Compute Economics are Changing
As infrastructure costs evolve, memory efficiency, storage performance, and network performance now play a central role in overall workload price-performance. The cost curve now favors applications and workloads that use memory more efficiently. At the same time, memory and storage now account for more of total workload cost and value, while network performance has become a key workload differentiator.
These shifts are raising the bar for cloud infrastructure. Customers want predictable workload-level price-performance and transparent pricing they can trust. They also want platforms engineered to improve real workload outcomes.
To meet these changing customer expectations, Oracle Cloud Infrastructure is introducing the next generation of OCI Compute Shapes, built on Oracle Acceleron. The new Acceleron family of compute shapes expands customer choice across AMD, Intel, and Ampere, with shapes designed for different workload needs.
This is more than a processor refresh. It reflects how OCI is evolving compute design to better serve modern workloads.
The new Acceleron family of Compute Shapes includes:
• E6 Standard Acceleron
• E6 DenseIO Acceleron
• X12 Standard Acceleron
• A4 Standard Acceleron
Built on Oracle Acceleron
At Oracle AI World 2025, we introduced Oracle Acceleron, OCI’s suite of network software and architecture, designed to help customers run workloads faster and more cost-effectively. Oracle Acceleron brings together new capabilities across fabric network architecture, fabric accelerators, and host network accelerators.
With the new Oracle Acceleron family, OCI is introducing the first host network accelerator capability from Oracle Acceleron into OCI Compute: Oracle Acceleron SmartNIC. In the new Oracle Acceleron shapes, Oracle Acceleron SmartNIC delivers two foundational benefits:
• Higher throughput with enhanced security properties: Helps increase throughput per core and per instance while lowering costs, with strong isolation, line-rate encryption, and no downtime network patching.
• Faster, higher-performance storage access: Enables NVMe-based storage acceleration, delivering 3X higher throughput per bare metal instance than prior approaches, without the cost and performance penalties of dual-NIC designs.
Designed for modern workload efficiency
Customers have long used OCI Flex shapes to tune infrastructure to their workload needs, and that flexibility remains important. But shape design still matters—especially for bare metal customers and for customers who want configurations that better reflect real workload patterns.
The new Oracle Acceleron shapes use core-to-memory ratios that better match actual workload needs, helping customers improve workload fit, use memory more efficiently, and avoid overprovisioning memory as it becomes a more important part of workload economics.
Performance highlights across the Oracle Acceleron family of Compute Shapes
The new Oracle Acceleron family also delivers meaningful performance gains across a range of workload types.
E6 Standard Acceleron builds on E6 Standard, delivering 16% higher per-core CPU performance while aligning memory-to-core ratio to customer workload needs with 8 GB per core.
E6 DenseIO Acceleron is the next generation of E5 DenseIO, with a CPU upgrade from AMD EPYC Gen4 to Gen5 for storage-optimized and data-heavy workloads. It delivers up to 13% higher base frequency, 33% faster memory, and 2x network bandwidth per bare metal instance.
X12 Standard Acceleron is the next generation of X9 Standard, with a CPU upgrade from Intel Xeon 3 to Intel Xeon 6. It delivers up to 2.5x CPU performance per bare metal instance and up to 50% better for VM instances, along with 42% higher memory bandwidth for high-performance and latency-sensitive workloads.
Transparent pricing for changing compute economics
As compute economics evolve, pricing clarity matters more than ever. Customers want to understand what they are paying for and how those costs map to the resources their workloads actually use.
With the new Oracle Acceleron family, OCI is realigning pricing to better reflect changing component economics. Across the new Oracle Acceleron family, OCPU-hour prices are up to 70% lower than the previous generation, while memory pricing now reflects the market reality of increasing memory costs. OCI continues to provide separate visibility into OCPU/hr and memory/hr pricing, along with a single worldwide pricing model, so customers can more clearly evaluate workload fit, price-performance, and the tradeoffs between compute and memory.
Learn more
The new Oracle Acceleron family brings together shapes designed for better workload fit, and transparent pricing to deliver stronger efficiency and clearer price-performance across a broad range of use cases.
To learn more about each new shape, read the dedicated launch blogs:
• E6 Standard/E6 DenseIO Acceleron
• X12 Standard Acceleron
• A4 Standard Acceleron