if you do your math well
the normal time for scanning the 66 bit range on an average cpu would take nothing less than 500 years
28 months running 1000 GPUs, nonstop.
Imagine that electricity bill at the household rate.

Ok then, lets materialize our imagination, if we want to be fair, we should say the time taken to solve 66 is 14 month and not 28 month. Right?
14 month = 420 days.
420 days = 10080 hours.
RTX 3090 rental price = $0.20/hr
1 RTX 3090 rented for 10080 hours = $2016
1000 RTX 3090 rented for 10080 hours = $2016000
We will use a discounted price of $2,000,000 USD.
Puzzle 66 contains 6.6 bitcoins, at a price of $30,000. 6.6 * 30,000 = $198000.
In the future when bitcoin is at $100,000 * 6.6 = $660,000.
Further in the future, bitcoin is at $300,000. tech has advanced, speed of key per second is now 10B/s. So we just need 100 GPUs.
100 * $2016 = $201,600. With discount is $200,000 cost of finding #66, and bitcoin at 300k, now puzzle reward is worth $1,980,000 - 200,000 = $1,780,000 profit.
Question, when will we see GPU speed jumping from 1B/s to 10B/s, and bitcoin price at $300,000?
Even then what about #67, two times harder than #66 in theory, #68, #69, #70?
It's either profitable to rent GPUs and find them, or will be profitable in the future, in both cases, nothing really changes, nobody is gonna come up with a solution and an algorithm to brute force keys with 10B/s rate in the next 10 years.