You are amazingly naive. People are still hacking COBOL, Fortran and PL/I code that was written 30 or 40 years ago. There are obvious and huge advantages to replacing these legacy systems, but nobody ever does, because why fix what isn't "broken"?
No matter how advanced your technology is, systems can never be 100% secure because of the monkey in front of the keyboard. Think idiots who use "123" as a password, put their password on a sticky note, respond to emails from Nigeria, etc. Or simple black hats abusing their access. Even if systems could be made 100% secure, nobody in the software industry has an economic incentive to do so. A lot of vendors charge for bug fixes and support. Microsoft makes money off antivirus software.
Your current fixation is on Lisp Machines. But really, hardware versus software is irrelevant. Either hardware or software could be constructed to be provably correct. Nobody ever does, except for things like elevator control hardware where it really is necessary.
General-purpose hardware always tends to win in the end, because the hardware business is a volume game. The more you ship, the cheaper it is. Your custom, one-of-a-kind Erlang machine might be beautifully designed, but Intel has armies of physicists working on shrinking transistors, and massive economies of scale. Arguably, microprocessors are one of the most advanced technologies the world has ever seen, but due to economies of scale, you can buy one for cheaper than the price of a fancy haircut. Think about that. You're not going to win this battle writing transistor diagrams on a napkin in your basement. Why don't you read about the original Lisp Machines. Or Intel's CPU that executed Ada directly.
If you have a lot of data, what you need to do is build a Hadoop cluster, not pay through the nose for some proprietary hardware. You can even do it without writing a line of unmanaged code.
In 2015, there will be more application developers using managed languages. There will also be more people writing drivers in C, and creating hardware using Verilog or VHDL. And there will be another crop of idealistic college students ready to believe that the world will convert to Scheme or Prolog in 3 years.