I think it doesn't make sense to build a cheaper EV and it's unlikely to happen in the next 10 years. Maybe second hand Model 3 and Y can address that market. Tesla will need a lot of batteries for S3XY, roadster, truck, Semi, bus, and energy storage.
There's plenty of reasons for cheaper EVs. Plenty of people can't afford a $35k car, even used (especially if they're just going to end up in the Tesla Network instead of being sold), and also don't have public transportation as a viable option (either lack it entirely for their needs, or it's poorly implemented, or ... etc), and even if the Tesla Network takes off, it's not necessarily going to be affordable enough for a daily commute for those who can't afford it.
That doesn't necessarily mean Tesla needs to be the one building BEVs in the <$30k range, but if nobody else gets around to it... someone needs to. Even if TCO (assuming 1 million miles is achievable goal, or even half that, it should) favored a $35k BEV over a $25k ICEV, a good portion of those who can only afford the $25k (or perhaps even only afford $15k) + years of gas car can't get the financing for the $35k even if the TCO would be lower.
I suppose an alternative, assuming TCO (and thus monthly payments over the long run) would work out, would be for a financing company to come along that is less concerned with profits and more about getting more people into BEVs. If you could operate such a thing as a minimally profitable (but not losing money in the long term, clearly - unless it's being subsidized by wealthy benefactors and/or governments) operation, even accounting for risks... Perhaps if we assume Tesla's safety features are going to reduce accident rates significantly to the point of the only major "loss" issue would be people failing their payments (hopefully outright theft would be minimal since the vehicles location is constantly tracked), then you could simply repo the car, and either add it to the Network to make money on it to cover it's costs or sell and finance it again used (at a possibly slightly reduced rate, or perhaps not if the original rate was bare minimum) to someone new. If the goal isn't to make money but to put more and more people into BEVs and save the planet, then offering very attractive financing for very long terms to reduce the monthly payments could go a long way to expanding the addressable market. Even if this was structured like a lease for a few years at a time, if we assume the value of the vehicle is 1 million miles and at least 10 years, that's a long time you can keep extending the lease and/or leasing it to someone else. Of course you may need to start replacing bits of the interior at some point, but a refurbishment every 5-6 years or so might still work out fine.
Such an operation (lowering the barrier to entry for those without great credit and/or lots of cash on hand) would certainly let Tesla sell 3's and Y's until the cows come home and further the mission beyond what they can otherwise (even with regular loan/financing programs), the only problem then would be keeping up with the demand ...
Of course also having a lower end car that is cheaper to make would help even more, reaching even more buyers. If the $35k Model 3 is making good margins at scale, imagine what price you could achieve if all you did was actually build it with the originally intended cheaper interior, sound system, roll the clock forward a couple of years (for lower costs on many things, hopefully), and maybe even shrink the overall dimensions (additional costs for new stamping forms and so on with lots of vehicle specific parts, but at a certain scale perhaps is worth the cost versus just a cost-reduced Model 3/Y)... (assumes large enough scale that the savings of less variants is outweighed by the cheaper parts of those variants)