Coming soon
Dedicated servers on our own network.
Not a control panel and a support queue. Bare metal, storage and GPU nodes behind the anycast edge we built and operate, with the option of the engineers attached.
Planned line-up
Three families, specified from real workloads.
Final configurations and pricing are not published yet — deliberately. We would rather build what the waitlist asks for than guess at a catalogue.
Compute
General-purpose bare metal for application, database and build workloads.
- Recent-generation AMD and Intel platforms
- NVMe local storage, optional Ceph-backed volumes
- Private L2 or routed interconnect between your nodes
- IPv4 and IPv6 from our own allocation
Storage
High-capacity nodes for object storage, backups and data retention.
- Dense spinning and hybrid configurations
- S3-compatible endpoint via Ceph RADOS Gateway
- Cross-site replication within our footprint
- Egress priced to be used, not to trap you
GPU
Inference and training nodes, sized from your workload rather than a catalogue.
- Single and multi-GPU configurations
- vLLM, Triton or your own serving stack preinstalled
- MIG partitioning where the hardware supports it
- Private model storage with no egress to third parties
Included
The network is the product.
Anyone can rack a server. The difference is what it sits behind.
- Announced from address space we own, so your reputation is yours
- Anycast edge in front, if you want it
- Out-of-band access and remote hands at every site
- Monitoring and alerting included, not an upsell
- Managed option with the full infrastructure practice attached
- No overselling — published capacity is real capacity
anycast topology
203.0.113.0/24 · AS<ours>
Waitlist
Tell us what you would actually buy.
Specification input is more useful to us than an email address. The more detail you give, the more likely the first batch fits your workload.
Include the workload, rough node count, regions you need, and what you pay for equivalent capacity today. That is the input that shapes the first batch.
Questions
Reasonable things to ask before a launch.
When does this launch?
We are building the capacity and the provisioning layer now. The waitlist exists to shape the specification — the configurations we build first will be the ones people on the list actually asked for.
How is this different from any other dedicated server provider?
Two things. It runs on address space and a network we own and operate, rather than rented IP space on someone else’s transit. And the infrastructure consulting practice is attached — you can buy a server, or a server with the engineers who would design what runs on it.
Will there be a control panel?
Yes, and an API and a Terraform provider. Anything you can do in the panel you will be able to do in code, because an infrastructure product that requires clicking is not a serious one.
Can we bring our own IP space?
Yes. If you have your own ASN and prefix we will announce it for you with the appropriate IRR and RPKI records in place. If you do not and you need one, that is a conversation we have regularly.
What about existing consulting clients?
Existing engagements get first access. If we already run your infrastructure, moving part of it onto our own capacity is a conversation worth having early.