The Oracle Authorized Cloud Environment policy is the document that decides what an Oracle BYOL deployment costs on AWS, Azure and GCP. Unlike the partitioning policy, the cloud policy has been updated repeatedly — most recently in 2024 — and customers running Oracle BYOL on a contract signed before the current version often find themselves licensed under terms that no longer match Oracle's audit position. This is the current state of the policy, the 1:2 core factor, and where the audit findings actually come from.
Oracle's "Authorized Cloud Environment" (ACE) policy nominates Amazon EC2, Amazon RDS, Microsoft Azure and Google Cloud Platform as recognised public-cloud environments for Oracle Database, Middleware and Application licensing. Customers running Oracle products in those environments under BYOL terms apply the policy's counting rules. Customers running outside an ACE-recognised platform need to license against Oracle's standard processor metric without cloud-specific concessions.
For AWS and Azure, the policy specifies a 1:2 ratio — two vCPUs in the cloud equal one Oracle processor licence. For GCP, the same 1:2 ratio applies. For OCI, every two OCPUs counts as one processor (although OCPUs and vCPUs are not equivalent units, which is why OCI BYOL maths often surprises customers). The 1:2 factor was reduced from a more favourable earlier ratio in 2017 and has been the audit baseline since.
A r5.4xlarge instance on AWS provides 16 vCPUs. Under ACE, that requires 8 Oracle Database EE processor licences. At list, that is approximately $380,000 plus 22% annual support. The same Oracle workload on OCI E4 Standard 4 shapes uses 4 OCPUs, requiring 2 Oracle Database EE licences. The asymmetry between OCI and non-OCI cloud is intentional and is the basis for Oracle's preference for OCI in any commercial conversation.
The 1:2 ratio applied to the wrong shape size doubles the licence position quietly.
The 1:2 ratio applies to Database Enterprise Edition and most other Oracle products on AWS and Azure under standard ACE terms. Several exceptions matter:
Includes Oracle cloud BYOL benchmarks and ACE migration scenarios.
A common point of confusion is that the ACE policy counts vCPUs on the underlying instance — not on the workload running inside the VM. A workload that uses 4 cores out of a 16-vCPU EC2 instance is still licensed against 8 Oracle processors. The fence that resolves this is to size cloud instances precisely against the licensed Oracle workload. Customers running Oracle BYOL on oversized shapes routinely have licence positions 2-3x larger than they would on right-sized shapes.
For Azure, Dedicated Hosts can provide additional fencing — but Azure Dedicated Hosts are not automatically recognised as more favourable than shared infrastructure under ACE. The relevant fence is shape sizing, not host type. AWS Dedicated Hosts can sometimes attract favourable treatment but the policy is silent on this; relying on it requires contractual language, not a policy assumption.
Amazon RDS for Oracle is the most common Oracle deployment pattern on AWS. RDS supports BYOL and licence-included options. BYOL is favoured when the customer has under-utilised perpetual licences; licence-included is favoured for tactical or short-term deployments. The licence-included pricing includes Database Enterprise Edition only — Partitioning, ASO, Advanced Compression and other options must still be separately licensed even under licence-included, unless using RDS Oracle's specific "Enterprise Edition" offering with restrictions.
Oracle's commercial strategy on cloud has been to use OCI Universal Credits as the financial vehicle for BYOL-to-cloud migration. A customer with significant on-prem Oracle commitments can typically convert a portion of those commitments to OCI Universal Credits at terms more favourable than equivalent AWS or Azure BYOL. The relevant levers in those negotiations:
The trade-off is operational. OCI is improving rapidly but lacks the breadth and ecosystem of AWS, Azure or GCP. Customers running Oracle as a strategic platform often find OCI compelling. Customers running Oracle as a legacy workload to be exited rarely do.
The maths is workload-by-workload. We've run it 200+ times.
We benchmark BYOL positions on AWS, Azure and OCI — and run the workload-by-workload economics. It pairs with our cloud contract advisory so the BYOL position still holds when the renewal lands.
Weekly compliance intelligence for IT leaders.