A hybrid estate cost model compares cloud, colocation, and on premises for the same workload on a fully loaded basis: compute, storage, egress, power, space, hardware refresh, plus the people and tooling each venue needs to run. The point is to decide placement per workload rather than by blanket policy. Steady, high utilization workloads that run continuously often favour owned or colocated hardware once capital is amortised, while spiky, growing, or uncertain workloads favour cloud elasticity. The model is the referee, and it only works if both sides count every cost, because a partial comparison always flatters the venue the analyst already preferred.
Here is what goes into the model, the costs each side tends to omit, and a worked comparison you can adapt.
Why compare on a fully loaded basis?
Cloud and owned infrastructure bill in completely different shapes. Cloud is a per hour operating cost with elasticity built in. Owned hardware is upfront capital plus the ongoing cost of running a facility and a team. Comparing a cloud invoice against a hardware purchase price is meaningless, because the invoice already includes power, cooling, staff, and refresh that the purchase price does not.
A fully loaded model normalises both into the same unit, usually cost per unit of useful capacity per month over a multi year horizon. Only then does the comparison tell you anything, and only then can placement follow economics rather than whichever number happened to be quoted.
What goes into the model?
Account for every line on each side. The common omissions are where comparisons quietly break.
- Cloud side. Compute and storage at the rate after commitments such as Savings Plans, Reservations, or Committed Use Discounts, plus egress and inter region transfer, which are the lines most often forgotten and the ones that move repatriation maths most.
- Owned and colocation side. Amortised hardware, the refresh cycle, power and cooling, rack space or facility cost, and the staff and tooling to operate it. Add the cost of headroom, because owned capacity must be sized for peak and sits idle the rest of the time.
- Both sides. Resilience and disaster recovery, network connectivity between venues, and the migration cost to get the workload where the model says it belongs.
When does each venue win?
The pattern of the workload decides. Steady, predictable, high utilization workloads that run around the clock are where owned or colocated hardware competes hardest, because you keep the asset busy and amortise it fully. Spiky, seasonal, or fast growing workloads favour cloud, because you pay only for what you use and never buy for a peak that may not recur. Uncertain workloads, where the future shape is unknown, favour cloud for the option value of not committing capital to a guess.
This is why repatriation is a per workload decision, not a strategy. The same estate can sensibly keep a stable database tier on colocation while leaving bursty front end and analytics on cloud, with connectivity costed between them.
A worked comparison
A Fortune 500 retailer assumed its entire estate was cheaper on cloud and was about to renew a large commitment to match. The hybrid model told a more precise story. A steady core of batch and database workloads, running at high utilization around the clock, came out lower on colocation once power, refresh, and operations were amortised, while the elastic front end and seasonal analytics stayed clearly cheaper on cloud thanks to elasticity. Placing each tier where the model pointed, and committing only to the cloud baseline that remained, cut the blended cost of the estate well into the range a typical optimization program delivers. Figures are verified against billing data and anonymised.
Frequently asked questions
What is a hybrid estate cost model?
When is repatriation actually cheaper than cloud?
What costs get missed in hybrid comparisons?
Build your placement model with us
We build the fully loaded model that decides where each workload belongs across cloud, colocation, and on premises, and we take zero provider commissions, so the answer follows your economics and not a vendor's. Our guarantee: we reduce your cloud spend or we reimburse our service fee, on either a Fixed Fee or a no risk Gainshare basis. More placement economics arrive through The Cloud Spend Navigator.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.