It's a valid approach in certain markets.
But if you're FAANG and launch that sort of MVP going against mature products from your competitors, often replacing a more feature complete product with an existing user base that you try to port over... well, you're going to get destroyed. As keeps happening to Google because they don't seem to realise that the market has moved on and you can't launch things like Gmail any more.
I think what you're describing though is if you cull too many features, i.e. if you try to launch _below_ the minimum that's viable for your product-market fit, you'll always get destroyed.
You get some more leeway if you're the only offering in a problem space, but going up against incumbents really raises the bar of what the viable minimum is.
That said, you _can_ offer less features but a better execution — e.g. there's a lot of challenge banks (Monzo, Revolut, Starling, etc.) doing well in the UK at the moment (in the sense of customer acquisition, at least).
I remember when they were applying for their banking licences, a lot of people working in traditional fintech were saying that they'd be a flash in the pan & that customers wouldn't switch to these new businesses until they had parity in terms of offering things like loans and mortgages. But the offering of trad banks was _so bad_ that people were willing to overlook a lot of missing features for a better UX on their day-to-day banking.
Like any tool, Lean Startup MVPs aren’t the right tool for every situation. You need extreme uncertainty.
If you’re a FAANG and you’re releasing a new version of a well established type of product like an email client, what giant existential risks do you have? FAANG have existing customers, we all know people use email. There’s no questions to answer, no uncertainty, so MVPs are the wrong tool for the job.
Why bother running experiments if we have 20 years of data showing us that we have customers and they like using email…
Then you haven't launched a viable product. You haven't made an MVP.
What is smallest product we can build to learn something from and iterate with. If you build something too small that no one uses because it doesn't have enough features, then you can't learn anything from it (apart from you need to build more).