The community has known about server-side collection for quite sometime now. You could run Google Analytics on any of the serverless environments since a year or two ago (I noted this on news.yc a year back [0][1]). Tag Manager server-side is Google throwing its own solution in to the mix.
DNS based content blocking was always DoA, there simply are too many chinks in the armour besides CNAME or HTTPS/SVCB or SRV or ALIAS record cloaking [2]. The worst I've seen reported to me by users is a tracker generating domains names on-the-fly (domain generation algorithms) and A/AAAA records pointing to different IP addresses each time [3].
That said, a firewall can still mitigate this offensive, while network security with just DNS was always going to be what it was: A stop-gap.
This isn't the end-game: I fully expect that IP address blocklists would crop up in no time, and will be painfully maintained by folks pouring their life in to it.
TFA points that Google's reverse-cloaking presumably with IP addresses, but the worse would be if multiple domains shared IP addresses (like in a CDN), reverse-cloaked with Server Name Identification. Even firewalls would have to blanket block IPs... and what if those IPs are shared with other Google front-ends like the AMP project / YouTube / Mail / Docs?
The firewalls would also have trouble with something like Ao1 [4]: If multiple websites were behind multiple IPs, or in the extreme, a single IP.
The firewall is bust, but that's good, now we simply de-Google / de-Cloudflare ourselves, and be luddites like they want us to be.
[0] https://news.ycombinator.com/item?id=26003654
[1] https://news.ycombinator.com/item?id=25169029
[2] https://news.ycombinator.com/item?id=26298339
[3] Ex: https://www.reddit.com/r/uBlockOrigin/comments/srza8x/changi...
[4] https://nitter.net/rethinkdns/status/1448738898998292495