A friend wrote as follows:
“In most places, people don’t advertise they have it - San Francisco is an exception. In most countries, the Public Health folks don’t want to admit the thing is completely out of control and there is no vaccine/cure in sight because the virus keeps changing. And the available numbers fluctuate widely from year to year (I have four years of the Economist handbook of statistics, and the AIDS numbers vary a lot over that period). Then there is the distinction between HIV and “full blown” AIDS, so depending on how the line is drawn, the numbers can be shifted in either direction. All in all, I’d be surprised if most AIDS numbers don’t have a 50% uncertainty (probably always on the high side - they could be 50% higher), and in some cases it could be a lot more than that.”
So, is there any reliable source of AIDS statistics in the US? What about the rest of the world? What kind of accuracy can we expect in the numbers we see in the media? How does this compare to statistics published for other diseases?
One thing to keep in mind would be the difference between HIV infection (a retrovirus) numbers and AIDS (a syndrome) numbers, which are presumably two different things but which get reported in a mixed fashion in a lot of media outlets.
In general the global projections (future) vary widely. Projections for the US from ten or more years ago were drastically overstated; HIV awareness groups argue that this is because they did a good prevention job and some critics charge that this was an overstated scare campaign based in worst case scenario projection models.
For several years I used to do global studies research for an NGO that collected a lot of annual country-by-country health data and I have to say a lot of the numbers struck me as fishy on the high side. The African numbers always struck me as inflated; it amazed me that societies in time of plague were also posting 3-4% or higher annual population growth rates with increasing GDPs and slightly higher life expectancies.
When you look at how the numbers are generated you see that instead of individual reports to a central authority (such as the CDC) for individual cases (even then there are probably a small minority of presumptive diagnoses, based in symptoms and not an HIV test), WHO and UNAIDS in Africa is pretty much taking small HIV test samples and projecting that over very large populations. The last thing anyone can afford in a lot of these places, especially several yrs ago when the testing was more expensive, where basic health needs go unmet are reliable HIV tests and and central reporting.
An Austrian doctor named Christian Fiala had this to say about the WHO projection method: “[A]ll registered AIDS cases worldwide are noted by the WHO in Geneva. As there is certainly an unknown number that are not registered, the WHO multiplies the registered cases in order to reach an estimate of the ‘actual’ figure. The multiplication factor, however, increases every year. In 1996, the WHO multiplied registered AIDS cases in Africa by 12. In 1997 this had jumped to 17. In the last one and a half years alone, 116,000 new cases of AIDS in Africa have been registered with the WHO. In the same period, the WHO has raised its statistics for the estimated cases in Africa by a whole 5.5 million, thus multiplying the reported cases by 47.”
There’s a whole lot more Western money out there for African governments’ health departments and international aid to fight AIDS than there is to fight the whole range of component diseases and health disorders that Africans with compromised immune systems actually die of, whether they are HIV+ or not. Malaria, for example, is an AIDS component disease in Africa (i.e. malaria + HIV = AIDS diagnosis, but either one by itself does not). If I’m working in the Ugandan health dept & I have a malaria patient, the sad fact is everyone involved has a better chance of getting assistance by just assuming & reporting that’s an AIDS case. That same money for malaria or even just malnutrition simply isn’t there. 
It dosn’t pay anyone to correct the record and say “We just had 50 patients die of treatable infections the West isn’t holding fundraisers for.”
One of the most interesting articles I’ve ever seen on this was by a South African journalist who was trying to do a story on the RSA as the “AIDS capital of the world” (seems that’s a different African country each year, another tip-off something in the process may be off) and came up with an article on how hard it was to locate an AIDS death. He also noted that the WHO figure for annual AIDS deaths in RSA was higher than the RSA’s own quite well reported total mortality from all causes for the year! (I couldn’t find this article in a brief web search but perhaps someone else could track it down?)
There’s also this letter to the BJM from a South African prison doctor.
Short answer to your OP: we don’t know, but in my judgement the (past and current) global numbers are probably high and the US numbers probably a tiny bit high. The projections, based upon earlier projections vs even the inflated reported numbers, have fortunately been very wrong, especially for the US.