Using 3 COTS (Commercial Off The Shelf) components with 90% reliability can ensure 99.9% redundancy instead of having a Unique very reliable and even costlier 99.9% custom component can really lower costs.
But it can happen only if mass is not a constraint.
The Perseverance Mars rover cost $2.4 billion, which works out to a few thousand salaries for just under a decade. Thousands of people are needed to build this rover because landing stuff on Mars is so hard that subsystem masses must be tracked to a tenth of a gram, on a system that weighs a tonne. The whole thing is meticulously handcrafted from custom silicon, PCBs, titanium tubes, motors, cameras, and other awe-inspiring instruments. Starship will be able to land 100 of them per flight. Now what? How can NASA feed a team that makes one feather light rover per decade for a billion dollars if the demand just jumped by a factor of a thousand and the unit cost fell by the same amount? Set up a production line? Work out how to make them with a team of ten? Build one every two weeks?
They should put that team to work on next-generation tech, the stuff that's not yet a commodity. They can start working on habitat construction materials and hardware, for instance.
Although it is interesting to consider that we've put a lot of expense into optimizing payloads that, in retrospect, would have been smarter to put into better launch vehicles. SpaceX probably isn't going to spend $2 billion developing Starship (even if Boeing would have.)
IIRC, SpaceX is getting $2.89B from NASA for the Artemis lander, which proposal is based on Starship. Although, the GAO put that on hold recently, after complaints from BlueOrigin and Dynetics. Hardly surprising, I guess.
From Casey Handmer's blog: https://caseyhandmer.wordpress.com/2021/03/04/sls-what-now/
The Perseverance Mars rover cost $2.4 billion, which works out to a few thousand salaries for just under a decade. Thousands of people are needed to build this rover because landing stuff on Mars is so hard that subsystem masses must be tracked to a tenth of a gram, on a system that weighs a tonne. The whole thing is meticulously handcrafted from custom silicon, PCBs, titanium tubes, motors, cameras, and other awe-inspiring instruments. Starship will be able to land 100 of them per flight. Now what? How can NASA feed a team that makes one feather light rover per decade for a billion dollars if the demand just jumped by a factor of a thousand and the unit cost fell by the same amount? Set up a production line? Work out how to make them with a team of ten? Build one every two weeks?