At 100 sat/vByte, each block costs more than 1 BTC in fees.

With 10 min block time, that's 144 blocks a day.

So they would need to burn 144 Bitcoin a day.

In current exchange rate, $5.7m.

$171m a month.

I guess it sounds doable but wouldn't that just increase the bitcoin price in $ terms, while also increasing fees in both sats and $?

It seems unsustainable in the end...

If the btc price doubles (because the attacker keeps buying), and fees double (200 sats/vByte), that's quadruple the cost... $684m per month.

Unsustainable if you ask me.