How do you arrive at several decades and billions of organisms? You can't extrapolate the current mutation rate to the spillover event; SARS-CoV-2 was already quite fit at infecting humans and didn't need to change much. In contrast, during spillover events there's heavy selection pressure to adapt to the new host.

There's also the possibility of recombination events, which can instantly cause significant drift. Now consider the case of the Mojiang miners, who contracted and slowly succumbed to viral pneumonia in 2012 after inhaling bat guano. WIV found, like, a dozen beta-coronaviruses in that cave alone (including RaTG13.) The miners could have been exposed to several at once, which recombined and adapted quickly to their new home in human lungs. Again, unlikely to have been SARS-CoV-2, since the pandemic almost certainly would have happened already, but whatever they had was virulent enough to kill half of them.