● Total Cost of Ownership

The Real Cost of
Public Cloud at Scale

Public cloud pricing is designed to be simple to start and expensive to scale. Private cloud flips that model — predictable costs that decrease per unit as you grow.

See Hidden Costs Request a TCO Analysis
The Economics

Why Costs Diverge at Scale

Public cloud follows a consumption model — you pay per VM-hour, per GB stored, per GB transferred, per API call. At small scale, this is efficient. At production scale, these micro-charges compound into significant and often unpredictable monthly bills.

Private cloud follows an asset model — you invest in infrastructure once, and the marginal cost of each additional workload approaches zero. There are no per-VM fees, no transfer charges, and no metering on internal traffic.

Industry analyses consistently show that the break-even point occurs in the range of 100-500 VMs, depending on workload characteristics. Beyond that threshold, private cloud costs significantly less per workload-month than equivalent public cloud configurations.

Public Cloud Cost Drivers

  • Per-VM or per-vCPU hourly billing
  • Data egress fees on every byte leaving the region
  • IOPS charges on high-performance storage
  • Inter-AZ transfer fees for HA architectures
  • Support tier fees (percentage of total spend)
  • NAT gateway and load balancer hourly charges

Private Cloud Cost Drivers

  • Hardware (one-time CapEx, amortised over 3-5 years)
  • Co-location or data centre space
  • Operations team or managed service fee
  • Network connectivity (fixed monthly, not per-GB)
  • No per-VM, per-GB, or per-API charges
What the Bill Doesn't Show

The Hidden Costs of Public Cloud

These costs are architecturally embedded — they cannot be optimised away without leaving the platform.

↑

Egress Fees

Every byte leaving a public cloud region costs money. Backups to external storage, API responses to users, data replication — all metered. On private infrastructure, outbound traffic costs are a fraction of a cent per GB on your own network uplink.

▲

IOPS Charges

Base block storage comes with limited IOPS. Database workloads requiring sustained high IOPS need provisioned tiers at significant additional cost per volume per month — on top of the storage capacity charge.

☎

Support Tiers

Without a paid support plan, you get community forums and documentation. Enterprise-grade support with SLAs, phone access, and a Technical Account Manager costs a percentage of your total monthly bill.

⇄

Inter-AZ Transfer

High-availability architectures that spread across availability zones — which every cloud provider recommends — incur continuous cross-zone data transfer charges on every replication, failover, and health check.

↺

NAT Gateway Costs

Each NAT gateway incurs hourly charges plus per-GB processing fees. A VPC with multiple private subnets and moderate outbound traffic accumulates significant NAT charges monthly — a cost category with no equivalent in private infrastructure.

🔒

Lock-In Premium

Proprietary services (managed databases, serverless, AI APIs) carry implicit switching costs. The deeper you integrate, the more expensive migration becomes — by design. Open-source infrastructure carries zero exit fees.

TCO Framework

How to Compare Costs Properly

A fair comparison must include every cost category — not just compute hours.

Cost Category Private Cloud (OpenStack) Public Cloud (AWS/Azure/GCP)
Compute Hardware amortised over 3-5 years Per-hour or reserved instance pricing
Storage Ceph cluster, capacity-based Per-GB/month + IOPS tiers + snapshot charges
Network egress Fixed upstream cost Per-GB metered, increases with volume
Software licensing Apache 2.0 — zero Included in per-unit pricing (opaque)
Operations Internal team or managed service fee Included (but support tiers add cost)
Scaling cost curve Decreasing per-unit cost as you grow Linear or increasing (volume discounts plateau)
Exit cost Zero — you own everything Egress fees + re-architecture + retraining
Honest Assessment

When Public Cloud Still Wins

Private cloud is not always cheaper. These scenarios favour public cloud economics:

  • Burst workloads — short-lived spikes where you'd otherwise overprovision hardware
  • Small scale — under 50 VMs, the operational overhead of private cloud outweighs savings
  • Global distribution — if you need presence in 20+ regions, building private DCs everywhere is impractical
  • Managed services — if your team lacks the skills to operate databases, queues, or AI infrastructure
  • Speed to market — when a prototype needs to be live in hours, not weeks

The Honest Question

The right question is not "which is cheaper" — it is "which cost model aligns with your operational reality."

If your workloads are predictable, long-running, and data-intensive, private cloud almost certainly costs less. If your workloads are spiky, globally distributed, and heavily reliant on managed services, public cloud may be the right choice.

We help you make that assessment objectively — without a financial incentive to push you either way.

Request a TCO Analysis

Share your current cloud spend and workload profile. We'll build a side-by-side TCO comparison specific to your environment — with real numbers, not marketing estimates.

Get in Touch