The cost is considerably lower if you watch out for a11y from the get-go, retrofitting it is more expensive since you have to retain the existing behavior, sometimes of existing and complex but non-accessible components. Add to that the need to e.g. caption all the existing pictures and it gets even worse.

It really depends on the specifics; for something like this I suspect it's a non-trivial investment. For a lot of other things it's not that hard.

For a lot of things a18y features are just good features in general; zooming text for example is something loads of people do, not just blind or low-vision people.

For the specific project of making this remote browser accessible, my wild guess is that if Cloudflare were to hire me to work on the project (no, not available at the moment), it could easily take a few months, but probably not more than a year. They could probably cut down that time if they hired away someone from the Chrome or Edge team who's actually an expert on Chromium accessibility specifically; I admit my main expertise is in Windows accessibility.

It’s part of every product team’s baseline requirements to own and assess. It’s considerably easier to do that up front than to retrofit. Think of it as analogous to security in this situation.