That's pretty broad conclusions for Razib Khan to be drawing from scant (and biased) data. (The population of Indians in the United States in a non-random sampling of the Indian genome, especially since our immigration policy is largely selected on the basis on already having family networks based in the US. I would expect to see a lot more clustering than in an Indian sampling).
This is a better and more holistic description. I'd quibble withs some aspects of its framing but it's mostly nitpicks. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3514343/
The gist of it is that there are a lot of discrete, endogamous subgroups within India formed by successive waves of migration by different groups to where they are functionally distinct ethnic populations. So it wouldn't be all that different from racial differences in a plural society like the United States. What's remarkable is the maintenance of this plurality, as well as in-group social norms, over a very long span of time rather than it becoming a genetic melting pot as we might have expected from experiences with Europe. But even there you still have distinct groups with strong endogamy norms (e.g. Ashkenazi Jews).
There's also additional complexity because in many places there isn't just a norm of endogamy, but consanguinity (cousin-marriage). Which is certainly going to make groups look more endogamous even though it would lead us to draw subtly different conclusions about what that means in terms of historic cultural norms.
TL;DR it's a really complicated matter that doesn't lend itself to summarization into a single, clean, generalizable narrative.