AWS launches Graviton5 memory-optimized EC2 R9g instances

Amazon Web Services (AWS) has introduced its latest memory-optimized cloud instances, the EC2 R9g and R9gd families, now generally available with AWS Graviton5 processors. The new instances are designed for memory-intensive applications such as databases, in-memory caching, real-time analytics, and containerized workloads, delivering up to 25% higher compute performance per vCPU compared to their Graviton4-based predecessors. Alongside improved processing power, the R9g instances feature faster memory, larger caches, and increased network and storage bandwidth, addressing growing demands for high-performance cloud infrastructure in data-heavy environments.

Performance and target workloads

The R9g family is positioned as a successor to earlier memory-optimized instances, offering enhanced efficiency for workloads that require low-latency access to large datasets. AWS highlights use cases including relational and NoSQL databases, distributed caching systems like Redis, and real-time analytics platforms. The R9gd variant adds local NVMe storage, catering to applications that benefit from high-speed, ephemeral data access. Both instance types are available in multiple sizes, allowing customers to scale resources based on workload requirements.

AWS has not disclosed specific pricing for the R9g instances, but the company’s historical approach suggests competitive rates relative to comparable x86-based offerings. The Graviton processor line, developed in-house by AWS, has been a key differentiator in the cloud market, enabling the provider to optimize hardware and software stacks for cost and performance. Graviton5 continues this strategy, with AWS framing the launch as part of its broader effort to reduce infrastructure costs for customers while improving compute density.

Market context and competition

The release arrives amid intensifying competition among hyperscalers to capture demand for high-performance cloud infrastructure. Microsoft Azure and Google Cloud have both expanded their own custom silicon offerings in recent years, with Azure’s Cobalt and Google’s Axion processors targeting similar workloads. AWS’s decision to focus on memory-optimized instances reflects ongoing demand from enterprises migrating large-scale databases and analytics workloads to the cloud, particularly in sectors like finance, e-commerce, and logistics.

The launch also follows AWS’s recent moves to strengthen its position in the analytics and AI infrastructure markets. In late August, AWS announced plans to deploy an additional two million NVIDIA GPUs, responding to surging demand for AI training and inference workloads. While the R9g instances are not directly tied to AI, their improved memory bandwidth and compute performance could benefit certain AI-related applications, such as feature stores or real-time inference pipelines.

What to watch

For cloud buyers, the R9g instances provide another option for optimizing workload costs, particularly for memory-bound applications. Enterprises already using Graviton-based instances may see immediate benefits from migrating to the new family, while those still on x86 architectures may evaluate the performance and cost trade-offs of switching. AWS’s ability to sustain its custom silicon roadmap will also be a factor in long-term cloud economics, as competitors continue to invest in their own hardware innovations.

The broader trend of hyperscalers designing custom processors underscores a shift in the cloud market, where hardware differentiation is increasingly tied to software and ecosystem integration. As AWS, Microsoft, and Google expand their silicon portfolios, customers may face more complex decisions around instance selection, workload placement, and cost management. The R9g launch is unlikely to disrupt the market immediately but reinforces AWS’s strategy of using custom hardware to drive cloud adoption and retention.