No doubt any replacement would have to be enough better that it's worth the cost (opportunity and otherwise) to both produce and switch to it. Perhaps security

No doubt any replacement would have to be enough better that it's worth the cost (opportunity and otherwise) to both produce and switch to it. Perhaps security/privacy issues might drive a serious rehatching of commonly used software. It occurred to me that there's enough inertia at this point that you'll likely only see ground-up architecture and/or implementation when some sort of hardware substrate is produced that requires it. Until then, we are on the karmic wheel of cpus that drive os/language design that drive cpu design.