Totally made up numbers (GOT will never sell 4 millions on PC, Horizon didn't in 5 years) and that's the best you can get. 65-70 millions doesn't even cover the marketing budget of a AAA.
It's best to at least try to have some known data when it comes to an argument of how much GoT might or might not sell.
Let us recall this official slide from an official presentation dating to Feb 2023.
Here we see that
Horizon Zero Dawn has sold 3.3m copies on Steam as of Feb 2023.
It is actually quite likely that it has passed 4M or the game is rather close to it with continued deep discount sales from March 2023 till May 2024.
It also has not been out for 5 years, as the PC port released in August 2020.
Closing in on a full 4 years in a couple months though.
Ghost of Tsushima is also having a larger launch than Horizon, with the CCU peak behind only behind GoW.
Doubly so because if we look at the CCU graph between Asia and EU, it is incredibly stable between the timezones.
That means that as Asia is going to sleep, instead of a sharper drop in CCUs, it is staying the course.
This means a greater amount of sales on a global scale than just looking at the singular CCU peak number.
TL;DR - Am I guaranteeing GoT will sell 4M+ and more? No.
But looking at launches of previous SIE games and the sales trajectory of their most successful titles, it's a very very strong possibility.
There's a reason a lot of people were saying that games like GoT/Bloodborne/Demon's Souls would probably overperform compared to the bigger IPs like Horizon/Spider-Man/GoW on PC relative to how much they sold on PlayStation.
These games are more aligned with PC taste even though, for some of them, the IP brand is weaker overall.