This is definitely possible as 2 cubes in the same JVM could communicate through the local machine network. This can be optimized further by using the LOCAL_MESSENGER, which will do less Serialization, but more clones.
<distributedActivePivot>
<Id>DistributedCubeName</Id>
<clusterId>ClusterId</clusterId>
<distributionType>POLYMORPHIC</distributionType>
<underlyingPivotId>CubeName</underlyingPivotId>
<properties>
<!-- <entry key="distributedMessengerKey" value="JGROUPS_MESSENGER" /> -->
<entry key="distributedMessengerKey" value="LOCAL_MESSENGER" />
</properties>
</distributedActivePivot>