Time to Abandon the Public Cloud?
As a bit of background to set things up, I regularly attend IT Virtual Roundtable discussions; it’s a regular thing where IT professionals gather and discuss technical topics in depth. Two topics that are always mentioned are “Cloud Adoption” and “AI Adoption.”
AI Adoption, well, we are all going through that. We are finding we are being forced to adopt it whether we want it or not, but suffice it to say, organisations can subscribe to a commercial AI solution to achieve some theoretical benefits of having a different method of understanding their own data.
“Cloud Adoption” is the process of moving an organisation’s workloads off of local on premises traditional servers to Cloud Platforms like Microsoft Azure, Google Cloud, or Amazon AWS. Essentially, instead of going through the considerable expense of setting up and maintaining a private datacenter and all the hardware, software and networking, just lease it.
I am not going to argue the advantages and disadvantages of Cloud Adoption, that’s been going on for over a decade now. But during the IT roundtable, as we were discussing AI adoption, I made the argument that adopting AI, as it stands right now, is a particularly pernicious form of vendor lock-in. For example, the most common solution is to use is Microsoft Co-Pilot. License it, let it ingest data, and tune and configure AI Agents against that data. As it stands, it does not look like that AI configuration is portable to another vendor. If an organization does find AI useful, but months or years later, for whatever reason, finds that Microsoft Co-Pilot is no longer an option, there is no way to port that useful solution to Google Gemini (as an example.) All that data, all that work, all those found efficiencies, gone.
Vendor lock-in is a huge issue in IT. Migrating away from a key vendor is usually between challenging and impossible. The rise of the Public Cloud has created its own form of vendor lock-in. At a certain scale of complexity and scope, moving from Azure to AWS could be a nightmare of a process, even when well managed.
As I pondered the vendor lock-in issue, the phrase “for whatever reason” kept replaying in my mind. I then realized all of us all face a new threat, one none of us saw coming. What if the entire nation that is the United States of America falls into chaos, instability and civil war?
The vast majority of the Public Cloud is US based or US reliant. What if it suddenly ceased to exist because of events? These massive data centers hosting the cloud are power hungry and resource hungry beasts. They require stable and functioning modern nation states to exist and run.
No one wants to have to maintain a data center in a war zone…
That vendor lock-in to a Cloud vendor has a new threat, a new way for things to go awry.
Granted, if you are an organization in the same war zone that your cloud vendor is in, you have multiple threats to deal with. Your own survival as a human being trumps requests from HR to ensure the employee absentee reporting portal is up and running.
But for those outside that zone, should you look at reshoring your systems back to on premises solutions? Or look for a cloud vendor that is not US based, and is not heavily reliant on a US parent?
Will Microsoft Azure, Google Cloud and Amazon AWS be reliable solutions that can be counted on during a period of civil strife? Are you willing to bet your entire organisation on that?
IT risk analysis is a weird beast at the best of times. It just got a lot weirder, and a hell of a lot scarier.
