Two announcements in one week show Oracle is betting on deeper multicloud distribution for its database business and vertical AI agents that execute work inside operational systems. Both carry practical implications for ERP teams well beyond the press release headlines.
The AWS Deal Graduates From Experiment to Commitment
On August 13, Oracle and AWS announced an expanded long-term strategic collaboration agreement. The announcement was made alongside the general availability of Oracle Exadata Database Service on Exascale Infrastructure on Oracle AI Database@AWS. One year after general availability, the service now spans 22 AWS Regions, with customers including CJ Olive Young, Kobalt Music Group, and the Metropolitan Transport Authority of Barcelona running business-critical workloads on it.
The detail that matters most for ERP teams sits in the capabilities list. Oracle now cites application-to-database round-trip latency as low as 165 microseconds using AWS High Performance Networking, and names ERP applications explicitly among the latency-sensitive workloads this supports. Add zero-ETL integration with Amazon Redshift, Oracle-managed backups to Amazon S3, and the new Exascale option that removes the need to provision dedicated database and storage servers, and the message is clear. Oracle wants ERP-class workloads running on its database inside AWS, not just archived there.
The economics are being pushed just as hard as the engineering. Exascale brings pay-per-use Exadata pricing to smaller databases, and the Oracle Migration Accelerator helps fund migration services, with partners such as Deloitte and Hitachi building practices around the offering. The subtext is competitive urgency. As ERP Today reported when AWS entered the AI factory race, hyperscalers are now fighting to be the venue where enterprise AI runs next to enterprise data. Oracle’s answer is to make its database a first-class citizen wherever that fight lands.
Oracle Health Shows What Production AI Agents Look Like
Six days later, on August 19, Oracle Health expanded its Clinical AI Agent with automated professional fee coding, clinician-controlled dictation, and AI-assisted chart review, all now available in the US. Oracle says note generation alone has saved physicians more than 400,000 hours in the nearly two years since launch.
ERP Today readers should not file this under healthcare news. The coding capability analyzes patient visit conversations and suggests professional fee charge codes within the orders workflow, with clinicians reviewing before submission. That is charge capture, a revenue cycle process, being automated at the point of work with a human approval gate. It is the same pattern Oracle is building into its business applications through Fusion Agentic Applications and the AI Agent Studio builder, where agents act on business objects inside existing security and audit controls rather than around them.
The Pattern Across August
The rest of Oracle’s August reinforces the strategy. The AI-powered Oracle Health Patient Portal reached general availability on August 12, keeping patient data inside the EHR environment while adding conversational access. On August 11, Oracle and Quantinuum announced a multi-year partnership to deploy the Helios quantum computer inside an OCI AI data center, with a quantum service preview planned in the coming months.
Taken together: distribute the database everywhere, embed agents in the workflow, and keep governance inside the system of record. For customers, the friction point is absorption. As ERP Today has covered, NetSuite Next is rolling out its AI-first experience, and the Cloud ERP 26C release is landing 137 Financials changes this quarter, 109 of which require customer action. Innovation velocity is now Oracle’s product. Managing it is the customer’s job.
What This Means for ERP Insiders
Multicloud database strategy is now a costed option, not a compromise. With Exascale economics, 165-microsecond latency, and funded migration offers, teams running Oracle databases under ERP workloads on AWS should re-price their architecture rather than assume OCI is the only performant path.
Judge AI agents by their approval gates, not their demos. Oracle Health’s coding agent works because qualified professionals confirm every recommendation. ERP leaders evaluating Fusion or NetSuite agents should demand the same review, audit, and rollback controls before automating revenue-relevant processes.
Release absorption is the hidden cost of the AI roadmap. Quarterly updates now carry AI capabilities, compliance changes, and UI shifts simultaneously. Organizations without a standing release-readiness discipline will struggle to consume what Oracle is shipping, whatever cloud it runs in.
