I think that's the point of the question: to see how you approach it. The 'correct' approach that they'd be looking for is 'Fermi estimation': https://en.wikipedia.org/wiki/Fermi_problem
Here's a brainteaser: spot all the flaws in that chain of "reasoning".