Move the right workloads off the public cloud and keep the savings
Cloud repatriation isn’t a retreat. It’s a strategic move to regain control of costs, performance, and compliance by shifting the right workloads off hyperscalers and onto infrastructure purpose-built for sustained demand.
Talk to our team
37signals left the cloud and now saves millions a year
The makers of Basecamp and HEY were spending $300,000 a month on AWS and couldn’t get anyone on the phone. They moved to Summit. Here’s what happened.
The numbers speak for themselves
Measured after migrating seven applications from AWS to Summit colocation in Ashburn and Chicago.
cloud spend
over five years
With AWS, we give them $300,000 a month and they don’t care about us. They don’t answer our calls. But we call Summit and they’re on the phone with us immediately.
Colocation
Identical footprints in Summit’s Ashburn and Chicago data centers: (4) 48U cabinets at each site housing ~90 Dell R7625 servers, with 40kW of power split across four cabs per location.
Network connectivity
Redundancy at every layer. External connections to the internet, AWS, and between data center sites, each using different providers and diverse physical paths for full resilience.
Disaster recovery
Critical applications run across Ashburn (active) and Chicago (standby). Either site can absorb full traffic during a disaster event with minimal transition time.
Public cloud made sense until it didn’t
The economics of hyperscale hosting have shifted. Companies with predictable workloads are finding that the flexibility premium no longer pays for itself.
Unpredictable bills
Egress fees, data transfer charges, and tier pricing make month-over-month costs nearly impossible to forecast.
Vendor lock-in
Proprietary tooling and licensing changes (like Broadcom’s 72-core VMware minimum) trap you in cost structures you didn’t sign up for.
Performance ceiling
Noisy neighbors and virtualization overhead cap what your workloads can do, regardless of how much you spend.
Support you can’t reach
Hyperscalers don’t answer the phone unless you have an enterprise contract. Summit engineers pick up, 24/7.
The workloads that win when they leave the cloud
Not every workload belongs off the public cloud. These ones usually do.
A proven process, not a leap of faith
Before we move anything, we audit your workloads, architectures, dependencies, and performance patterns. The Summit Repatriation Framework ensures every migration is de-risked at every stage, with fit-to-purpose solutions and minimal business disruption.
37signals pulled seven applications out of the cloud in six months. We’ve run this same process for companies migrating from AWS, GCP, and Azure. The result: right-sized infrastructure and performance you can measure.
Start your assessmentWorkload audit
We analyze your current bills, architectures, peak loads, dependencies, and compliance requirements to identify which workloads are ready to move and which should stay.
Infrastructure design
We design a fit-to-purpose environment: private cloud, bare metal, colocation, or hybrid. Configurations are scoped to your actual workloads, not a template.
Incremental migration
We move workloads in stages. Each is validated before the next begins, which rules out big-bang cutovers and surprise downtime.
Ongoing optimization
Once live, we monitor performance, right-size resources, and surface opportunities to improve cost and throughput through the customer portal.
The modern data center is nothing like the one you left
Today’s private infrastructure matches the best of what public cloud offers, without vendor lock-in, noisy neighbors, or unpredictable pricing.
Private cloud
Dedicated infrastructure with cloud-like flexibility. API-driven provisioning, Infrastructure-as-Code support, and full environment control. Best for teams that need elasticity without the hyperscaler price tag.
Learn about private cloudBare-metal servers
Single-tenant dedicated hardware without virtualization overhead. Every core, every GB, every IOPS is yours. Best for AI, GPU workloads, database hosting, and compute-intensive applications.
Learn about bare metalColocation
You own the hardware, we provide space, power, cooling, and connectivity in Tier III-exceeding facilities with biometric security, 24/7 engineering teams, and U.S.-based Remote Hands.
Learn about colocationBring us your cloud bill and we’ll show you what Summit costs
There’s no commitment. Customers typically save 30 to 70% on infrastructure costs after repatriating to Summit.
Signs it’s time to have the repatriation conversation
You don’t have to be in crisis to benefit from repatriation. These patterns show up before the pain gets serious.
Your cloud bill climbs even when usage is flat
Egress line items you don’t recognize, tier changes you didn’t agree to. That’s the signal.
You got hit by Broadcom’s VMware licensing changes
The 72-core minimum means paying for cores you’ll never use. Summit supports Hyper-V and Proxmox as VMware alternatives if you want to exit the Broadcom ecosystem entirely.
Peak loads happen at predictable times
If your traffic patterns are stable, you’re paying for elasticity you don’t need. Dedicated infrastructure wins on price-to-performance for consistent demand.
Compliance audits keep flagging shared environments
HIPAA, SOC 2, and PCI DSS are easier to demonstrate on single-tenant infrastructure where you control the entire stack.
You’re scaling AI or GPU workloads
Cloud GPU instance pricing escalates fast at scale. Dedicated GPU nodes on bare metal deliver the same throughput for a fraction of the per-hour cost.
Support is gated behind enterprise contracts
If your hyperscaler won’t take your calls without an upgrade, that’s not a support model. Summit’s 24/7 support is included, not upsold.
Companies that made the move
We pulled seven applications out in six months. You can do that too. The tools are there. They’re free. So don’t stay in the cloud because of the hype.
Everything you need to think through repatriation
AI recruiting platform lowers costs and improves performance via cloud repatriation
A cloud-native company repatriated from AWS and cut monthly costs by 30% while gaining 50% more processing capacity on the same budget.
Why companies are leaving the public cloud
Cost myths, technical challenges, migration best practices, and real results. A clear guide to doing repatriation right.
BlogThe data center you left isn’t the one you’re returning to
Think cloud repatriation means going backward? Modern data centers match the best of public cloud, without the lock-in.
BlogOnly fools rush in: how incremental cloud repatriation ensures optimal performance
Before you race to repatriate, consider the bigger picture of your infrastructure, business goals, and ideal future state.
BlogCloud repatriation is back thanks to Broadcom’s 72-core requirement
Broadcom’s VMware licensing overhaul is forcing companies to rethink their cloud strategy. Here’s how to sidestep the core tax.
BlogLosing sleep over public cloud bills? Time to consider repatriation
Outrageous bills aren’t the only reason to rethink your hosting infrastructure. A hybrid, managed setup can future-proof your organization.
BlogCloud repatriation isn’t a retreat, but a reconsideration
What’s needed isn’t a retreat from the cloud, but a clear-eyed reassessment of how best to use it for each workload.
Questions we hear before every migration
Cloud repatriation is the process of moving workloads from a public cloud provider (like AWS, GCP, or Azure) to private infrastructure: managed private cloud, bare metal servers, or colocation. It’s not about abandoning cloud entirely. It’s about putting the right workloads in the right environment for cost, performance, and control.
Disruption is the most common concern we hear. That’s why we use an incremental approach: moving workloads in stages and validating each one before proceeding. 37signals pulled seven applications out of the cloud in six months using this method, with no material downtime to Basecamp or HEY customers.
No. Most of our customers end up in a hybrid model: some workloads on Summit infrastructure, others remaining on public cloud where it makes sense. The audit phase of our process identifies which workloads benefit from the move and which should stay put.
Bring us your cloud bill and we’ll show you. Customers typically see savings of 30 to 70% depending on workload type and current pricing. 37signals cut cloud spend by 60% and projects $10M in savings over five years. An AI recruiting platform we work with reduced monthly infrastructure costs by 30% while gaining 50% more processing capacity on the same budget.
Yes. Summit offers private VMware cloud environments, so if you want to stay on VMware but leave the public cloud, we can host that. We also support Hyper-V and Proxmox as VMware alternatives if you’d rather exit the Broadcom ecosystem entirely and avoid the 72-core minimum penalty.
Summit holds a SOC 2 Type II (AT-101) audit and supports customers with HIPAA, PCI DSS, and other compliance requirements through documented controls, BAAs, and shared responsibility frameworks. Single-tenant infrastructure makes it significantly easier to demonstrate compliance than shared cloud environments.
Ready to see what your workloads would cost off the cloud?
Bring your current cloud bill. We’ll come back with a cost comparison and infrastructure recommendation within one business day.