“We’re moving everything to cloud.” Said as if it’s self-evidently the right answer for everything. It isn’t.
Cloud is genuinely the right call for a lot: anything variable, bursty, growing, or spread across regions. Renting capacity you can scale in minutes beats owning a datacentre you sized by guesswork. That value is real and not in question.
But “cloud-first” quietly became “cloud-only,” and that’s where it stops being strategy and starts being dogma. Some workloads are steady, predictable and heavy, running flat out, all year, no spikes. Rent that long enough and you’re paying a premium for elasticity you never use. A few well-known companies have done the maths and moved exactly those workloads back. Not because cloud failed — because that workload was never a good fit for renting.
Repatriation isn’t a cloud backlash. It’s the market growing up, realising the question was never “cloud or not,” it’s “which workloads, where, and why.”
The honest version of cloud-first: cloud where it earns its place, owned where it doesn’t. Decided workload by workload, not by slogan.
If your cloud strategy is “all of it,” that’s not a strategy. It’s a default with a press release.