Years of advancements in the Semantic Web are determining a technological gap between the Linked Data levels of the traditional Semantic Web vision, and its higher layers. While the core knowledge representation and interlinking mechanisms have consolidated rather rapidly, standardisation efforts for reasoning, unifying logics, proofing and interaction are striving to reach maturity. This has given rise to alternative schools of thought concerning the nature of the Semantic Web infrastructure, some of which are even putting the very need for ontology languages in question. Part of this phenomenon is due to unexpected results in interpreting combined Linked Data along with their schemas, alignments and other ontologies, with subsequent declining trust in high-level semantics from application developers. This talk will explore some possible research directions that can help keep ontology management on track with the evolution of Linked Data. One such effort will be described in greater detail, which proposes virtualisation as a technique for dynamically assembling multiple semantic data sources into makeshift ontology networks. Experiments on the interpretation of virtual ontology networks have shown promising results in several recurring distribution scenarios of Linked Data statements, with the highest possible axiom expressivity being reached with a reduced assembly effort.