I don’t see how that math works. The first case in China was in late November, now over 100 days ago. With your every 3 day double billions would have it by now.Never said doubling time was one day. I used 3 days which is on the lower end of the observed range globally (3 - 6 days in most countries excluding China). If the double time was one day then it would take 23 days to go from first case to 4 million. However, no outbreak can sustain exponential growth indefinitely as it eventually runs out of uninfected people to infect. See below for data. https://ourworldindata.org/coronavirus#the-growth-rate-of-covid-19-deaths
The number with positive tests is obviously lower than the actual number but whether it is off by 10, 100, or 1000 isn’t settled yet. Hopefully the number that had it and recovered or were asymptomatic but never tested is higher because that would depress the mortality rate and mean that we are further into the outbreak without yet overloading hospitals. That would mean fewer left to get it so we in theory are closer to the inflection point then if testing was perfect.
The truth is we have no idea how many people already had it, have it now, or will get it. Everyone is just plugging numbers that are almost certainly wrong into equations and spitting out answers that are almost certainly wrong