I don't think "eclipse" is the right frame, because drones and nukes do different jobs. A nuke is about how much damage you can promise; a drone swarm is about how cheaply you can reach things. The interesting overlap is that the things drones can now reach include nuclear forces.
There's already a real data point: Ukraine's Operation Spiderweb in June 2025, where small drones launched from trucks deep inside Russia hit parked strategic bombers -- aircraft that are part of Russia's nuclear delivery force. Nobody escalated, but it showed that a few thousand dollars of quadcopter can put a strategic asset at risk without crossing any traditional nuclear threshold.
So my answer to "what would count as evidence" is: watch posture, not battlefields. If nuclear states start dispersing bombers, hardening shelters, moving to faster launch-on-warning, or treating attacks on early-warning radars as nuclear-level provocations, that is drones mattering strategically. Hardening and dispersal are stabilizing; hair-trigger launch is not.
The safeguard that seems most tractable to me is narrow: a mutual commitment not to target each other's nuclear command, control and early-warning systems with conventional or cyber means. It doesn't require banning drones, which won't happen, and it addresses exactly the "use them or lose them" pressure the SIPRI piece worries about.
#
There's already a real data point: Ukraine's Operation Spiderweb in June 2025, where small drones launched from trucks deep inside Russia hit parked strategic bombers -- aircraft that are part of Russia's nuclear delivery force. Nobody escalated, but it showed that a few thousand dollars of quadcopter can put a strategic asset at risk without crossing any traditional nuclear threshold.
So my answer to "what would count as evidence" is: watch posture, not battlefields. If nuclear states start dispersing bombers, hardening shelters, moving to faster launch-on-warning, or treating attacks on early-warning radars as nuclear-level provocations, that is drones mattering strategically. Hardening and dispersal are stabilizing; hair-trigger launch is not.
The safeguard that seems most tractable to me is narrow: a mutual commitment not to target each other's nuclear command, control and early-warning systems with conventional or cyber means. It doesn't require banning drones, which won't happen, and it addresses exactly the "use them or lose them" pressure the SIPRI piece worries about.