Amazon Web Services (AWS) today announced the general availability of its highly anticipated Amazon EC2 M9g and M9gd instances, powered by the fifth-generation AWS Graviton5 processors. This marks a significant milestone in cloud computing, offering customers access to AWS’s most powerful and energy-efficient processor to date. The M9g instances provide a balance of compute, memory, and networking resources for a broad range of general-purpose workloads, while the M9gd instances add high-speed, low-latency local NVMe SSD storage for I/O-intensive applications. This launch follows a successful preview period, first announced at re:Invent 2025, during which early adopters reported substantial performance gains across diverse production environments.
A New Era of Cloud Performance and Efficiency
The journey of AWS Graviton processors has been characterized by consistent innovation, with each generation delivering marked improvements in compute performance, price-performance ratios, and energy efficiency. From their initial introduction, Graviton processors have represented AWS’s strategic commitment to custom silicon, offering a compelling alternative to traditional x86 architectures by leveraging the Arm instruction set architecture. This strategy has allowed AWS to optimize hardware and software integration specifically for cloud workloads, resulting in superior performance characteristics and cost benefits for customers.
The Graviton5 processor is the culmination of this continuous investment, engineered to meet the escalating demands of modern cloud applications, particularly those driven by artificial intelligence (AI) and complex data analytics. It boasts architectural enhancements designed to accelerate processing, minimize latency, and improve data throughput, positioning it as a foundational technology for the next wave of cloud innovation.
During the preview phase, several prominent customers rigorously tested the M9g instances, providing compelling evidence of their capabilities. ClickHouse, a leading open-source columnar database management system, reported an impressive 36% performance boost compared to previous-generation M8g instances, achieved without requiring any code changes—a testament to the processor’s seamless integration and raw power. Similarly, Honeycomb, an observability platform, documented a 36% improvement in throughput per core over Graviton4-powered instances during an extensive six-month A/B test involving production observability workloads. HubSpot, a leading customer relationship management (CRM) platform, deployed M9g instances for their MySQL databases and observed query durations plummet by up to 60%, highlighting the processor’s effectiveness in accelerating database operations. These results underscore the tangible benefits that Graviton5 brings to critical business applications.
Graviton’s Expanding Footprint and Strategic Importance
The general availability of M9g and M9gd instances solidifies AWS’s leadership in custom silicon development within the cloud industry. After five generations of processors and eight years of dedicated research and development, Graviton now powers over 350 instance types, serving more than 120,000 customers worldwide. This expansive adoption spans a diverse ecosystem, from agile startups to multinational enterprises, supported by a robust network of independent software vendor (ISV) partners and a broad array of managed services that leverage Graviton underneath.
Graviton’s versatility enables it to support a wide spectrum of workloads, including web applications, microservices, analytics, databases, machine learning (ML) inference, electronic design automation (EDA), gaming, and video encoding. As enterprises increasingly migrate compute-intensive and data-driven workloads to the cloud, the demand for greater processing power, coupled with enhanced network and storage bandwidth, has become paramount. The Graviton5 instances are meticulously designed to address these requirements, efficiently packaging compute, memory, and I/O capabilities to maximize energy utilization and minimize operational costs.
Architectural Innovations for the AI Era
The advent of agentic AI—where AI systems not only answer questions but also take autonomous actions, execute code, utilize tools, evaluate results, and orchestrate complex multi-step tasks—is driving an unprecedented surge in demand for CPU compute. Graviton5 is specifically architected to excel in this evolving landscape. With an impressive 192 cores, a 5x larger L3 cache, up to 33% lower inter-core latency, and high-bandwidth DDR5 memory, Graviton5 helps AI agents process more instructions, manage vast numbers of concurrent environments, and keep accelerators like GPUs fully utilized, thereby reducing time spent on CPU-bound steps.
In a landmark strategic move, Meta, a global leader in social media and AI research, is deploying Graviton at scale, commencing with tens of millions of cores to bolster its agentic AI initiatives. This makes Meta one of the largest Graviton customers globally, underscoring the processor’s critical role in powering cutting-edge AI development. Agentic AI workloads, characterized by their CPU-intensive nature for real-time reasoning, code generation, and multi-step task orchestration, directly benefit from Graviton5’s superior compute performance, larger caches, higher memory bandwidth, and impressive core density.
Key Innovations in M9g and M9gd Instances
Built on the sixth-generation AWS Nitro System, M9g instances are powered by AWS Graviton5 processors that deliver substantial performance gains over their predecessors. Graviton5 offers up to 25% better overall compute performance compared to Graviton4-based instances. Specific workload optimizations yield even greater benefits: up to 35% faster performance for web applications, up to 35% for machine learning inference tasks, and up to 30% for database operations.
The Graviton5 processor is a technological marvel, being the first CPU in the AWS fleet to support the latest generation of PCIe Gen6 and DDR5-8800 memory. This combination delivers the fastest memory of any processor instances in the cloud, coupled with a staggering five times more L3 cache compared to the previous generation. These architectural advancements translate directly into improved application responsiveness and efficiency. Crucially, these performance enhancements are delivered with superior energy efficiency, enabling organizations to meet their sustainability targets without compromising on capability or scalability.
Networking and storage bandwidth have also been significantly expanded to keep pace with the enhanced compute capabilities. M9g and M9gd instances offer up to 15% higher network bandwidth and 20% higher Amazon Elastic Block Store (Amazon EBS) bandwidth on average across all instance sizes. For the largest instance sizes, network bandwidth can reach up to twice that of previous generations.
A notable new feature is Instance Bandwidth Configuration (IBC), which allows customers to fine-tune the allocation of bandwidth between Amazon EBS and Amazon Virtual Private Cloud (Amazon VPC) networking for an Amazon EC2 instance by up to 25%. This granular control is invaluable for optimizing performance for workloads with specific bandwidth requirements, such as high-volume database reads and writes, rapid query processing, and extensive logging operations, ensuring faster data movement and improved throughput.
Pioneering Cloud Security with Nitro Isolation Engine
Security and isolation are fundamental pillars of the AWS cloud infrastructure. Within the AWS Nitro System, the Nitro Hypervisor is meticulously designed to provide robust isolation between instances and from AWS operators. With the introduction of M9g and M9gd instances, AWS is further elevating its security posture with the Nitro Isolation Engine.
The Nitro Isolation Engine is a groundbreaking enhancement to the Nitro System, purpose-built to enforce instance isolation with mathematical precision through formal verification. This component is solely responsible for mediating all access to virtual machine memory, CPU register state, and I/O devices via a minimal set of APIs. Formal verification, a technique that mathematically demonstrates that hardware or software behaves exactly as intended, even in edge cases, is employed to rigorously prove the isolation properties of the Nitro Isolation Engine. This intensive verification establishes Nitro as the first formally verified cloud hypervisor, setting an unprecedented standard for mathematically proven cloud security and offering customers unparalleled assurance regarding the integrity and confidentiality of their workloads. Further details on the Nitro Isolation Engine and its formal verification results are available in dedicated AWS blog posts and technical whitepapers.
Instance Specifications and Workload Suitability
M9g instances are designed with a balanced ratio of one vCPU for every four GiB of memory, making them highly versatile for a broad range of general-purpose workloads. These include application servers, microservices architectures, midsize data stores, gaming servers, caching fleets, containerized applications, large-scale Java applications, code repositories, web applications, and, critically, agentic AI workloads.
For applications demanding high-speed, low-latency local storage, the M9gd instances are the ideal choice. They provide up to 11.4 TB of NVMe SSD storage and boast 30% higher IOPS and overall storage performance compared to Graviton4-based M8gd instances. M9gd instances are particularly well-suited for general-purpose workloads requiring a delicate balance of compute and memory alongside high-performance local storage. This includes application servers, microservices, gaming servers, midsize key-value data stores, caching fleets, data logging, media processing, batch and log processing, and any application that benefits from temporary storage such as caches and scratch files.
The M9g family offers a comprehensive range of instance sizes to cater to varying workload demands, from medium with 1 vCPU and 4 GiB of memory to the powerful 48xlarge and metal-48xl instances, featuring 192 vCPUs and 768 GiB of memory. These larger instances also provide up to 100 Gbps of network bandwidth and 72 Gbps of EBS bandwidth, enabling massive scalability for the most demanding applications.
Here are the key specifications across the M9g family:
| M9g | vCPUs | Memory (GiB) | Network bandwidth (Gbps) | EBS bandwidth (Gbps) |
|---|---|---|---|---|
| medium | 1 | 4 | Up to 15 | Up to 12 |
| large | 2 | 8 | Up to 15 | Up to 12 |
| xlarge | 4 | 16 | Up to 15 | Up to 12 |
| 2xlarge | 8 | 32 | Up to 17 | Up to 12 |
| 4xlarge | 16 | 64 | Up to 17 | Up to 12 |
| 8xlarge | 32 | 128 | 17 | 12 |
| 12xlarge | 48 | 192 | 25 | 18 |
| 16xlarge | 64 | 256 | 34 | 24 |
| 24xlarge | 96 | 384 | 50 | 36 |
| 48xlarge | 192 | 768 | 100 | 72 |
| metal-48xl | 192 | 768 | 100 | 72 |
M9gd instances include local NVMe SSD storage, with instance storage capacities scaling from 1 x 59 NVMe SSD for the medium instance up to 3 x 3800 NVMe SSD for the 48xlarge and metal-48xl instances. Compute, memory, network, and EBS bandwidth specifications for M9gd instances are identical to their M9g counterparts.
| M9gd | vCPUs | Memory (GiB) | Instance storage (GB) | Network bandwidth (Gbps) | EBS bandwidth (Gbps) |
|---|---|---|---|---|---|
| medium | 1 | 4 | 1 x 59 NVMe SSD | Up to 15 | Up to 12 |
| large | 2 | 8 | 1 x 118 NVMe SSD | Up to 15 | Up to 12 |
| xlarge | 4 | 16 | 1 x 237 NVMe SSD | Up to 15 | Up to 12 |
| 2xlarge | 8 | 32 | 1 x 475 NVMe SSD | Up to 17 | Up to 12 |
| 4xlarge | 16 | 64 | 1 x 950 NVMe SSD | Up to 17 | Up to 12 |
| 8xlarge | 32 | 128 | 1 x 1900 NVMe SSD | 17 | 12 |
| 12xlarge | 48 | 192 | 3 x 950 NVMe SSD | 25 | 18 |
| 16xlarge | 64 | 256 | 1 x 3800 NVMe SSD | 34 | 24 |
| 24xlarge | 96 | 384 | 3 x 1900 NVMe SSD | 50 | 36 |
| 48xlarge | 192 | 768 | 3 x 3800 NVMe SSD | 100 | 72 |
| metal-48xl | 192 | 768 | 3 x 3800 NVMe SSD | 100 | 72 |
Availability and Migration Resources
M9g and M9gd instances are currently available in key AWS Regions, including US East (N. Virginia), US East (Ohio), US West (Oregon), and Europe (Frankfurt). Customers can procure these instances through various purchasing options, including Savings Plans, On-Demand, Spot Instances, Dedicated Instances, or Dedicated Hosts, providing flexibility to optimize costs based on workload predictability and usage patterns. Detailed pricing information is available on the Amazon EC2 pricing page.
To facilitate adoption and migration, AWS provides a suite of resources. The AWS Graviton Getting Started Guide offers comprehensive technical instructions for building, running, and optimizing workloads on Graviton-based instances. For organizations keen on tracking their financial benefits, the Graviton Savings Dashboard helps monitor and measure cost savings achieved by migrating to Graviton. Furthermore, AWS Transform, an AI-powered service, streamlines the migration of Java applications from x86 to Graviton-based Amazon EC2 instances by automating compatibility analysis, recompilation, dependency updates, and validation, significantly reducing the effort and time required for transition.
The general availability of Graviton5-powered M9g and M9gd instances represents a critical leap forward in cloud infrastructure, offering unmatched performance, energy efficiency, and security features. As the cloud continues to evolve, driven by advancements in AI and data-intensive applications, Graviton5 is poised to become a cornerstone technology, empowering businesses to innovate faster, operate more cost-effectively, and build the next generation of intelligent applications.
