Once you certify out of a ULA, the unlimited right ends and every cloud instance must run inside the perpetual entitlement you declared. Cloud that did not count at exit still consumes that entitlement, so the time to plan cloud licensing is before certification, not after it.
By the Meridian advisory team, former Oracle LMS and GLAS licensing analysts. Updated 4 June 2026.
After certification you hold a fixed number of perpetual processor licenses and the unlimited deployment right is gone. From that point, every place the named products run, on premises and in cloud, draws on that fixed entitlement. Cloud instances in authorized environments convert to processor licenses under Oracle's cloud policy, commonly two vCPUs to one processor with hyper threading on, and that consumption has to fit within your certified count. As long as total deployment stays inside the number you certified, you are licensed. The moment cloud growth pushes total consumption past it, you are over deployed and the excess must be licensed with a deliberate purchase. The post exit world is therefore a capacity management problem: a fixed pool of entitlement against a deployment that can move.
You have a fixed pool of certified processors. Cloud and on premises both draw from it. Stay inside the pool and you are licensed. Grow past it and you buy more, deliberately, never by a panic return to a ULA.
This is where many organisations get caught. Cloud counting is contract specific, and a workload that did not count toward the baseline, because the contract excluded public cloud or the instance had not met a continuous run condition, does not vanish at exit. It still runs, and once the unlimited right ends it draws on your certified entitlement like any other deployment. If you certified without accounting for it, that cloud is now consuming processors you needed elsewhere, or it pushes you over the count. The way this is handled is the difference between a clean exit and a post certification scramble. We set out how cloud counts at the exit itself in counting vCPUs in authorized cloud environments.
The fix is always pre exit. Where cloud will not count toward the baseline but the workload is permanent, there are three deliberate moves. Repatriate the workload on premises before the certification date so it counts toward the baseline as on premises deployment. Move it to OCI, where Oracle's treatment can let it count, before exit. Or size the certified count to cover the cloud consumption you will continue to run, so the entitlement pool is large enough to hold it afterward. Each is a planned decision made while the unlimited right is still live, which is why cloud strategy belongs in the exit project, not the year after.
After certification, managing cloud is about keeping deployment inside the pool. That means tracking what runs in authorized cloud, converting vCPUs to processors consistently, and reconciling total consumption against the certified count on a regular cadence. Audit risk rises in the first two years after certification, and cloud is a frequent focus because it is easy to grow quietly. A workload scaled up in AWS or Azure without checking the entitlement pool can move an organisation into over deployment without anyone deciding to. The discipline is a periodic self review of cloud against entitlement, the same review that protects the rest of the estate.
Take an indicative case. An organisation certifies 1,200 processor licenses, of which on premises uses 1,000, leaving 200 of headroom. It also runs an authorized cloud estate converting to 150 processors that the contract excluded from the baseline, so those 150 were not part of the certified number. After exit, that cloud now draws on the 200 of headroom, leaving 50. A planned cloud expansion of 100 processors then pushes total consumption to 1,250, fifty over the certified count, requiring a deliberate purchase of 50 processor licenses. Had the cloud been repatriated or moved to OCI before exit so it counted, or had the certified number been sized to include it, the purchase would not arise. The figures are indicative, but they show how excluded cloud quietly erodes headroom after the exit.
| Move before exit | Effect on cloud workload | When it fits |
|---|---|---|
| Repatriate on premises | Counts toward the baseline | Workload can run on premises |
| Move to OCI | Can count under Oracle treatment | OCI suits the workload |
| Size the certified count | Entitlement pool covers it | Cloud stays where it is |
Which move is right depends on your contract, your cloud policy version, and the workload, so test each against your own position rather than assuming one answer.
Licensing cloud after a ULA is the reward, or the cost, of decisions made before the exit. Plan the cloud position while the unlimited right is live, certify a count that reflects the cloud you will keep running, and manage consumption against the pool afterward. The mistakes that shrink a cloud count, and the over deployment they cause later, are covered in the cloud counting mistakes that shrink your count. The full exit method lives in our pillar, the ULA exit strategy guide.
After a ULA, cloud runs inside a fixed entitlement pool, and cloud that did not count at exit still draws on it. Plan the cloud position before certification by repatriating, moving to OCI, or sizing the count to cover it, then manage consumption against the pool with a regular review. Growth beyond the count is licensed deliberately. The organisations that plan cloud into the exit avoid the post certification purchase that surprises the ones that did not.
After certification the unlimited right ends, so cloud instances must run inside the perpetual processor entitlement you certified. vCPUs in authorized cloud convert to processor licenses under the cloud policy, and that consumption must fit within your certified count. Growth beyond it has to be licensed with a deliberate purchase.
Cloud that did not count toward the baseline still consumes entitlement once the unlimited right ends. If the contract excluded it, that capacity now draws on your certified processors or needs new licenses. The fix is planned before exit: repatriate on premises or move to OCI so the workload counts, or size the certified number to cover it.
Book a confidential assessment and we will set your cloud position so it counts at exit and stays inside your entitlement afterward.