Multi-Cloud
Deliberately running across more than one cloud provider — a decision with a much higher cost than the lock-in it is usually adopted to avoid.
Worth separating the versions, because they are not equally costly. Best-of-breed — using one provider's speciality service alongside another's core platform — is common and usually fine. Portfolio — different workloads on different providers, often through acquisition — is a management overhead, not an architecture. Portable — the same workload able to run on either — is the expensive one, and the one people mean when they say multi-cloud.
Portability costs: you can only use the lowest common denominator of managed services, so you build and operate what you could have bought; teams need expertise in two ecosystems; tooling, identity, networking and observability are duplicated; and inter-cloud data transfer is billed both ways, which frequently exceeds any discount the negotiating leverage produced.
The honest questions: is the lock-in cost real (price it as exit cost — how long to move, what to rewrite), and is there a specific requirement driving it, such as a regulator or a customer contract demanding provider independence? Absent one, a single provider with deliberate portability at the boundaries — containers, open formats, avoiding proprietary DSLs — captures most of the benefit for a fraction of the price.