If you define several clouds with the same labels you expect not only one cloud to be used but to share the load between all of them. However, Jenkins always tries to provision a node on the first cloud with matching labels. It will go to other clouds only if provisioning fails - It's a common Jenkins behaviour.

Several clouds with similar labels will make only the first cloud to be used in a non failure scenario.

Ideally, you should share the load and even be able to have more "expensive" clouds later in the list and the "cheaper" clouds will be used first until they reach capacity