The lack of any chipset that would have tolerable information for producing those or any lifetime or willingness to deal with someone smaller than a Sony or LG. But yeah, drivers is close to where that ends up breaking.

The thing I don't really understand is why no one is doing this. Fine, Qualcomm sucks, but why is there no phone SoC from SiFive?

The TDP of a lot of existing phone chips (e.g. various Snapdragons) is often 10-15W. Ryzen 7 5800U and various others are configurable down to those numbers. Why has no one ever put one in a phone or in a tablet with a cell modem?

Where is Dell or Framework or any of these young companies from China that want to take a dent out of Samsung? Where, for that matter, is Samsung?

Trolls come out to say there is no demand for it, but do the math. Having open source drivers doesn't make your chip less attractive to the people who don't care about that, so if you design the chip you still get all of those sales and the ones from the people who do care. Meanwhile the other incumbents are hidebound and myopic, so if your hardware creates a software ecosystem that depends on it, you'll have it to yourself for a decent while before the recognition that you're getting 100% of those sales because you're the only one providing the hardware documentation they need can permeate through the relevant layers of the incumbent's bureaucracy. And by then you've reinvested that revenue into making a better chip than them.

First mover advantage is valuable.

Which is why phone projects should start with new hardware.

wouldn't that mean they would be starting with new hardware that doesn't exist? So they'd have to make their own?

Maybe Framework might one day be able to support the market with new hardware. We need an open hardware consortium to promote software freedom. If we don't have an open hardware environment that is a foundation for free software, we are fighting a losing battle.

Yes, that's exactly it.