Fuck, now that I have the sources I couldn't resist looking into it a little. Here's their methodology on the PEPFAR numbers (which is the USAID program to counter AIDS/HIV):

The state department says 20.6 million people are receiving anti-virals, and then the CGD uses a different paper (Glaubius et al 2021) which, according to the CGD, says that basically 8% of people without anti-virals die a year. They do this math: 20.6 * .08 = 1.648 million per year.

Except, the Gluabius paper doesn't actually specify an annual case fatality rate. Instead, it just says that the median survival rate for a 15-24 year old that is not on anti-virals is 12.5 years. I didn't understand how the CGD paper got 8% out of this, so I asked Gemini 2.5, and it explained that basically if the median survival rate is 12.5 years, that means 50% of them will have died by 12.5 years. So in each year, 1 year/12.5 years = 8%. 8% will die each year.

Except, of course, as we know, HIV does not kill at a steady rate. It takes quite a few years without treatment for HIV to turn into AIDS, so the vast majority of those deaths would NOT be in the first year like the tweets and the NYT article claim. There might be relatively few deaths in the first year, but there would be TONS of deaths in the next 12.5 years. Around half of the 20.6 million receiving anti-virals would be dead, and probably more than that due to increased transmission (of course, assuming no other countries start their own funding to pick up the slack). To be fair to the CGD, it probably is fair to say that PEPFAR saves 1.6 million lives a year, but it's NOT fair to say that 1.6 million people will die to AIDS this year without PEPFAR, but they're not making that claim.

So yeah, you were right to call bullshit. The tweets and the NYT article were both dead wrong, at least with the AIDS aspect (lets cut the NYT a little bit of slack, it is an opinion piece and therefore didn't have a rigorous vetting). Although, some of the other things listed, like Malaria, have more immediate effects, and actually might lead to deaths in the first year around those numbers, but I don't really know enough about them to look into them.

PS: ALSO the CGD seems to have their math wrong. If the median survival is 12.5 years, then you would expect half of the 20.6 million to be dead by year 12.5. So, it's 10.3 million divided by 12.5 years = .824 million dead per year, or 4%. It's half of what they said, which was 1.6 million, or 8%. That being said, it's a really rough estimate to begin with, and the 12.5 year median survival only applies to young people, and that survival rate is much worse for older people, so the numbers might not be THAT far off in the end anyway.