> Is my understanding correct?
No.
> Would the fix have to be deployed on every computer that existed, or just on cloud?
Cloud providers had some ability to deploy the fix en masse. I wouldn't count out virtual hosts either hosted in your own data center (there's rich API's for Hyper-V and VMWare and others).
It's the bare metal servers which could be a challenge. There are solutions which let you have an IP based KVM and tap into it with automation (potentially a harder solution).
To be honest, if you're asking this question, you're not fully understanding the technical details. CloudStrike is widely deployed for security and compliance both on servers and client machines, both physical and virtual.
> As I understand, air traffic systems are not dependent on software so they would not be affected.
They have computers - of course they do. The air traffic control systems were not the only things impacted - booking, scheduling, ticketing systems were all impacted as well. The digital signage with flight status was impacted. It was all over the place.
> My understanding is that if the airline uses Microsoft's workspace and also Azure for cloud, all of those things would have been affected and several other dependent systems would also have been compromised
This was not an Azure outage specifically. Everyone running Crowdstrike Falcon for Windows was impacted by it, regardless of where the machine was running. If you were running Windows Defender or other endpoint protection on-prem or on Azure or anywhere else for that matter, you weren't impacted.