each site: 200–500 kW. Low latency to major metros. Carrier diversity is non-negotiable. | Wholesale options in 4–6 US metros simultaneously. Carrier-neutral facilities. Consistent contract terms. Negotiated rate card for expansion. |
| Managed service provider / hosting company | Building owned infrastructure to serve enterprise customers. Need 250 kW–1 MW to achieve unit economics for competitive managed service pricing. | Wholesale $/kW that pencils out for their margin model. Minimum 5-year term. Ability to provide managed services from within the suite. |
| International company entering US market | APAC or European HQ expanding to US infrastructure. Need 100–500 kW as a US anchor site. Not familiar with US market pricing or availability. | Carrier-rich US market. Provider with international company experience. Normalized pricing in USD with clear NRC/MRC breakdown. |








