Oracle and AWS Expand Partnership as Oracle AI Database@AWS Scales to 22 Regions

Oracle and AWS have expanded their long-term strategic collaboration agreement and announced the general availability of Oracle Exadata Database Service on Exascale Infrastructure — bringing Exadata-class performance to Oracle AI Database workloads of any scale across 22 AWS Regions, one year after the service first reached general availability.

Exadata Economics at Any Scale — What Exascale Changes

The structural limitation of traditional Exadata deployments has always been cost: dedicated database and storage server configurations made the economics work only for large enterprise workloads. Oracle Exadata Database Service on Exascale Infrastructure changes that model. Instead of provisioning dedicated hardware per workload, every database spreads across pooled storage servers — customers specify the compute and storage capacity they need, and the infrastructure handles distribution for high performance and availability automatically. The result is that Exadata performance and economics become practical for mid-market workloads, multi-region high availability deployments and disaster recovery configurations that could not previously justify the cost.

The GA release also adds instant thin cloning: database copies for development and testing spin up with full Exadata performance without duplicating storage. For engineering teams working on large Oracle schemas, that alone reduces the infrastructure overhead of maintaining separate dev and staging environments substantially. As Oracle's technical documentation notes, the same Exascale architecture now underpins the AI workload layer — meaning the performance gains extend directly to vector search and agentic AI applications running on live business data.

AI Built Into the Database Layer

Oracle AI Database@AWS now encompasses two generally available managed paths: the new Exadata-powered service and Oracle Autonomous AI Database Serverless, which lets customers build and run fully managed databases without provisioning or managing any infrastructure. Sub-200 microsecond application-to-database latency — as low as 165 microseconds using AWS High Performance Networking with EC2 placement groups — means latency-sensitive OLTP systems, ERP applications and AI-enabled operational workloads can run adjacent to the database without performance compromise.

The AI integration layer is delivered through Oracle AI Vector Search, which enables retrieval augmented generation directly from SQL without moving data out of the database. Amazon RDS for Oracle now supports Oracle Database 26ai with Amazon Bedrock integration — giving developers access to foundation models including Anthropic Claude, Amazon Nova and Meta Llama, plus Oracle's Select AI feature for generating and running SQL queries from natural language prompts. The AWS agent plugin ecosystem and the broader Anthropic Claude integration via Bedrock mean that enterprises can now run agentic AI workflows directly on their Oracle data without the data movement and latency costs of separate AI infrastructure.

Enterprise Customers Already Running Business-Critical Workloads

Three anchor customers illustrate the range of industries already committed to Oracle AI Database@AWS. CJ Olive Young — South Korea's leading beauty and wellness retailer operating over 1,380 stores — is migrating core workloads to the cloud using the platform, integrating live retail data with AWS analytics and AI services. Kobalt Music Group, a technology-first music rights management company managing one of the largest independent catalogues globally, has moved business-critical data workloads to the service. The Metropolitan Transport Authority of Barcelona is using Oracle AI Database@AWS to put critical mobility data and advanced analytics to work on public transportation optimisation.

The Barcelona MTA case is notable specifically because transportation authorities represent some of the most conservative infrastructure buyers in any market — highly regulated, politically accountable and operationally intolerant of downtime. Their commitment to Oracle AI Database@AWS is the kind of reference that enterprise sales teams use to unlock financial services and government workloads. The deal mirrors the pattern seen in the Ryanair–Google Cloud partnership, where large incumbents in regulated, operationally critical sectors are making multi-year cloud AI commitments rather than pilot deployments. The Goldman Sachs–Nvidia infrastructure financing context shows just how much capital is now flowing toward exactly this layer of enterprise AI infrastructure.

Global Partners and the Migration Acceleration Push

The expanded strategic collaboration agreement includes a formal commitment from both Oracle and AWS to accelerate customer migration and adoption. Oracle Migration Accelerator funds migration services, allowing enterprises to offset migration costs while partners scale larger transformation projects. Eligible members of both the AWS Partner Network and Oracle Partner Network can resell Oracle AI Database@AWS through AWS Channel Partner Private Offers, and qualified managed service providers can operate the platform on a customer's behalf.

Deloitte, operating across both Oracle and AWS practices, is leading enterprise engagements that combine industry domain knowledge with Oracle database migration and AWS AI services. Hitachi, drawing on more than 30 years of Oracle experience, has validated Oracle AI Database@AWS against the requirements of mission-critical financial services systems as part of its Lumada modernisation initiative. The combination of an expanded long-term commitment from both hyperscaler-scale vendors, GA availability of Exadata economics at any scale, and a funded migration path from a 22-region global service creates exactly the conditions for large-scale enterprise database migration to accelerate significantly through 2027. For CIOs currently running Oracle workloads on on-premises infrastructure, the economics and risk profile of staying off-cloud are getting harder to defend — particularly as independent AI evaluation shows that the AI capability gap between cloud-connected and isolated enterprise data is widening every quarter.