Most infrastructure debates start with a provider: which cloud, or how many. We start one step earlier. Each workload has its own requirements, and those requirements decide where it should run — which is why most enterprise estates end up mixing models rather than picking one.
Five factors that decide placement
Latency: how close the workload needs to be to its users, its data and the systems it talks to. Data residency: where the data is allowed to live, and who can reach it — including UU PDP obligations where they apply. Compliance: what evidence auditors will ask for. Cost: not just the monthly bill, but the cost of running and changing the platform. Team capabilities: what your engineers can operate confidently at 3 a.m.
The five models, and when each fits
Data center — modernize and consolidate existing infrastructure when the hardware, contracts and skills are already in place. Private cloud — dedicated capacity, access control and data residency for regulated or sensitive workloads. Hybrid — keep stable systems where they are and extend selected workloads outward. Multi-cloud — use specific provider capabilities, but only where the added complexity is justified. Public cloud — elastic capacity and managed services for workloads that benefit from them.
Multi-cloud is a decision, not a default
A second provider adds tooling, controls and skills to maintain. If you are evaluating multi-cloud, the question isn’t “which two providers” — it is “what specific capability or risk are we solving for that one provider doesn’t cover.” If there isn’t a clear answer, a well-run single platform plus a tested recovery plan is usually the better outcome.
Whatever the mix, the layers above have to connect: compute and network, platform, integration, data and AI, with security and operations running across all of them.