The AWS cloud computing landscape has been significantly reshaped with the general availability of Amazon EC2 M9g instances, alongside the introduction of the new M9gd instances. These latest additions to the Amazon Elastic Compute Cloud (EC2) family are powered by the fifth-generation AWS Graviton5 processors, representing a monumental leap in compute performance, price-performance efficiency, and energy sustainability. First unveiled in preview at re:Invent 2025, the M9g instances have undergone extensive testing by a diverse range of customers, yielding remarkable performance improvements across critical workloads, with the M9gd variants now offering integrated high-speed, low-latency NVMe SSD local storage for demanding I/O-intensive applications.
A Legacy of Innovation: The Evolution of AWS Graviton Processors
The journey of AWS Graviton processors began in 2018 with the introduction of the first-generation Graviton, marking Amazon’s strategic foray into custom silicon for cloud workloads. This initiative was driven by the vision to deliver superior price-performance and energy efficiency tailored specifically for the unique demands of cloud environments. Each subsequent generation – Graviton2 in 2019, Graviton3 in 2021, and Graviton4 in 2023 – has built upon this foundation, consistently pushing the boundaries of what is possible with Arm-based architecture in the cloud. These iterations have brought incremental yet significant advances in compute capabilities, memory bandwidth, and overall system efficiency, steadily solidifying Graviton’s position as a preferred choice for a vast array of workloads.
The announcement of Graviton5 at re:Invent 2025 underscored AWS’s unwavering commitment to innovation and its leadership in custom silicon development. The preview period following this announcement allowed key partners and early adopters to rigorously evaluate the performance characteristics of M9g instances. The positive feedback and impressive benchmarks shared by these customers served as a strong validation of Graviton5’s capabilities, paving the way for its general release. This continuous, eight-year investment has culminated in a robust ecosystem, with Graviton now powering over 350 instance types, serving more than 120,000 customers globally, from agile startups to expansive enterprise operations. The broad adoption is further supported by a thriving Independent Software Vendor (ISV) partner ecosystem and integration across a wide spectrum of AWS managed services.
Graviton5: Architectural Brilliance Meets Unprecedented Performance
At the heart of the new M9g and M9gd instances lies the AWS Graviton5 processor, engineered to be the most powerful and energy-efficient processor AWS has ever developed. Graviton5 boasts an impressive 192 cores, a staggering five times larger L3 cache compared to its predecessor, and up to 33% lower inter-core latency. This architectural prowess is complemented by the integration of DDR5 memory, delivering unparalleled bandwidth, and making Graviton5 the first CPU in the AWS fleet to support the latest generation of PCIe Gen6. These combined advancements result in the fastest memory available in any processor instance in the cloud, setting a new benchmark for performance.
The performance gains are substantial and span a wide range of application categories. Compared to Graviton4-based instances, Graviton5 offers up to 25% better general compute performance. Specific workload optimizations yield even more impressive figures: web applications can experience up to a 35% faster performance, machine learning inference workloads see up to a 35% speed improvement, and database operations benefit from up to a 30% boost. These enhancements are not merely about speed; they also translate into superior energy efficiency, enabling organizations to achieve their sustainability targets without compromising on operational capability.
Customer Validation: Real-World Performance Impact
During the preview phase, leading technology companies put the M9g instances through their paces, reporting significant performance uplifts. These real-world results underscore the tangible benefits of migrating to Graviton5.
ClickHouse, a popular open-source column-oriented database management system, reported a remarkable 36% performance boost when running on M9g instances compared to M8g, achieved with zero code changes. This substantial improvement highlights Graviton5’s ability to accelerate data processing and analytics workloads, critical for real-time insights and large-scale data aggregation. The larger L3 cache and higher memory bandwidth of Graviton5 are particularly beneficial for data-intensive applications like ClickHouse, allowing for faster data retrieval and processing.
Honeycomb, a leading observability platform, conducted a rigorous six-month A/B test involving production observability workloads. Their findings revealed a 36% better throughput per core on Graviton5 compared to Graviton4. This indicates a significant increase in efficiency, allowing Honeycomb to process more telemetry data per unit of compute, which is crucial for monitoring complex distributed systems and ensuring application performance at scale. The improved core efficiency and lower inter-core latency of Graviton5 are key factors in achieving such throughput gains in highly concurrent environments.
HubSpot, a prominent customer relationship management (CRM) platform, deployed M9g instances for their MySQL databases and observed query durations dropping by up to 60%. This dramatic reduction in query time signifies a substantial improvement in database responsiveness and overall application performance for their users. For a platform like HubSpot, where real-time access to customer data is paramount, such performance gains can directly translate into enhanced user experience and operational efficiency. The faster memory, larger caches, and improved I/O capabilities of Graviton5 contribute to accelerating complex database queries.
These testimonials from diverse and demanding customers provide compelling evidence of Graviton5’s transformative potential across a broad spectrum of cloud workloads.
Expanding the Horizons of AI with Graviton5
The rapid evolution of Artificial Intelligence (AI), particularly the shift towards agentic AI, is driving an unprecedented demand for CPU compute power. Agentic AI models, which are designed to move beyond mere question-answering to performing complex actions, running code, using tools, evaluating results, and orchestrating multi-step tasks, are inherently CPU-intensive. Graviton5 is purpose-built to address this burgeoning demand. With its high core count (192 cores), expansive L3 cache, reduced inter-core latency, and high-bandwidth DDR5 memory, Graviton5 significantly reduces the time agents spend waiting on CPU-bound steps. This enables faster instruction processing, efficient handling of numerous concurrent environments, and seamless operation of accelerators.
A testament to Graviton5’s pivotal role in the future of AI is the strategic partnership with Meta. Meta is deploying Graviton at an immense scale, starting with tens of millions of cores, to support its ambitious agentic AI initiatives. This makes Meta one of the largest Graviton customers globally, a clear endorsement of the processor’s capabilities for cutting-edge AI research and deployment. Agentic AI workloads, which require sophisticated real-time reasoning, efficient code generation, and the intricate orchestration of multi-step processes, stand to gain immensely from Graviton5’s superior compute performance, larger caches, higher memory bandwidth, and impressive core density. This collaboration highlights Graviton’s emergence as a critical infrastructure component for the next generation of AI applications.
Advanced Networking, Storage, and Unprecedented Security
Beyond raw compute power, M9g and M9gd instances integrate significant advancements in networking, storage, and security, all powered by the sixth-generation AWS Nitro System.
Networking and storage bandwidth have been substantially expanded to keep pace with the increased 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 instance sizes. For the largest instance sizes, network bandwidth can even double. A groundbreaking feature, Instance Bandwidth Configuration (IBC), allows customers to fine-tune the allocation of bandwidth between Amazon EBS and Amazon Virtual Private Cloud (Amazon VPC) networking by up to 25%. This flexibility is crucial for optimizing performance in workloads with specific bandwidth requirements, such as database read/write operations, complex query processing, and extensive logging, ensuring faster data movement and improved throughput.
Security and isolation remain paramount in cloud computing. The AWS Nitro System, with its Nitro Hypervisor, has long been designed to rigorously isolate instances from each other and from AWS operators. With M9g and M9gd instances, AWS introduces the Nitro Isolation Engine, raising the bar for cloud security even further. The Nitro Isolation Engine is a purpose-built enhancement to the Nitro System, specifically designed to enforce instance isolation with mathematical precision through formal verification. This innovative component is responsible for mediating all access to virtual machine memory, CPU register state, and I/O devices via a minimal set of APIs. Formal verification is a rigorous technique that mathematically demonstrates the intended behavior of hardware or software, proving its correctness beyond specific test cases. This intensive verification process establishes Nitro as the first formally verified cloud hypervisor, setting a new industry standard for mathematically proven cloud security. This unparalleled level of assurance provides customers with greater confidence in the integrity and confidentiality of their workloads running on M9g and M9gd instances.
For workloads demanding high-speed, low-latency local storage, M9gd instances deliver up to 11.4 TB of NVMe SSD storage, offering 30% higher IOPS and overall storage performance compared to Graviton4-based M8gd instances. This makes M9gd instances ideal for applications requiring a balanced blend of compute and memory with extremely fast local storage, including application servers, microservices, gaming servers, midsize key-value data stores, caching fleets, data logging, media processing, batch and log processing, and any applications that rely on temporary storage like caches and scratch files.
Broad Application and Accessibility
M9g instances, with their balanced configuration of one vCPU for every four GiB of memory, are versatile general-purpose instances suitable for a vast array of 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 increasingly, agentic AI workloads.
Both M9g and M9gd instances are now generally 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 flexible purchasing options, including Savings Plans, On-Demand, Spot Instances, Dedicated Instances, or Dedicated Hosts, catering to diverse budget and operational requirements.
To facilitate adoption and optimization, AWS provides a suite of resources. The AWS Graviton Getting Started Guide offers comprehensive technical guidance on building, running, and optimizing workloads on Graviton-based instances. The Graviton Savings Dashboard helps organizations track and quantify the cost savings realized from migrating workloads to Graviton. Furthermore, AWS Transform, an AI-powered service, automates code transformations for migrating Java applications from x86 to Graviton-based Amazon EC2 instances, streamlining compatibility analysis, automated recompilation, dependency updates, and validation, thereby significantly reducing migration effort and time.
Broader Industry Impact and Future Outlook
The general availability of Graviton5-powered M9g and M9gd instances solidifies AWS’s leadership in the custom silicon market and marks a significant inflection point for Arm-based computing in the cloud. While other players in the industry have introduced Arm-based instances, AWS’s Graviton ecosystem stands unmatched in its breadth and depth. The continuous innovation across five generations and eight years of investment has created a mature, high-performance, and widely supported platform that challenges the traditional dominance of x86 architecture.
The economic implications are profound, with Graviton processors consistently offering superior price-performance, allowing customers to achieve more compute power for their budget. This efficiency, combined with significant energy savings, positions Graviton as a critical component for organizations looking to optimize both their operational costs and their environmental footprint. As cloud workloads continue to grow in complexity and scale, the demand for highly efficient, purpose-built processors will only intensify. Graviton5 is not just an incremental update; it is a foundational technology designed to meet the compute challenges of the next decade, particularly in the burgeoning fields of AI and high-performance computing.
The strategic partnership with Meta further validates Graviton’s capability to handle the most demanding and cutting-edge workloads. This collaboration serves as a powerful testament to the processor’s scalability, performance, and suitability for the future of AI development. As cloud providers increasingly invest in custom silicon, AWS Graviton remains at the forefront, driving innovation and setting new standards for what is achievable in cloud infrastructure.
In conclusion, the general availability of Amazon EC2 M9g and M9gd instances powered by Graviton5 processors represents a pivotal moment for cloud computing. By delivering unparalleled performance, enhanced security features like the Nitro Isolation Engine, superior energy efficiency, and a robust ecosystem, AWS continues to empower customers to build, deploy, and scale their most demanding workloads with greater efficiency and confidence.
