Clojure (the first talk) - Rich Hickey at LispNYC (2007) 

lolol. wish i would've given Clojure some time in 2007 or so.

What info, please?

Interesting observation. But I don't agree with your conclusion. Virtual machines provide you a big toolchain not available on the target OS. Well, OS could be also considered a VM, as well as C++ standard (it describes C++ as a VM). So I don't think that it matters whether applications written for JVM actually compile to bytecode and not native (i.e. via Graal). I think what Rich Hickey meant is that C ABI and its ecosystem is not a good enough environment to build a language for, but JVM and other VMs are. And, you can look at Rust and its ecosystem as a counterexample of this :)

For another case of success of VMs, look at JavaScript, environment around it and languages built on top of it

Maybe we can remove the OS too. I’m thinking of the Unikernel approach: embedding only the functionality of the OS you need (eg the TCP/IP stack) directly in your app (aka micro/service), leaving out what you don’t need (eg a shell, device drivers, process schedules, etc). You get an executable that is launched by the Hypervisor, which is what cloud apps run on usually. mirageOS is a notable example of this approach. It’s been developed by Xen and later bought by Docket. Maybe worth considering what role the JVM/GraalVM and the standard library could play to achieve this goal.

Graal offers the best of both

Reality is somewhat more complicated. The JVM's recent advancements (e.g., green threads, value types, AOT compilation) are actually a response to Go and Rust whether people like to admit it or not. These changes are necessary because people are using the JVM for things it wasn't made for i.e., it's not the general purpose VM we've come to believe it to be. In other words, the issues are specific to the JVM, not the general concept of VMs or runtimes. Modern runtimes are really performant and the containerized world you've hinted at is actually going full-throttle at near-native-performing WebAssembly runtimes. WASI is the new C ABI, and is ushering in a new age of interoperability between basically every major ecosystem in existence. Once WASM runtimes mature (get their own GC and so on), it's going to be pretty natural for languages new and old alike going forward to race towards first-class support for it.