There are various numbers out there but according to the NFPA (National Fire Protection Association), around 117,000 cars catch fire each year in the USA. This works out to roughly 4 in 10,000 cars each year. This is for all cars on the road not just new ones. If your bone stock 1955 Chevy catches fire for any reason it is included in this number. I do not know if that includes arson. What are the rates of fire for electric vehicles? I don't know. I do know that if one burns up 10 states away I will see it reported. Yet the other 116,999 fires seem to happen without fanfare. That sounds like an agenda to me. Picking just one EV maker, Tesla, that has produced about 7.25 million vehicles as of 2024, we would expect that they would have had about 2900 fires if this same .04 percent number holds up. These would be world wide numbers. Have they? I strongly suspect we do not have good data on that but the only number I could find was a reported 255 Teslas have caught fire according to carjunkya.com. I would be willing to bet the number is substantially higher. But is it 10 times higher? That is what it would have to be to catch up to the average.
A couple notes:
There are various timing problems with the data. In other words not all the data came out on the same day, or in some cases even the same year. With Tesla's almost exponential sales increases this is a major problem.
The 117,000 number includes EV's of all kinds. But since the install base is so small I think that can be ignored.
There are no 70 year old (1955) Tesla's out there. Does the relative newness of the Tesla fleet increase or decrease it's rate of fire? I would think if it is a manufacturing or assembly problem, IE some have it and some don't, then it would increase it's rate as those vehicles quickly eliminate themselves from the pool. If it is an inherent problem with the design then as the install base get's larger/older the real rate will become known over time.
The actual data for this is horribly incomplete. Leaving the possibility that the numbers are very different than what is available. With such a small install base this could really change the statistics/percentages. Tesla probably knows the real numbers but even if it is relatively good they wouldn't want to release it because the vast majority of people will only hear two words, Tesla and fire.