I don't think that's true. If you simply removed the de facto state endorsement of parties - if you only listed candidates by name on the ballot and moved to a California type open primary instead of having partisan primaries - you probably wouldn't end up with two parties like we see today.
Party primaries are a good example of this. They're also FPTP, but that hasn't lead to the emergence of two subfactions because such a thing is hard when it's not entrenched in the ballot. Instead primaries have much more loose coalitions of groups and endorsements that shift around in every election. Even in a particular election there are often more than two.
I mean the video explains it way better than I ever could, and there are a bunch of other political systems and the maths behind them explained ELI5 style.
It's... a useful model. It turns out not to work in practice everywhere, though, as there are complicating features with geography and ideology that tend to add parties. Canada, for example, has 2 and a half effective parties (Liberal, Conservative, NDP) in general, and more in Québec. The UK also has two and a half (Con, Lab, Lib Dem) plus powerful Scottish and Welsh nationalist parties, plus an entirely different party system in Northern Ireland. So 'tends towards' is going a lot of work.
EDIT: Also, if the system tries to equalize at two parties, why we do not see a similar process inside parties themselves?
Having said that, electoral politics within parties doesn't tend to be as important as other things: bureaucratic control of the party machinery, for example.
FPTP as in the US will always yield a two party system in the long run as thats the only way you can “win” in it.
More elaborate systems would get you an equilibrium of multiple parties, with much better overlap of what people want to their representatives.
Grouping would indeed always happen, but you wouldn’t need to go to 50% of society to get what you want, so the wants and political speech can be much more varied and relevant.
The best explanations of how it all works I’ve seen were from CGP Grey’s videos https://youtu.be/s7tWHJfhiyo