When the US decommissions a warhead for treaty purposes, they remove the plutonium trigger and put it on a shelf somewhere. The trigger assembly doesn't count as a weapon, and therefore it can't be deployed. However, since the plutonium trigger is not actually destroyed, these could be rapidly assembled into a deployable weapon. You would not expect any of these to be participating in a nuclear exchange. The US has relative few operational warheads but it probably maintains an inventory of 10,000+ plutonium triggers.
For treaty purposes, they often count delivery platforms, like missiles and bombers. This makes sense when you consider that it is easy to re-hydrate a warhead on short notice if you have the necessary components in inventory but not so easy to produce an operational delivery system which may have a lead time of years.
The point of keeping inventory is that it is possible, if deemed necessary, to quietly scale up the number of warheads prior to any nuclear exchange, should the situation warrant it. Also, there is very limited capacity to produce new warhead assemblies, so keeping the old triggers around provides optionality for a relatively low cost while also creating a patina of uncertainty about capacity for adversaries.
It's also a little complex in terms of the count. Bombers count as one independent of how many warheads they carry (the US has non-trivial nuclear bomber forces)
So I think a significant amount of nukes will not be used at all.
There’s a rich and horrifying literature on nuclear bargaining that’s suddenly relevant again, right up to a bunch of people (particularly in Russia) rehashing old HAK theories about being about to control the nuclear escalatory ladder in the midst of combat. It’s not a comforting time right now.
I wish I could remember where I read it. It still isn't particularly comforting, but maybe better than the all-out-nuclear-war scenario.
Unnervingly, Russian (and Soviet) policy has always assumed a smooth ramp from conventional to nuclear war and back, while US policy since Kennedy has seen nuclear escalation as an irrevocable step change in warfare (more reassuringly, China is, like the US, also unlikely to add nukes into a conventional conflict).
Russian doctrine therefore encompasses using a limited demonstration strike to compel an end to conventional hostilities on Moscow’s terms. It’s hard to imagine NATO not responding, though, because the obvious effect of not responding is Europe living under Russian control enforced by the constant threat of nuclear destruction. What that response would look like is uncertain, though warplans certainly exist, but it would absolutely set us up on an escalatory spiral.
What Russian doctrine says this? This sentence is a complete fabrication.
Over the last decade, Russian leadership has repeatedly floated public hypotheticals (including the Zapad military field exercises) in which nuclear threats (or, in their war games, simulated nuclear strikes) are used to deter NATO from operations “beyond the edge of the Russian empire” (such as majority-Russophone regions in the Baltic states), which is in keeping with an evolving strategy generally termed “escalate to de-escalate” in the West. In fact, where Western game theorists developed an intricate theory of nuclear deterrence vs compellance, the Russian literature refers to both using сдерживания, seeing threats to compel and threats to deter as functionally identical, strategically speaking. (Compare with the PLA’s “Science of Military Strategy,” both editions of which discuss the PLA’s nuclear arsenal in deterrent terms in the event of a general war in East Asia against Western forces.)
This is what unnerves game theorists about the Ukraine war and Russia’s poor performance there — if it weren’t for global public opinion (crucially China and India at this point), their doctrine certainly encompasses an escalatory demonstration strike (a tactical-yield weapon on a largely uninhabited area of farmland, for example) used to quickly “de-escalate” the war, freeze the lines at a position of relative Russian advantage, and force Ukraine into negotiations.
https://thebulletin.org/2022/03/russian-military-doctrine-ca...
I'm a bit skeptical of this article, which seems excessively optimistic, since the estimation it does at the very beginning:
> Arms-control treaties limit both nations to 1,550 deployed warheads, although FAS estimates 1,588 for Russia and 1,644 for the US
seems to interpret numbers differently (deployed strategically/readily deployable).
Edit: the two articles complement each other in a very interesting way, for example: navalgazing.net assumes that nuclear weapons would target strategic targets, cornell.edu instead densely populated areas (I think reality would tend towards the first case).
The navalgazing.net article states that the soot research is from 80s/90s, while the reserch referenced by cornell.edu is from 2020 (I think in this specific case, the navalgazing.net is at fault).