Exponential growth extrapolations always look great, but get beaten down by the sheer scale of new additions required to keep the growth rate up as the base gets bigger. We have seen this with wind and solar, where the "hot" nation goes through rapid growth from a low base and then the growth rate falls rapidly. Journalists have tended to jump to another "hot" market (it was Germany, now its China, now its not!). Government policies also change as the cost of subsidies grows exponentially to the point where they get cut (as in China last year). The major solar and wind manufacturing associations see a continued deceleration in solar and steady under 10% growth in wind.Then there are all the required grid changes as solar and wind go to higher shares of generation.
Same with EVs, which are still at a very low market share (2% in the US, 4% in Europe, 6% in China) and sales fell in 2019 in the US and China. The non-extension of Federal purchase subsidies will be a drag in the US in 2020. China plans for 20% EVs in 2025, which will be a significant accomplishment. Truly self driving cars are suffering from the usual real-world messiness and software development complexities, the hype around them has rapidly dropped reflecting reality.
The other variable is oil prices. If EVs take off and cut global oil demand, the price of oil will crash and cause a speed-bump to the adoption of EVs, unless countervailing taxes are put in place. I still remember C$0.30 per litre gas in the 1990s when oil hit US$10/barrel, anything like that would give a short-term lease of life to the ICE.
So I see more toward 2030 as the cross-over point, although the other possibility is that people just keep driving their current cars, which will have mostly been depreciated already, while they wait to see what is going to happen with ICE vs. EV. I personally will keep driving my low-mileage 2014 BMW until at least then. I did tell my friend's daughter not to train as a car mechanic, given the probable collapse of that occupation in the 2030s. Another thought is that rapid technological advancements will actually destroy the resale value of old EVs for a while, until things stabilize at the top of the technological "S" curve. The difference between lets say a new Model S and a five-year old one may be so great (leaving out possible production quality issues) that the depreciation becomes quite rapid. Such volatility and inability to forecast could only help people decide to keep their current vehicles until the market becomes clearer and more stable - a nightmare for the car manufacturers.