I'm not "saying" pace of improvement, I'm "seeing" pace of improvement.
It has been getting harder to solve edge cases for years already, yet, the pace of solving edge cases has ramped up thanks to the drastic leap from V11 to V12.
Right now, based on data, the disengagement rate of FSD is approximately two orders of magnitude worse compared to Waymo's. If you have better data, I'm happy to re-consider my position.
Any OEM that doesn't make at least one model capable of being licensed for Tesla FSD will be behind any other OEM who does.
The OEM will get all the profit from the vehicle sale. This should keep the lights on for them.
I don't think that's very likely. Tesla just had to drop the price in half to improve its take rate. So as of today, FSD is barely a factor even for the majority of Tesla buyers. I don't think OEMs are under any time pressure to make a decision today or even over the next couple of years. It might very well be that Waymo manages to reduce the price of their sensor stack and then it makes more sense to license their technology.
This makes the best kind of sense for those OEMs as it won't require them to invest billions trying to reinvent the FSD wheel, yet, allow them to keep selling cars with their marque on them. (along with the "Tesla Inside" sticker, of course)
But, you didn't actually answer the question regarding what timeline is your thesis based upon?
Again, if I were an OEM, I'd hedge my bets and not commit to a solution before autonomy chooses its winner. There's no point in rushing. European manufacturers have plenty of regulating power in Europe so they'll be able to delay it long enough to get a license to put it on their cars once it happens.
Regarding the timeline. I think Waymo is better positioned right now to produce commercial autonomous taxis at scale. But no matter the winner, I don't think autonomy will produce serious profits sooner than 7-10 years.