Giles Thomas makes the case that the Python programming community numbers in the low millions. This seems right to me: that's a large community, but it's not quite at the level of the most popular programming languages. That size is supported by this chart, which has impressed me as "feeling right" when it comes to the popularity of various languages.
One point, though, is that Python has made very significant inroads in the scientific community, which I believe is a key influencer and leading indicator: the libraries that scientists build become building blocks for future work. When you look at the history of programming languages, you see that scientists and engineers were clear driving forces behind FORTRAN and although C and C++ were broadly popular, their performance benefits made them extremely popular in labs as well.
I'm not sure that the popularity of Python in labs is going to be captured by metrics that focus on the professional programming community, so if anything, I suspect that the Python community might even be a moderate amount larger than Giles suggests.