As you'll see, the three biggest impacts on that graph, as measured by dollars, are: Labour, Extreme Temperature Mortality, and Coastal Property. To a first approximation, each of those are a quarter, and everything else is another quarter. I was going to do Labour, but the first one along was ETM, so I thought I'd do that instead. Note first that The study also only considers deaths related to extreme temperatures, though extreme heat will... do other stuff too; so this isn't complete.
$141B/yr is a Big Number, but we all known that Big Numbers without some kind of context or referent are meaningless. Let's try to put it into place. As the report says, This analysis estimates the number of deaths over the course of the 21st century attributable to extreme temperatures in 49 cities in the contiguous U.S., which account for approximately one third of the national population. So that's a bit funky: you might at least expect them to multiply it by three or something. That number - by 2090 - represents 9,300 deaths/yr which is ~$15M/life by a quick in-my-head calc; and looking at footnote 142 I see I'm right. Soooo... how might we assess that? Imagine (I do this just to wind you up, you understand) we were talking not about people, but about industrial plant. Then we'd want to compare those 9,300 people to the replacement rate; which is currently abut 1%, of about 300M, which is to say the population is growing by 3M/yr. Against which 9,300 is about 0.3%, if I have my maths right. Scaling that, we might expect a similar number from the GDP figures: $141B is large, but the US GDP in 2018 was ~$20Tr, of which $141B is about 0.7%, which is close enough, as I haven't been very careful about what year I'm looking at.
The next interesting thing to consider is Mortality from extremely hot days decreased more than 50% under both RCP8.5 and RCP4.5 in 2050 and 2090 when the human health response to extreme temperatures was evaluated using Dallas’ threshold for extreme heat (in all cities with thresholds initially cooler than Dallas), as a sensitivity analysis to consider the effect of adaptation. Which is a good place to remind ourselves that, for simplicity, the initial figures are with no adaption, a not very plausible scenario. Unfortunately they don't really explore Dallas-world in depth, so it is nothing but a sensitivity analysis; but if you think $141B is large then you presumably think $70.5B is also large, and so should be very interested in exploring an effect that large.
If you follow the spiral of bubbles down to near the centre you'll eventually come to "shellfish" at $23M/yr. Now I quite like shellfish, some of my best friends are shellfish, but compared to the uncertainty in $141B, 23M is less than a rounding error, so why they bothered - other than pressure from the shellfish associations - I really don't know.
I should also point out that these are RCP8.5, which you can call too high if you like, and can consider RCP4.5 if you prefer, in which case you get to cut the numbers in half.
Update: I didn't explicitly note it here, but their estimate of the (small) changes due to winter mortality don't look plausible to me, especially in the face of stuff like Excess winter deaths in England and Wales highest since 1976 from the Graun.
* The management apologise for any inconvenience - aka Estimating economic damage from climate change in the United States by Solomon Hsiang et al., Science 30 Jun 2017.
* 10% of GDP? - ATTP