Will nuclear ICBMs in their current form be obsolete soon? Here's the argument:

• ICBMs' military utility is to make the cost of intercepting them totally impractical for three reasons:
  • Intercepting a reentry vehicle requires another missile with higher performance-- the rule of thumb is an interceptor needs 3x larger acceleration than the incoming missile. This means interceptors cost something like $5-70 million each depending on range
  • An ICBM has enough range to target basically anywhere, either counterforce (enemy missile silos) or countervalue (cities) targets, so to have the entire country be protected from nuclear attack the US would need to have millions of interceptors compared to the current ~50.
  • One missile can split into up to 5-15 MIRV (multiple independent reentry vehicles) which carry their own warheads, thereby making the cost to the defender 5-15x larger. There can also be up to ~100 decoys, but these mostly fall away during reentry.
• RVs are extremely fast (~7 km/s of which ~3 km/s is downward velocity) but their path is totally predictable once they enter the boost phase, so the problem of intercepting them basically reduces to predicting exactly where they'll be, then putting something ~100 g or larger in their path. The RV's own kinetic energy will completely destroy it.
• My understanding is that sensing had historically worked like this. The launch is first detected by primitive satellites, but there are too many decoys to tell which are the warheads and the resolution is poor. Interceptors thus require extremely high performance, meaning expensive systems like thrusters for moving sideways at tens of g at the last minute, radar, and cooling systems for IR cameras; including all the fuel the interceptor masses thousands of times what it needs to.
• Sensing is quickly being solved. One new generation of satellites (HBTSS) can track warheads after their booster burns out (at which point they become much dimmer), and advertises "weapons quality track for fire control" even for hypersonic glide vehicles, which are much harder to track than ICBMs. The first two were just launched last year, but it would not surprise me if hundreds will be operational soon: US has well over 1000 Starshield satellites in orbit, many equipped with radar and infrared cameras. Through SpaceX, they have the ability to launch tens of satellites per month if they needed to. China doesn't have the boost capacity of SpaceX but they would still be able to launch 1000 sensing satellites.
• The ultimate cost floor for intercepting ICBMs in the limit of perfect sensing is interceptor drones. Ukraine and Russia are in the middle of a war where most casualties are caused by drones, providing extremely high incentives to develop interceptor drones, which already have a top speed of 300 km/h, max altitude of 5 km, and basic infrared cameras at a unit cost of $1k-10k. The idea is the space-based sensors know the missile's velocity well enough to locate the target within 100 meters with 15 minutes of warning, the drones launch to a position 3 km in front of the target and are guided into exact position by radar and their own sensors as the RVs reenter, with sufficiently low error that they never need to go above their top speed or accelerate more than 1 gee.
• The other thing keeping ballistic missile defense from being practical during the Cold War was treaties like the ABM treaty, but today there are basically no active nuclear treaties between Russia, the US, and China.
• The internet seems to think it's still really hard to do missile defense against ICBMs. It seems like this was true in the Cold War but we can no longer be confident it's true today after 40 years of peace, with Boeing and American drone manufacturers vastly less competent than SpaceX and Ukraine's/Russia's. My guess at why the meme has persisted is that it is highly threatening and destabilizing to advertise one's ICBM defense capabilities. But many officials are talking openly about the predictability of ballistic missiles, and there is an arms race for other hypersonics for a reason.

My current guess is that the SM-6 and PAC-3 missiles either are already reliable or would be if the US had more practice, and if sufficient incentive exists, the cost of terminal ballistic missile defense will gradually decline as sensing improves, from the SM-6's $5M unit cost to lower and lower cost missiles like $400k AIM-9X until one could defend every city and silo in the US or China with a fleet of 100,000 interceptor drones. This would last until there are enough hypersonic glide vehicles and cruise missiles to reestablish deterrence (because they can outmaneuver cheap interceptors). But MIRVs would never return to their previous state of heavily cost advantaging the attacker, because Russia's state of the art ICBMs only claim to be able to carry 3 HGVs, vs 15 purely ballistic MIRVs. Also HGVs and HCMs are vulnerable in other ways.

Edit: Another option pointed out by commenters is that ICBMs are retrofitted to carry reentry vehicles with limited maneuvering ability, which could happen faster.

Curious to hear if I'm missing something major, as I'm by no means an expert here.