Its a lot easier to fix one problem at a time than 10 problems at the same time

You get up in the morning and your car won't start. You run through the quick checklist. Fuel in the tank? Yup. Key in the ignition? Yup. Battery charged? Nope

So you jump the battery and the car starts and you're off

This is what happens when they slowly go through the process nodes. Each node has a new quirk that needs to be identified, triaged, and fixed and once that's done they can start getting product

Its a lot harder if your car just went through a wild science experiment and won't start. If your battery is dead, ignition switch is broken, fuel pump is missing, and spark plugs are bad its going to be a whole lot harder to troubleshoot that car and get it going since you have to identify and fix all problems before you get any results

It doesn't make business sense to go from 90% yield and then hop to a node with 2% yield that slowly ramps up to 90% over the next decade. Take a sequence of small steps that each have a small step rather than a giant leap that either works or utterly bankrupts the company