We can't assume good will and behavior from an industry that is built on deceiving and manipulating the user. The GDPR is a good first step at regulating these practices, but it's too vague, and it's applied far too leniently. It also obviously only applies to EU citizens, and not to the global industry.
I wasn't familiar with the privacy "budget", but it sounds like Google is trying to define privacy as a scale, where some amount of fingerprinting is OK. Users can be identified with just a few data points, and some are more valuable, depending on the context. Some might even be required for the site to function, so will there be "legitimate" exceptions to the budget in those cases? It sounds like a backwards approach that will be difficult to manage, so I'm not sure it will be a win for protecting privacy.
More importantly, I don't trust that an adtech company will go out of its way to implement solutions that go against its bottom line. These companies have a track record of abusing user data, and the only reason they take these initiatives is for good PR, which is again protecting their bottom line. The entire industry needs much broader and stronger regulation for any of this to actually improve.