David Heinemeier Hansson

Basecamp has had one foot in the cloud for well over a decade, and HEY has been running there exclusively since it was launched two years ago. The cloud excels at two ends of the spectrum, where only one end was ever relevant for us.

Why we're leaving the cloud

There are complexities that cloud services do provide that can get you caught between a rock and a hard place. Edge computing is one instance of this. By server static content across many servers and getting closer to your consumer you drastically decrease the lag time between calls to your server. So you will forever be tied to mega computing services that have the capability to do this.
Consumers want their product on demand and don’t want to wait, a second or two does matter in the world of computing. While I have always valued David’s approach of simplification, there is such a thing as to simple. It was one of the reasons I moved away from Hey and went back to Google for my company email. We needed more robust product offerings that tied back to to traditional use and that customers and clients. I believe this i being overlooked here.
Amazon is a business too and mentioning their 30% profit margin, while true, isn’t the whole story. They achieve that 30% because of efficiency of scale, efficiencies most businesses cannot achieve because they will never be able to house all the products under one roof.

There are Web3 ‘decentralized’ projects which have 80% of their nodes on AWS .

This has always been true. The cloud is not, and was never meant to be, a solve it all. For a lot of businesses it makes sense to use the cloud for non-mission critical data (email, HR, etc…). For some it makes sense to run in the cloud until you reach a size that it might be better to goto a hybrid model or on prem. There are also things to consider if you do business in mulitple countries… these are complex decisions not taken lightly.

Wow! My sentiments exactly for several years - kudos, not the bravery to take a stand and make a move, but rather being logical about maintaining your own autonomy by returning to the roots of the internet that has served us well…Also, I appreciate the creation of Hotwire (Turbo &amp; Stimulus.js) and HEY, which are heading in the opposite direction as the rest of the development herd by returning to the client-server response model that served us well at the inception of the internet and simply needed solutions like you are developing to solve the original problems. We do not need to complicate web development with SPA-style solutions, that, in the end, get complicated further when they realize that, OOPS, static website generation, and relational database models are the correct paradigms, so we must further complicate our libraries and frameworks by creating SSR frameworks for our frameworks (Next.js/Nuxt.js).
Decouple and de-complicate. Rethink and return…to the web/internet foundation of paradigms creating uncomplicated, simple solutions for complicated problems like Turbo, Stimulus, HEY, Basecamp, Shape-up, and of course the new Ruby on Rails 7.0…
As always, David, my hat is off to you and your 37 Signals team!

I always run my own servers (20 years) and always was saying that cloud does NOT make sense for SMBs (if only you are not really small with no traffic and with no human resources).
Like… manly people pay for the AWS, GCP, Azure certifications, i mean - entire CERTIFICATIONS about the fancy cloud providers API abstractions. But how about passing real networking certification, without any fancy abstraction? Isn’t that more valuable to know the fundamentals of these things?
I really saw people, who don’t know how firewall works, but he knows how to setup VPC! That is alerting to me.
On top of all this, we have a great tooling to manage your own infrastructure. PXE, terraform, ansible, iLO, iDRAC, SDN, etc., etc. We have multi-cloud initiatives and approaches for cases when you really need an rapid scaling.
And lastly, #feedingthebeast was never a good strategy.