The market effects of linguistic diversity work both ways: having more languages increases the potential demand for translation tools, but the sheer number of long-tail languages also makes it hard to justify compiling a corpus and making a considerable effort to study each of these languages enough to build tools for them. Some language communities may almost completely lack the kind of highly literate people who would make this task surmountable, and those who could help, may already be working another more stable or higher paying job.
My landlady needs to know this. She still thinks we all speak "Indian".
No, by the 2001 census it's about 50% of Indians who do not speak Hindi.
I don't see a reason for this fraction to have changed significantly.
I am not certain about entire Andra and Telengana. People in Hyderabad speak Urdu.
As a side nodes, some existing researchers who have an interest in other languages start work on them even if it’s not their native language. I know researchers who did that with african languages: by working with students speaking them, there is transfer of knowledge and preliminary work done. Mentoring the first generation of native speaker scientists is the key imo.