Since 2017, we ended up spending almost $22m on cloud. Peaking at $3.7m in 2021. Now that we're all out, we'll spend a little over a million on everything we own ourselves. Same team size. Better performance. Saving $2m+ per year! 🎉

Oct 21, 2025 · 9:51 PM UTC

Replying to @dhh
now, build your own DIY storage cluster and save even more! :D
4
1
197
Replying to @dhh
I remember you announcing that you would be leaving the cloud, and I can’t believe it was 3 years ago. Time flies!
1
15
Replying to @dhh
Can you share how you do it and what you use. I mean not the details, but are you using your own servers in redundant datacenters? How are they connected? Where do you store the backups? Would be interested in how you do it, if you can and are willing to share this.
1
11
Replying to @dhh
What do you use to orchestrate VPS instances, on demand storage and allocating serverless compute? OpenStack looks doable but seems complicated to rollout and maintain
2
10
Replying to @dhh
Is it ok to ask, what app are you running with so much revenue to justify these hosting costs? Your apps have such a huge usage and so many paying customers? This is not AI infrastructure, just storage and some CPUs? I tried to Google what you sell but couldn’t figure out..
3
6
Replying to @dhh
What solution do you use to host the hardware? I remember listening to your great talk with Lex that you don’t host the hardware itself on your own premises, if I’m not mistaken?
1
3
Replying to @dhh
How do I handle regional distribution your selfs
3
2
We use DCs in the mid-west, US east, and Europe. Distributing between these regions is similar to what it was in the cloud. You need geo load balancing, database replication, etc.
2
21
Replying to @dhh
What are you doing with the extra money saved?
1
1
Keep it!
4
76
Replying to @dhh
What did you replace S3 with? And how do you do CDN? I’m sure you mentioned it but that’s been a hold up for me making the switch to Kamal.
We run our own Pure Storage that has S3 compatible API. Then we use Cloudflare for CDN.
1
6
Replying to @dhh
What’s the main services that you still rely on AWS?
Today? Zero. But up until this summer, we were still stuck with a 4-year commit to S3.
I swear we didn't plan this, but just today, we finally NUKED our entire AWS account at the 37signals company meetup! This followed moving out compute+dbs in 2023, then S3 this summer, and now finally the entire account is GONE 🎉
1
1
23
Replying to @dhh
Excellent. I even bet your platform or infrastructure team is probably the same size or less. Also, how was your up time looking in comparison? I’m assuming better.
Same team. Here's our uptime.
1
15
Replying to @dhh
You rock man. Just rewatched your lex video yesterday and posted about it. Ironically AWS went down the day before. You called it!
1
Replying to @dhh
If you didn't exit your S3 commitment, what would the price increase to?
1
Replying to @dhh
These totals are wild 😳
Replying to @dhh
Does your hardware provide the same level of security as Nitro? Besides that I’ve been the one to personal go reboot the servers in the middle of the night. And you have to manage people maintaining secure systems in multiple regions. I’d rather focus on my application code. But if you want to manage all that might save some money. No SCPs. No global policy infrastructure. No secrets manager, KMS, GuardDuty. Undoubtably less secure. Having been there done that, no thanks. docs.aws.amazon.com/whitepap…
Replying to @dhh
better performance and better maintainability plus you don’t have to worry about cloud providers randomly deprecating services betting on your team is the way