The first issue is that the RDBMS code base in ancient. The second issue is that we keep pushing the envelope on the RDBMS model; examples include OLAP and RDF triple-stores.
Some folks want to say "reports of the death of the RDBMS are premature."
Like COBOL, the relational model, and words like "DASD", some technologies will be with us for decades after their useful life.
The decline of COBOL and the Relational Database will be protracted, painful, inevitable and asymptotic with actual death. The old one-size-fits-all COBOL is being replaced by many other languages. Similarly, the one-size-fits-all RDBMS will be fragmented into more specialized data stores. Further, legacy technology never completely goes away.