The Microsoft AI research division accidentally leaked dozens of terabytes of sensitive data starting in July 2020 while contributing open-source AI learning models to a public GitHub repository.
I do not think anyone belive clouds are cheaper. For a stable workload probably 2x as expecive. Especially when you also count the new finops department you need to know what you are actually paying for in the cloud.
What cloud do give is virtualy infinite capacity, infinite scale out performance, instant availabillity and scaleabillity up to a global presence, no up-front cost, no tear down cost, bragging rights, no long running contracts and api’s for EVERYTHING.
Let me add another important point: outsourcing responsibility. In case of a data breach, you have someone to sue and you don’t need a whole internal team to be up to date on the latest security topics. Instead, they just have to be able to manage the web interface (not saying that is easy, just less subject to changes)
I do not think anyone belive clouds are cheaper. For a stable workload probably 2x as expecive. Especially when you also count the new finops department you need to know what you are actually paying for in the cloud.
What cloud do give is virtualy infinite capacity, infinite scale out performance, instant availabillity and scaleabillity up to a global presence, no up-front cost, no tear down cost, bragging rights, no long running contracts and api’s for EVERYTHING.
Edit: I did see you write theoretically ;)
Let me add another important point: outsourcing responsibility. In case of a data breach, you have someone to sue and you don’t need a whole internal team to be up to date on the latest security topics. Instead, they just have to be able to manage the web interface (not saying that is easy, just less subject to changes)
Ding ding ding. It’s all about outsourcing accountability as much as possible. Always need a finger to point at if things go wrong.