The cloud argument of the 2010s was elasticity. The cloud argument of the mid-2020s is the invoice. Steady-state workloads — the ones that run every hour of the quarter — do not need a burst story. They need a unit cost. When that unit cost is higher than a colo row plus a small platform team, finance starts using the word “repatriation,” which is a polite way to say “we are moving it.”
This is not a morality play about hyperscalers. It is a matching problem. Some products are still perfect for the public cloud: spiky demand, global footprint, managed data stores you do not want to staff. Others were lifted and shifted in a hurry and never earned their keep.
The bill is a map
If you cannot say which service, which team, and which product own the top twenty lines, you are not ready to move anything. You are ready to argue. FinOps that only produces dashboards is theater. FinOps that can attach a cost to a SKU can choose.
Egress and data gravity do more damage than compute. The workload looks cheap until you try to leave, at which point the architecture reveals it was a hotel, not a house.
Repatriation without an ownership model is just a more expensive outage.
What actually comes home
Batch analytics, predictable inference, and stateful systems with clumsy data gravity are the usual candidates. Customer-facing apps with a global edge usually stay. The hybrid is the real destination: reserved cloud for the bursty surface, owned capacity for the furnace.
People are the constraint. A company that outsourced operations cannot “bring it home” without hiring, and hiring is slower than a board slide. The successful moves staff the platform team first, migrate a boring workload second, and only then hold a press quote.
How to decide
Run a twelve-month fully loaded comparison: cloud list minus real discounts, versus colo, hardware, staff, and the cost of being wrong. Include the feature you will not get — managed failover, a global backbone — as a product risk, not a rounding error.
If the savings only appear if nobody leaves the company, the savings are fictional. If they survive two resignations and a bad disk year, you have a strategy.