We're only testing those who are hospitalised. Given that we can expect only 20% of known CV-19 cases (from the Chinese data) end up in hospital, that implies that there are 240,000 cases in total.
This, though, is likely the tip of the iceberg. I was careful to say "20% of known cases". That's because many cases, probably the majority, never reach the attention of the medical authorities. I've seen estimates run from only 1 in 4 cases being spotted all the way up to 1 in 19 being spotted. Ferguson in his paper modelling the epidemic assumed that 66% of all cases were spotted (rather high, and contrary to all the other published analyses I've seen). Chinese data suggests that only 1 in 7 infections were ever discovered by the authorities. Taking that into account suggests a true number of infections of between 360,000 and 1.6 million (using the Chinese data as the high estimate).
Many of these people will have recovered from the infection (though some still have the virus, and it's not clear as to whether or not they're still infectious). Many will be asymptomatic. And many will have sub-clinical symptoms: a slight snivel or sore throat, perhaps, or just feeling subpar for a few days. At the moment, it's impossible to tell, and without a intensive campaign of antibody testing to tell us who's been exposed to the virus and detailed medical histories from those who test positive we won't know.