The Fundamental Licensing Difference Between OCI and AWS

When enterprises lift Oracle Database workloads to the cloud, the choice of cloud platform has a direct and material impact on how many Oracle licences are required. This is not a minor technical nuance โ€” for large deployments it is a multimillion-dollar difference, and it is one that Oracle's sales team rarely volunteers when discussing cloud migration options.

The core difference comes down to how Oracle counts processor licences on each platform. On Oracle Cloud Infrastructure, one Oracle Processor Licence maps to one OCPU, and one OCPU equals one physical core. On AWS, Oracle's BYOL policy counts two vCPUs as equivalent to one physical core, meaning one Processor Licence covers two vCPUs โ€” but AWS instances are hyperthreaded, so the effective licence requirement for the same workload capacity is systematically higher. An 8 vCPU AWS EC2 instance requires 4 Oracle Processor Licences. The equivalent compute capacity on OCI requires only 2 Processor Licences. The Oracle Knowledge Hub covers this calculation in detail, and it is one of the most consequential factors in cloud platform selection for Oracle workloads.

At Oracle Database Enterprise Edition list pricing of $47,500 per processor, the difference for a 100-processor equivalent deployment is $2.375 million in incremental licence cost on AWS versus OCI โ€” before support fees are applied at 22% annually, adding a further $522,500 per year. Over a three-year cloud contract, that gap compounds significantly.

OCI vs AWS Oracle Licensing Analysis

Redress Compliance provides independent analysis of Oracle licensing economics on OCI versus AWS, including BYOL counting, Support Rewards modelling, and total cost of ownership comparison.

Speak to an Expert

BYOL Rules: How Each Platform Counts Licences

Understanding Oracle's BYOL rules on each platform requires reading Oracle's cloud licensing policy document alongside each provider's instance type specifications. The rules are clear in principle but easy to misapply in practice.

On OCI, Oracle defines an OCPU as one physical core with hyperthreading enabled. For BYOL purposes, one Processor Licence covers one OCPU. If you deploy a 4 OCPU VM shape on OCI, you need 4 Processor Licences. This is a direct 1:1 mapping between the compute unit Oracle sells in OCI and the licence unit you bring. The simplicity of the OCI model is intentional โ€” Oracle designed OCI's compute primitives to align with its licence counting rules.

On AWS, Oracle's guidance states that 2 vCPUs count as 1 physical core, and therefore 1 Processor Licence. However, AWS EC2 instances use Intel or AMD processors with hyperthreading, presenting vCPU counts that reflect hyperthreads rather than physical cores. A c5.2xlarge has 8 vCPUs, which Oracle counts as 4 physical cores requiring 4 Processor Licences. For Bring Your Own Licence deployments on AWS RDS, Oracle manages the licence counting automatically โ€” but for EC2-based deployments, organisations must count correctly themselves. The Oracle licensing on AWS guide explains the vCPU-to-processor mapping for common EC2 instance families.

One important nuance on AWS: Oracle has confirmed that the 2 vCPU = 1 physical core rule applies to hyperthreaded instances. If you deploy Oracle on a dedicated host with hyperthreading disabled, you can count actual physical cores โ€” but this requires explicit configuration and increases the instance cost. Most organisations run hyperthreaded by default, making the 2:1 vCPU-to-processor ratio the standard assumption.

For Oracle Standard Edition 2, the rules differ: SE2 requires licensing by socket, and SE2 cannot be deployed on instances with more than 2 sockets. This effectively limits SE2 BYOL to smaller instance types and makes SE2 unsuitable for large-scale cloud deployments where EE is genuinely required.

Oracle Support Rewards: The OCI Advantage AWS Cannot Match

The licence counting difference alone would be a compelling case for OCI in Oracle-heavy environments. But there is a second, equally significant financial dimension: Oracle Support Rewards. This programme allows organisations running OCI Universal Credits to offset a percentage of their Oracle technology support bill directly.

Standard Oracle technology customers earn 25% of their annual OCI Universal Credits spend as Support Rewards, credited against their Oracle support invoice. ULA customers earn 33%. The practical effect is substantial: an organisation spending $1 million per year on OCI Universal Credits reduces its Oracle support bill by $250,000 to $330,000 annually. Over a three-year contract, that is $750,000 to $990,000 in support savings โ€” savings that have no equivalent on AWS. AWS offers no mechanism to reduce Oracle support costs, regardless of how much is spent on AWS services.

The Oracle Support Rewards guide explains the programme mechanics in detail, including how rewards accumulate, how they are applied against support invoices, and the contractual conditions that can affect eligibility. For organisations with large Oracle support bills โ€” particularly those paying Oracle 22% annually on a significant on-premises database estate โ€” the support rewards programme can make OCI migration economically compelling even before compute costs are compared.

There is a critical caveat, however. Oracle Support Rewards create a dependency on continued OCI spending to maintain the benefit. If OCI spend decreases or stops, the support cost reduction disappears. This creates a lock-in dynamic that is worth modelling explicitly when evaluating OCI versus AWS from a long-term commercial perspective. The Oracle cloud contract renewals guide covers how to structure OCI agreements to preserve commercial flexibility.

Oracle Cloud Licensing Assessment

Model the Oracle licensing economics on OCI versus AWS for your specific workload profile.

Run the Assessment โ†’

Total Cost of Ownership: A Worked Example

Combining the licence count difference and the Support Rewards benefit, the TCO gap between OCI and AWS for Oracle workloads is consistently material. Consider an enterprise running Oracle Database Enterprise Edition with Real Application Clusters on a workload that requires 20 processor licences worth of compute.

On OCI, 20 OCPUs of compute requires 20 Processor Licences brought from on-premises BYOL. Annual OCI Universal Credits spend for that compute at OCI's database pricing might be approximately $400,000 per year. Support Rewards at 25% on that spend generate $100,000 per year in support cost reduction. The effective annual cloud cost โ€” compute plus the net support reduction โ€” is therefore $300,000 in incremental OCI spend, offset by $100,000 in support savings.

On AWS, the same workload at 40 vCPUs (requiring 20 Processor Licences under the 2:1 rule) on comparable EC2 compute might cost $350,000โ€“$400,000 in compute annually. There is no support offset. The base compute cost is similar, but the all-in cost is higher because AWS provides no mechanism to reduce the Oracle support bill โ€” which continues at Oracle's standard 22% rate, increasing by 8% per year.

For organisations with existing Oracle on-premises workloads and a large Oracle support bill, the Support Rewards programme can shift the OCI TCO calculation decisively. Organisations that have modelled both options carefully โ€” including licence count, compute cost, and support rewards โ€” have consistently found OCI 30-40% cheaper for Oracle-specific workloads over a three-year horizon.

That said, AWS is a larger, more mature cloud platform with stronger ecosystem services, more geographic regions, and broader tooling for non-Oracle workloads. Organisations running mixed workloads may find a hybrid strategy โ€” Oracle databases on OCI, non-Oracle applications on AWS or Azure โ€” delivers the best commercial outcome. Redress Compliance's Oracle licensing advisory service provides the independent modelling to make this decision with confidence.

Audit Risk in Cloud Deployments

One dimension of OCI vs AWS that receives less attention is audit risk. Oracle's ability to audit BYOL cloud deployments is contractually preserved in both OCI and AWS agreements. However, the mechanisms differ in important ways.

On OCI, Oracle has direct visibility into compute consumption because OCI is Oracle's own infrastructure. While Oracle does not audit OCI customers in the same confrontational way as on-premises LMS audits, the data is available and licence compliance is a condition of the BYOL terms. Non-compliance on OCI can result in Oracle requiring additional licence purchases or restricting access to services.

On AWS, Oracle audits BYOL compliance through its standard LMS process, requesting deployment data from the customer. AWS does not provide Oracle with direct visibility into EC2 usage โ€” that is the customer's responsibility to disclose accurately. Organisations that have miscounted vCPUs or failed to apply Oracle's processor counting rules correctly face the same retroactive exposure on AWS as on-premises, including 8% annual support increases on any gap identified. The Oracle audit risk assessment covers cloud deployment scenarios explicitly.

Written by Fredrik Filipsson Co-Founder, Redress Compliance โ€” 20+ years enterprise software licensing