I think it was in a way.

Having an old school style IT structure was a pain. Maintaining your severs, networking gear if needed, OS updates ... on and on.

For very small teams (maybe not the well funded start ups) that could eat a lot of time and trouble.

You used to be able to scale up to millions of users in less time than you could memorize all the AWS acronyms, and with lower hardware and bandwidth costs to boot.

It’s true that it would take a couple weeks to order all the hardware and a couple more to install and configure it, but that’s about the same amount of time you spend writing all the terraform and Kubernetes scripts and configuring cloudformation and VPCs and IAMs.

Yeah, but then you wouldn't have high availability to the same level, for example. I mean, you could, but then tack on another 100 non-AWS acronyms to your on-premise solution.

Looking back 20-30 years, it absolutely was a primary limiting factor. Massive capital outlays were required to serve the kind of traffic you can get with the free tier of modern cloud infra providers.

That depends on what you are doing. Serving static websites needs a great pipe, but doesn't need much CPU.

Yup, just gotta lay good pipe.

Absolutely not.

not really, i built websites w hundreds of thousands/millions of users on a VPS that cost me like $30/mo.