A source for what you are saying is the linked SF article [1] which indicates that of ~3k evidently sick people with flu symptoms, 0.1-1% were confirmed to have COVID-19. Though note this is not random population sampling, on an interesting scale, so ignores asymptomatic question. Worldwide analysis of death rates [2], indicates #infected-so-far = 700-1600 x #dead so far. And data from Iceland [3], the only country with widespread sampling of asymptomatic population, is aligned with these numbers. Based on the analysis and support from Iceland, Covid-19 testing to see if patients actively have the disease are likely inaccurate and also a policy of only permitting testing of severe symptom patients is depressing reported numbers of once infected.

[1] https://www.sfgate.com/coronavirus/article/Think-you-got-the...

[2] Analysis of #infected today from #dead today, using following assumptions:

1. death statistics are reliable, since cause of death tests can adequately collect fluid samples (a major source of inaccuracy in COVID-19 testing), source = talking to medical staff at Bay Area sites.

2. E[days between first infection and death] = 14 (lost the source at the moment), Therefore, number of dead today D => number of infected 14 days ago = 100 D

3. rate of death increase consistently trends around 25%/day or 15% with full lockdown, source CDC situation reports worldwide excluding regions with restricted access to journalists

4. therefore, #infected today = #infected 14 days ago growing at 25% daily = 22X, or 7X if your region was locked down for last 2 weeks

5. so #infected=2200D or 700D

[3] Last week, CDC reported 2 deaths today for Iceland, and admittedly-non-random sampling by DeCode indicated upto 1% of population have/had virus, upto 3600. Iceland has done strict quarantining of confirmed and suspected cases, and extensive testing since February, so despite no lockdown, I would put them at the 700X end of the estimate. They report ~1000 confirmed cases, so they would actually estimate 3600-1000 infected so far. My estimate would have indicated 2200-700 X D (= today's deaths, 2) = 4400 to 1400 cases.