I will counter that by contending that the Semantic Web is here, just not fully so, and furthermore, that a more and more complete Semantic Web is inevitable.
My feeling is that you couldn't go wrong learning RDF, SPARQL, etc, AND, more specifically, focus on NLP. Why NLP? Because NLP is the bridge between the gazillions of petabytes of "stuff" that's out there that isn't part of the Semantic Web, and the structured data that does make up the Semantic Web. When you can use NLP (and related tech) to extract semantic meaning from free-form text and then make it part of the Semantic Web, that's pretty powerful stuff.
Take a look, for example, at Apache Stanbol[1] and the stuff they're doing with extracting structure from text.
Also, look at things like dbpedia[2] and the Linked Data[3] initiative. Seriously, seriously cool stuff is going on...
The answer of the user "mindcrime" is a direct answer to the Web3.0 question, but I thought that you should know the background. As the background knowledge behind the answer is sometimes more helpful than the answer itself.
Building Services is trending right now and building technologies and architectures that are able to scale to the requirements of these services are part of the trend.
Key characteristics are: Low barriers to entry, little or no capital expenditure, massive scalability, multitenancy, device independence, location independence. But it can be reduced to "benefit" only aswell.
Everything as a service (EaaS, XaaS) is a concept of offering Services through technology. It is a subset of cloud computing, but not limited to. IaaS, PaaS and SaaS are the most popular form of Service.
The most common and successful example is software as a service (SaaS), but the term as a service has been associated and used with many core components of cloud computing including communication, infrastructure, data and platforms. Humans as a Service (HuaaS) and other creative ways of using technology and architecture to solve problems are exciting forms of these Services.
Tl;Dr:
Learn what you think will make the world (a little) better as a computer scientist. Semantic Web, Web3.0, Online Education and Home Schooling are the horses I would bet on. If you don't trust us, hack Google Trends and "similar services" http://www.google.com/trends/
Mobile java+droid / IOS OR
Rapid Web Django+Python or ROR+Ruby
Pick a vertical, then a technology and do a project.