OCI compute is billed by shape, and the flexible shapes are the reason rightsizing is easier on OCI than on the hyperscalers. A flexible shape lets you set the exact number of OCPUs and the exact gigabytes of memory per instance, so you size to the workload rather than to the nearest fixed instance type. That removes the rounding tax that inflates compute bills elsewhere, where moving down means jumping to a smaller family. On top of that, the Ampere A1 Flex shapes run on Arm cores priced materially below x86, so workloads that run on Arm get a second, structural saving. Read utilization, dial the shape down, and move what you can to Arm, and OCI compute cost falls without touching throughput.
This is how OCI shapes are priced, why flexible shapes change the rightsizing math, when Ampere pays, and how to find the candidates.
What is an OCI shape and how is it billed?
A shape defines the resources allocated to a compute instance. The key unit is the OCPU, equal to one physical core. On x86 shapes one OCPU gives two vCPUs through multithreading; on Ampere Arm one OCPU is one physical Arm core. Flexible shapes, the E series on AMD, the X series on Intel, and the Ampere A1 Flex on Arm, bill per OCPU per hour plus per gigabyte of memory per hour, so the bill is a direct function of the cores and memory you allocate. Fixed shapes such as bare metal bill for the whole machine. The flexible families are where rightsizing lives, because they are the ones where you control the dials.
Why do flexible shapes change rightsizing?
On a hyperscaler, sizing down usually means moving to a smaller instance type, which changes both cores and memory in fixed steps and often forces a tradeoff you did not want. On an OCI flexible shape you set OCPUs and memory independently within a supported ratio band, so an instance that idles at low CPU but holds a lot of memory can drop cores without losing memory, and vice versa. That precision is the lever: instead of paying for a rounded up bundle, you pay for the cores and memory the workload actually uses. The same property makes scaling up a smaller decision, which encourages teams to start small and grow into demand rather than provisioning headroom they never use.
When do Ampere Arm shapes pay?
Ampere A1 Flex shapes are priced well below comparable x86 shapes per OCPU, and OCI also includes a generous Always Free allocation of Ampere capacity. For web and api tiers, microservices, JVM applications, build agents, and many container workloads, Arm performance is competitive and the price per core is lower, so the move is a structural saving rather than a one time cleanup. The cost is engineering: your images and dependencies must be built and tested for Arm. For modern, containerised stacks this is often a small change; for older binaries with native dependencies it can be more work. Treat Arm migration as a standing program for the workloads that qualify, not a blanket switch.
| Shape family | Architecture | Where it fits |
|---|---|---|
| Ampere A1 Flex | Arm (Ampere) | Lowest price per core; web tiers, microservices, containers, JVM |
| E series Flex | x86 (AMD) | General purpose x86 workloads at a strong price point |
| X series Flex | x86 (Intel) | Workloads that need Intel specific features or licensing |
| Bare metal | Whole server | Licensing by physical core, isolation, or maximum performance |
How do you find the rightsizing candidates?
Start in OCI Monitoring with CPU and memory utilization over a window that captures real peaks, not a quiet afternoon. Cross check against the usage reports so you are reasoning about billed capacity, not just live metrics. The native OCI advisors will recommend changes, but they recommend and do not decide: a workload that looks idle may carry a periodic batch peak that matters, and optimization that breaks that peak is not optimization. The discipline is to size to the real demand envelope, validate against the workload owner, and apply the change through the flexible shape so the saving is exact. Where the workload runs on Arm, fold the Ampere move into the same pass.
A Fortune 500 retailer ran a fleet of x86 flexible instances provisioned at fixed OCPU counts copied from an old hyperscaler sizing standard, most of them idling well below half their cores. We reduced OCPUs and memory to match measured demand using the flexible shape, then migrated the stateless web and api tiers to Ampere A1 Flex after validating the Arm builds. The combination cut compute cost while holding latency at peak, and the freed headroom removed the need for a planned capacity purchase. Figures are verified against billing data and anonymized.
Where to go next
Go deeper on the instrument itself in flexible shapes, OCI's quiet advantage, and see how the same precision applies to the database tier in autonomous ECPU sizing and auto scaling. The full OCI lever set is the OCI cost optimization guide, and the cross cloud view of rightsizing sits in the cross cloud cost optimization guide.
Frequently asked questions
What is an OCPU in OCI?
Are OCI Ampere shapes cheaper?
How do you rightsize OCI compute?
Take the next step on your cloud spend
We are an independent buyer side advisory that cuts public cloud spend across AWS, Azure, GCP, and OCI. We hold over $2.4B in annual cloud spend under management, take zero provider commissions, and deliver a 31 percent median reduction in the first 90 days. Our guarantee is simple: we reduce your cloud spend or we reimburse our service fee. Pricing is either a Fixed Fee scoped up front or Gainshare, a share of verified savings with no retainer and no risk to you.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.