Colocation Pricing Explained: Why We Charge in Amps
How Fast2Host colocation pricing works — why power is billed per Amp, what your monthly cost covers, the benefits of capacity-based billing, and how rack space + A+B feeds come together in Cambridge.
Colocation quotes can look confusing: rack units, ports, IPs — and then amps. At Fast2Host we use per-Amp (capacity-based) power billing because power and cooling are what actually drive a serious colo cost, not just “how many U you rent”.
This post explains why we charge in amps, what you pay for, how it works, and why that model is better for real hardware — from a single 1U server to dense GPU racks in our Cambridge Tier III facility.
How colocation costs break down
When you colocate with us, the monthly charge covers a bundle. Roughly:
| Cost piece | What it is |
|---|---|
| Space | 1U–5U, quarter, half or full rack in a secure APC cabinet |
| Power capacity (amps) | Reserved draw on dual A+B feeds — UPS, generator and distribution sized for your bracket |
| Cooling | Precision cooling (Stulz / containment) to remove the heat your amps create |
| Network | Port (1Gbps / 10Gbps), bandwidth commit, IPv4/IPv6, BGP where applicable |
| Protection & ops | Always-on DDoS, monitoring, remote hands, physical security |
Headline packages on our rackspace hub (from about £55/mo for 1U up to full cabinet figures) already include a sensible baseline of space, power feeds, connectivity and DDoS. Final quotes scale mainly with how many amps you need and bandwidth — not surprise “electricity bill roulette” after a GPU training week.
Use the power calculator on the colo pages to model draw (amps → kW → indicative monthly energy + overhead).
Why we charge in amps (not “free power” or pure kWh)
1. Amps = reserved capacity we must hold for you
An amp rating on a PDU is not a marketing badge. If you need 16A (or 32A) continuous capability, we must lock that capacity across:
- Utility / intake paths
- N+1 UPS strings
- Standby generator
- Distribution and the PDUs in your rack
Billing per Amp ties your monthly price to the capacity reserved 24/7 so that when your servers hit peak load, the power is physically there — not shared hope on a soft “average kWh” product.
2. Heat follows amps — cooling is part of the price
Every amp drawn at the rack becomes heat the plant must remove. AI and dense compute (for example Dell PowerEdge XE9680 / NVIDIA HGX-class builds) can sit far above classic “web server” racks. Our facility plans cooling (precision CRAC and hot-aisle containment, up to multi‑tens of kW plant capacity) against that draw.
Per-Amp billing bundles electricity with cooling effort to keep a stable environment (around 20°C in the cold aisle) so hardware does not thermally throttle.
3. Predictable finance — no bursty bill shock
Metered-only kWh models look fair until training jobs, render farms or storage rebuilds spike overnight. A capacity bracket (amps) gives finance a fixed monthly power figure for that bracket. When you grow, we move you to the next amp tier — transparent step, not a surprise utility invoice.
4. Transparent scaling with live metering
We use networked APC metered PDUs. You (and we) can see real draw. Approach your limit, and we plan the next Amp bracket with you. No opaque “facility factor” you cannot check.
That is why per-Amp capacity billing is the gold standard for AI and enterprise colo, not an accounting trick.
Amps in plain English
At UK single-phase (~230V):
Watts ≈ Amps × Volts
Example (illustrative):
| Amp capacity | Rough power (230V, cosφ ~1) |
|---|---|
| 10A | ~2.3 kW |
| 16A | ~3.7 kW |
| 32A | ~7.4 kW |
Three-phase and high-density full racks go higher (full cabinets can be sized toward 10–12kW+ with the right feeds — we quote those properly).
A+B means two independent feeds. You size so a single feed can carry the load if one path is in maintenance — that is Tier III thinking, not “plug both cords into the same strip”.
Typical Fast2Host baselines:
- 1U–5U / single server — dual A+B feeds with metered PDUs
- Quarter / half rack — dual 16A A+B PDUs as standard (higher on request)
- Full rack — dual 32A A+B as a common high-power starting point
You tell us nameplate / measured draw; we confirm the amp bracket and cooling fit before go-live.
How pricing works in practice
Step 1 — Size the space
How many U? One server → rack units. Growing stack → quarter or half. Private cabinet → full rack. See also our buyer’s guide.
Step 2 — Size the power (amps)
Sum expected idle + peak draw of servers, storage and switches. Add headroom (~20–30%) for growth and A+B failover. Prefer measured draw from a PDU at your current site over optimistic datasheet marketing.
Step 3 — Pick network commit
1Gbps or 10Gbps port, unmetered commit (e.g. 100Mbps–1Gbps depending on package), IPs, optional BGP / cross-connects.
Step 4 — Quote = package + amp/power profile + term
Indicative web prices:
| Footprint | From (indicative) | Power highlight |
|---|---|---|
| 1U | £55/mo | A+B power |
| 2U–5U | £115–£220/mo | A+B, rising bandwidth commits |
| ¼ rack (10U) | £250/mo | Dual 16A A+B |
| ½ rack (21U) | £425/mo | Dual 16A A+B, 10Gbps port |
| Full rack (42U) | £650/mo | Dual 32A A+B, private cabinet |
Final pricing depends on power draw (amps), bandwidth and contract length. Longer terms often fix power/bandwidth rates for budgeting.
Step 5 — Deploy
Ship or deliver kit → we rack, dual-power, cable, hand over IPs → you manage the OS. Remote hands cover power cycles and cable work when you cannot attend.
Benefits of the per-Amp model (for you)
- Guaranteed peak capacity — UPS and generator share is reserved, not oversold on averages.
- Cooling matched to load — dense GPUs stay in a designed thermal envelope.
- Budget certainty — finance knows the power line item each month.
- Fairness — a 1U web box does not subsidise a 12kW AI sled (and vice versa).
- Clear upgrade path — next Amp bracket when live PDU graphs show you are ready.
- Honest quotes — we ask for draw up front instead of a low rack fee with painful extras later.
What is included vs what scales
Usually in the colo package: rack space of the tier you buy, dual A+B feeds at the stated amp class, cooling share, IP allocation, base bandwidth, free DDoS, remote hands allowance, 24/7 NOC, physical security, power SLA.
Scales with your design: Amp bracket / kW density, larger ports, extra IPs, cross-connects, 3-phase or DC feeds, cages/adjacent racks for growth.
Common questions
“Can’t you just include unlimited power?”
Unlimited power at rack density is a myth that ends in overloaded PDUs and thermal trips. Capacity-based amps are how serious DCs stay honest.
“Do I pay for idle amps I don’t use?”
You pay for the capacity bracket you reserve — that is the point of reliability. Right-size with measured draw; don’t book 32A “just in case” if 16A with headroom is enough.
“Is kWh still in there somewhere?”
Yes — energy is real. Our model folds energy plus the plant overhead (cooling, UPS efficiency, generator readiness) into a predictable Amp-tied monthly figure rather than a wild metered swing.
“How do I estimate before ordering?”
Use the on-site power calculator, check server PSUs and PDU logs, then ask sales for a bracket. Bring nameplate watts or a week of PDU averages — we respond with a clear quote.
Bottom line
We charge colocation in amps because power capacity and cooling are the scarce, expensive resources that keep your hardware safe at peak. Space in U gets you into the cabinet; amps get you a guaranteed, cooled, UPS-backed path for the load you actually run — with predictable monthly cost and transparent PDU metering.
Ready to size a rack? Start at Colocation & rackspace, read the UK colo buyer’s guide, or request a quote with your U-count and amp target — our Cambridge team will map space, A+B feeds and bandwidth without the mystery extras.