This is bad data analysis. It has all kind of issues: using correlation to imply causation without any proposed causal mechanism; various selection bias; no context given for how the companies analyzed were selected; laughably large ranges for data points; etc.

One of the worst parts, in my opinion, is how the author misunderstands the meaning of "Actually Indians," which is about the data labeling workforce, not the white collar technical workforce.