As few of you mentioned this method works for many years already, I did found it like 5 years ago, and did reverse engineering on it. And did some tests few years ago.
Basically as some of you mentioned it's number game. And from my opinion it's thousand od domains daily that is added in system. Huge base of keywords and scraped content, all random without any logic behind kw targeting or something.
And yes they use fresh domains, expired domains and hacked websites in this method.
This all together helps pages to get indexed fast. That's why they getting this amount indexed pages per domain, because every page on every domain doing internal and external linking to each other all in random, every refresh of page will change links to other URL's.
Probably there are more then 1 team doing this, but it cost to start, even if you going just with fresh domains, but still you need some domains that are already in index (that's why they are using hacked domains) when they start with blast, they start earning money instantly via smartlink redirects, it's like media buying, your costs are not ads, but domains.
Some of domains will not rank, some of them will rank medium and some will rank very high. It's all meter of match of some unknown long tail keywords on SERP.
Indexing on this way works like a charm, I did test with 20 fresh domains and it works. But 20 domains to work on this method is not enough. I would recommend at last 1000 for start, it's not small invesment if you are not sure that you have all in place.
Do some fresh google search with some random long tail, choose 24h results or 1h results and you will find this websites.
Summary from my point of view:
1. Get at last 1000 domains
2. Get millions of keywords
3. Scrape SERP for months and mix all this content on radnom
4. Add from 10k to 100k pages per website
5. Interlink all on random and let it be dynamic so to be able to index as much as possible in short time
6. Make a script to submit on radnom all the time anonymously to GSC
7. Monetization is not a problem.