Either way, you have to build more production capacity, it could be done with 18650s or 2170s or prismatic cells.
Tesla didn't spell this out but (Maxwell) DBE may work better with wider diameter cells. (Reports of cracking if wound too tight some years ago).
And the capex efficient scaling described in Battery Day is only optimal with DBE.
If you are saying a form factor doesn't have magical properties, I agree, and many overstate the importance of form factor.
In fact, an example of Tesla at their best is the regular Plaid Model S with 18650 cells, the specs are great.
Of course Plaid+ is likely to have 4680s, but we can't attribute the difference in specs purely to form factor.
What form factor does deliver is lower costs and a bit more energy density due to less packaging.
But long term Tesla decided on 4680 for their own cell production and as a form factor that suits vehicles and energy storage.
Compared to the CATL LFP pack a pack built with Tab-less 4680 cells has the following advantages:-
1. It can be a structural pack (lower weight, and cost)
2. Faster changing - due to Tab-less (really multi-tab)
3. Better integration with Tesla BMS, (easier to manage SOC, balancing, thermal management.)
4. In house production, or multiple external cell suppliers.
5. Structural packs built with 4680 cells are probably signicantly easier to recycle. (Standard form factor helps here)
So I think Tesla will gradually migrate to most or all products being built with 4680 cells, but this may take years.
Outside suppliers with 18650, 2170 or prismatic cells may migrate some of those production lines to 4680.
18650, 2170 or prismatic cells will still have some uses.
Eventually I can see Panasonic migrating most/all 2170 production at GF Nevada to 4680.