One monthly figure.
No egress surprises, no per-request maths.
The crossover
Public cloud is excellent, until the bill is mostly idle capacity.
Past a certain steady-state utilisation, dedicated capacity is both cheaper and faster. The trick is knowing where your line sits.
Single tenant.
Your workload is not sharing a host with anyone.
In-country by default.
Residency you can point a regulator at.
We will tell you not to.
Below the crossover, stay on public cloud.
The sizing exercise finds your crossover on real usage data, before anything moves.
The tenancy
What you get in a tenancy.
Every layer configured together, not bolted on afterwards.
Compute
CPU nodes sized to your steady load, with headroom agreed rather than guessed.
Storage
Block and object tiers, snapshotted daily, with offsite copies included.
Networking
Private links to your sites and clouds, segmented so a breach does not spread.
Security
Hardening, access control, encryption at rest and in transit, monitoring and a tested recovery plan. Not a separate line item.
Security & Data →Fit
When it is the right answer.
And when it is not — the second list matters more.
A good fit
- Inference runs most hours of most days
- Data residency or single tenancy is required
- Finance needs a fixed monthly number
- Egress charges have become a real line
Not yet
- You are still validating whether the product works
- Load is spiky and idle most of the week
- You depend on managed services we would have to rebuild
- Nobody has measured current utilisation yet
Engagement
How a move actually runs.
-
Sizing
Real utilisation, measured — not a guess from a capacity spreadsheet.
-
Provision
Capacity reserved for you, in the region named in the contract.
-
Migrate
Workload by workload, with a rollback at each step.
-
Run
Managed under SLA, with your team keeping admin access to their own workloads.
Questions
The ones that decide whether this is worth a sizing.
Where is the hardware?
In the datacentre region you nominate, with the specific facility named in your contract. Capacity can be placed in other regions where you need it, but the default and the common case is in-country.
Is this just a rebadged public cloud?
No. It is dedicated hardware allocated to a single customer, managed by us. You are not sharing a hypervisor, and capacity is reserved rather than burst from a shared pool.
How is it priced?
A monthly figure based on the capacity reserved, quoted after the sizing exercise. No per-request or per-GB-egress charges. If usage grows past the reservation we agree the increment before it applies, not after.
What if we want to leave?
Everything runs on standard tooling — containers, standard storage protocols, infrastructure as code. Migration out is an exercise, not a rescue. We will help with it, and there is no exit fee.
Who manages it day to day?
We do: patching, monitoring, backup, capacity and incident response, under the SLA in your contract. Your team keeps admin access to their own workloads.
Can we run some things here and some on public cloud?
Yes, and that is common. Steady inference and regulated data here; spiky batch and managed services on the hyperscaler. We build the private links between them.
Get a sizing on your real usage.
Send utilisation, not a guess. We will tell you which side of the crossover you sit on.