Rented Square Footage: When a Cloud Migration Changes the Address but Not the Architecture
Photo: Ana Las Heras, CC BY-SA 4.0, via Wikimedia Commons
There is a moment familiar to many CIOs: the migration is complete, the legacy data center contract has been terminated, and the cloud migration is declared a success in the quarterly business review. Then the first full-cost invoice arrives, and it is higher than the one it replaced. The servers are the same. The applications are the same. The 2 a.m. incident calls are the same. Only the geography has changed.
This is not a cloud strategy. It is a colocation agreement with better marketing — and the industry's enthusiasm for 'cloud-first' mandates has produced a remarkable amount of it.
Five Signals Your Migration Was a Relocation
1. Your cloud bill grows with your business, not your usage. Cloud economics reward elasticity. If costs scale linearly with headcount and traffic with no unit-economics improvement, the workloads are likely running in a fixed, over-provisioned posture — rented hardware behaving exactly like owned hardware, minus the capex advantages.
2. Nobody has touched the architecture since migration. Lift-and-shift has a legitimate role as a first step, but it is a step, not a destination. If the application topology, database design, and deployment model are unchanged eighteen months later, the organization has paid a migration premium for an infrastructure swap.
3. Change lead times are identical to the old world. One of the core promises of cloud is velocity. If releasing a new environment still takes six weeks and a change advisory board, the organization has moved the data center without moving the operating model. The constraint was never the hardware.
4. Reserved-instance commitments mirror old hardware refresh cycles. When a company locks into multi-year reserved capacity matching its previous procurement cadence, it has recreated capital planning in a new vocabulary — often with worse economics than a well-run on-premises estate.
5. FinOps is a dashboard, not a discipline. Genuine cloud operations involve engineering teams accountable for the cost of what they deploy. If cost visibility lives in a monthly report nobody acts on, spending drift becomes structural.
A Decision Framework, Not a Mandate
The corrective is to retire the phrase 'cloud-first' in favor of something more honest: workload-appropriate placement. For each system, ask three questions. Does the workload have variable demand that elasticity can monetize? Does the application architecture — or the roadmap — support modernization toward managed services? And does data gravity, residency, or latency genuinely constrain placement?
Workloads that score high across all three belong in the cloud, and usually benefit from rearchitecting rather than relocation. Workloads that score low — stable, monolithic, latency-sensitive, or bound to regulated data with limited flexibility — may be better served by a modernized on-premises or dedicated hosting model. The hybrid middle is not a compromise; for many US enterprises in manufacturing, healthcare, and financial services, it is the rational steady state.
The uncomfortable conclusion is that some of what was migrated should, in hindsight, have stayed put — and some of what stayed should have moved. Neither admission is a strategic failure. Declaring victory on an address change, however, is. Technology leadership is measured not by where the servers sit but by whether the operating model, the architecture, and the economics actually improved. Anything less is just rent.
