It's a case that we've been making for a long time now. In articles, case studies, interviews and news updates we've shown how AWS and its hyperscale peers are slipping. Call it the Cloud Exit. Customers are saving money when they leave AWS and co for more down-to-earth infrastructure.
High cloud bills are the push, and ever more cost-effective hardware is the pull.
The latest twist came when AWS admitted, after 18 years, that on-premises hosting can be cheaper and more attractive. Somewhere during our article about that, our total output on the Cloud Exit and improving server hardware climbed past 15,000 words.
A long read, or a quick listen
To wrap your head around all this Cloud Exit stuff you could set aside a lot of reading time, or you could ask the content-summarisation tool of the moment. We turned our words over to Google's NotebookLM and asked for a 9-minute “deep dive podcast”.
Massive advances in AI are somehow both astounding and unsurprising these days. Our verdict here is the NotebookLM is very impressive - it has a convincing 'understanding' of the content that it's read, and the voices are climbing out the other side of Uncanny Valley - but inevitably imperfect.
We don't believe in AI for the sake of AI, and we don't usually order the flavour of the month, but technology like NotebookLM is worth investigating. It's good to know what is, and what still isn't, possible. Here's your quick rundown on the Cloud Exit, brought to you by a reading, talking machine:
A few things to keep in mind while you're listening
This audio is a fairly good summary of the content we gave it, but as you'd expect mistakes and misrepresentations still sneak in. One example: As 37 Signals moved to AWS and back out again, the team that managed their infrastructure never changed size much. They had anticipated a drop in numbers when they moved to the cloud, but it never happened. NotebookLM got confused by that.
That's not the only reason to retain a healthy level of skepticism. If this is your first brush with NotebookLM's podcast presenters (only two American-accented voices are available for now), you'll notice that it's getting harder and harder to distinguish machines and people. Stay alert to that.
On the plus side, NotebookLM does a good job of citing sources as it repeats facts or points of view from the articles that this podcast is based on. That's a useful way for AI-generated content to prove its trustworthiness.
The sources
If you want to keep digging into the Cloud Exit, or you want to see what NotebookLM based its podcast on, here are the articles we used.
Cloud Exit trend:
AWS:
AWS finally agrees that customers can save money by leaving the cloud
AWS charges a new unavoidable IPv4 toll, calls it "modernisation"
Hardware:
Dedicated CPUs over the years: There's improvement, and then there's this
Benchmarks show how fast our High Performance Virtual Servers are
Thinking about your own exit?
As always, we're here to talk if you want to see how your hosting bills would change if you joined the Cloud Exit. Contact us for a realistic, no-obligation conversation about your options.
Image by Artie Blur from Pixabay