The ad infrastructure can still exist -- it would just have a restricted set of data (IP, device fingerprint, the surrounding content, and whatever info the first-party publisher voluntarily submits about you) to decide what ad to serve. Small, niche websites may do better than big news sites under this regime since you can infer more about their visitors by the fact that they chose to visit.

I could see bigger sites expending a lot of energy trying to bring the tracking and inference in-house, and even federating these efforts, creating a kind of soft-paywall that requires you to "pay" by validating an email address or some other stable identity marker in exchange for temporary access to content, so they can watch what you browse and build a shared model of you that they can feed back into the ad networks. I could see the NYT continuing to manipulate and fine-tune its headlines and graphics, trying to sort its visitors into cohorts based on what appeals to them to squeeze every last cent out of a pageview.

At the same time, so much content discovery and consumption happens in the belly of the beast (Facebook, Google, Youtube) that most ads will continue to be targeted based on the considerable information those websites have about you, regardless of what browsers do or what happens to third-party tracker networks.