The insurance business model is not about distributing costs from high risk customers more evenly with low risk customers, but in distributing costs more evenly between unlucky customers and lucky ones [1]. After all, most insurance companies will access the risk of their customers and assign premiums accordingly - your insurance premium is the expected value of payouts, plus overhead, plus profit.
Wildfires can destroy large swaths of land, and they are certain to occur every year, but not at every susceptible house. My mother has lived in a house in a high fire risk area of Northern California since the early 1980s, but her particular neighborhood hasn't burnt. Neither has the houses of my father-in-law, grandma-in-law, and two brothers, which are all in similarly high risk areas of Nor-Cal. Likewise, in March, nobody would have predicted that Paradise, CA would all but be destroyed, nor would they have predicted the damage in Santa Rosa the previous March. But we could still have concluded that those areas are at risk.
An insurance company can assign a value to your property (in fact, they probably already have), can estimate the probability of loss due to forrest fire, compute the expected value of payouts, add overhead, add profit, and sell you a policy. Damages are correlated on a local level, but not on a statewide level - that there was a fire in Paradise, CA doesn't increase the odds that Quincy or Fort Bragg will experience a fire too this year.
With fire damages in the tens billions per year but maybe only millions of insured to spread around the cost, people might not like the resulting premiums, but they can choose to lower their risks (and hence insurance premiums) or roll the dice.
[1]: And in managing the large pile of money you need to have on hand in case you end up with more unlucky customers than expected.