Personally, this makes me suspect bad architecture. If they've got well-factored APIs, the additional cost of maintaining a modest web presence should be pretty low. From what they describe, mobile is now somewhere between 40 and 60% of their traffic. Even if they were pretty sure that the trend to mobile would continue, why wouldn't they just hedge their bets by investing less in the HTML front end until the web traffic is really negligible? Few people use faxes anymore, but a lot of businesses still publish fax numbers just in case.
Assuming that this isn't purely internal politics, then my best guess is that they have a lot of terrible web code. Somebody saying, "Mobile is the one and only future! Look, shiny future!" gives people a chance to declare bankruptcy on a lot of technical debt while appearing as forward-thinking leaders, rather than bozos who made a giant mess.
I do think that some amount of bad architecture is involved, but flipkart does have a lot of money, and a lot of good engineers. They even have cool stuff on github [0], which needs good people. They have likely fixed their architecture issues by now.
Flipkart has money, and lots of it (current valuation is $11 billion). There is absolutely no way that they are considering this on a cost-analysis basis.
Your theory that people with lots of money and stuff on Github have no architectural problems does not match my experience.
Also, I like you using the web to tell me the web is dead. Do go on.
There are some big social media companies whose mobile traffic so far outstrips their web-usage, that they consider from time to time dropping the html frontend. It seems inconceivable to us, but it makes a certain amount of sense.