I view this as a serious failing of the internet that we collectively should have done a better job of avoiding. In most cases I believe the content itself is in fact still available somewhere and it’s simply the link that broke. Some kind of two layer system like the DOI system used for libraries would be helpful for cases like that:

https://nuim.libguides.com/referencing/DigitalObjectIdentifi...

But it would need to handled automatically to maintain the full usefulness and convenience of URLs. Not sure how that could work though.