Well, no one can ensure that same text they found will be found again on a link. It can simply change since the last time you saw it.
Now, building an index means that after you are done you can drop the data you had and use that space for getting newer data, so you might not even have it. It's probably smart to keep the data to help the crawler be less aggressive at downloading the whole internet though, but this data is globally replicated and ready to serve, might be just in 1.5+ hard disk drives that are not publicly reachable because they are dedicated to run the indexing service, not serve end users.
Could Google do it? Yeah, sure, but they are busy doing LLM things.