So to put finer point on when peak ICE will happen, consider that the global new car market grows about 2% to 3% per year. When the growth in EVs becomes large enough to consume all the growth in the market, there is no growth for ICE. That is,
Growth_ICE = (Growth_All - Share_EV * Growth_EV) / (1 - Share_EV)
So clearly Growth_ICE < 0 when
Share_EV > Growth_All / Growth_EV
So let's assume that EVs continue to grow about 50% around the peak and that the whole auto market grows around 2.5% around this peak as well. Then the peak happen when EV marketshare hits 5%.
Now one issue that makes this tricky is that the growth in the entire market is quite variable from year to year. So in a bad year, it could look like ICE is post-peak before the difinitive peak. And in a particularly good year post-peak, ICE could have a little growth. So we may have a few bumpy years near the peak, a bumpy plateau, if you will. However, one thing that works against that is that demand growth for EVs may still out pace the growth in supply. Suppose EV marketshare is at 4%, but 7% of the market wants to buy an EV the next year while the supply will only be enough for 6%. What will that surplus demand of 1% do? Will these excess EV buyers settle for a new ICE, or simply wait until the next year? We know many people who are waiting until the Model X or the Model 3 to buy their next new car. They'll make due with older cars or even buy a used car to bide their time. So it may well be that close to peak many buyers will delay purchases in anticipation of buying an EV. If this is the case, it puts downward pressure on the growth of the new car market leading up to the peak. Moreover, just one year after EV share hits 5%, the share goes to 7.5% and then over 11% the year after that. So within two years of hitting 5% share, EVs are decisively gobbling up any growth in the entire new car market.
Thus, I am pretty confident that the peak will happen within a year of EVs reaching 5% marketshare. So the next question then is, when will this critical share be reached? This largely depends on how competitive the EV market becomes. At the extreme, we could envision a scenario where Tesla must do battle alone. Thus, Tesla alone must build 5M cars or so. After hitting 500k in 2020, it must continue to grow at 50% for 5.7 years. Thus, 2026 is my most remote sceario. More likely, Tesla will have only half the EV market at peak ICE, whence Tesla needs only reach 2.5M cars as other EV makers complete the rest. This moderate scenario puts the peak at 2024. Finally for an aggressive scenario, suppose Tesla has only 25% EV share at peak ICE, then Tesla needs to get to 1.25M, which happens in 2022. So here's the irony if the auto industry comes out with a bunch of Tesla Killers so as to confine Tesla to a 25% EV marketshare, then they hasten the ICE apocalypse to as early as 2022. Otherwise they can forestall their fate by at most 4 years, but enter the post-apocalypse utterly unprepared. I don't think it's realistic that the industry will let Tesla walk away with more than 50% of the EV market, so the ICE peak will most likely happen 2022 - 2024. Even so the important questions are: when will EV reach 5% penetration? And who will have how much share of the emerging EM market at that time?