The board mandates multi-cloud to avoid lock-in. What do you present back?

What the interviewer is testing

Whether you can engage with a strategic concern honestly rather than either complying or dismissing it.

Establish the actual objective

"Multi-cloud" covers four very different things, and they have wildly different costs:

Portability — able to move if needed, over months. Redundancy — active workloads on two providers for availability. Best of breed — different providers for different workloads. Negotiating leverage — a credible threat to move.

The board almost always means portability or leverage. Engineering hears redundancy, which is the most expensive interpretation by a wide margin.

The cost of active-active multi-cloud

Every design falls to the lowest common denominator — no managed services with no equivalent elsewhere, which is most of what makes cloud productive. Two sets of operational expertise, tooling, security controls and compliance evidence. Cross-provider data transfer, charged continuously. And the honest reliability point: the added complexity introduces more failure modes than it removes, and correlated failures in your own abstraction layer are the likely outcome.

What I would propose instead

Portable architecture as the default: containers, open table formats, standard protocols, infrastructure as code, and a deliberate register of provider-specific services with a named alternative and assessed migration effort for each. Cheap, and it covers a real portion of the risk.

Data portability demonstrated — an actual extract and load elsewhere, with time and cost measured. Data gravity is the real lock-in, not APIs.

A tested exit plan for critical services, since regulators increasingly require demonstrated capability rather than documented intent.

Best of breed where it is genuinely better, accepting the integration cost consciously.

Concentration risk accepted explicitly at board level, with the reasoning recorded.

What a strong answer adds

Quantifying the ask: the estimated cost premium of full active-active multi-cloud versus the portable approach, and what each buys. That converts a mandate into a decision the board can actually make, which is what they were trying to do.

Common weak answers

Complying and building an abstraction layer over both providers. Dismissing lock-in as a non-issue.