Of course, it's just about where it comes from. There's not a GCC premium edition, or a hidden LLVM test suite, etc. I would never question the value of money in development. I think that the business model for languages and runtimes has (for a while now) clearly shifted towards no-strings-attached usage because the sponsoring/stewarding corporate interests align with the open ones. It seems the only barriers towards language and runtime use are put up these days by those clinging to business models of the past.
Thanks to Oracle, there is no premium OpenJDK or a hidden test suite as in the Sun days (Oracle contributed the remaining commercial "premium" features to OpenJDK a few weeks ago, including those that came in through the BEA acquisition). There is premium support offered by several companies (Oracle, RedHat, Azul, IBM), and a commercial JCK, which is not the OpenJDK test suite but a Java specification test suite for implementations that are not OpenJDk. The JCK can be freely used by OpenJDK.
Premium is something like discontinuing JS engine people rely on, move to Graal where certain performance features are not free for everyone. Hidden test suite is like the JCK I can't download to test my non-OpenJDK-based open source Java impl against (or Harmony or whatever). Regardless of details, the point is that it isn't more open like a MIT licensed Rust, Go, (recent) .Net, etc. There are obviously differing levels of openness by language and runtime regardless of specifics and differing levels by stewards of these languages/runtimes to make money on them.
We're not discontinuing Nashorn, simply not investing more in its maintenance any with the resources we have, which we can put to better use; if someone else wants to step up -- they're more than welcome to. Rust, Go and .Net Core (I think .Net isn't open source) are much, much smaller than OpenJDK. They require far smaller investment from the companies/organizations behind them.
Check how many of their key developers are on pay slips from multinationals, including Oracle.
Bills need to be paid and advanced compiler optimization research isn't doable as hobby.