AWS has unveiled Graviton5, its most advanced generation of custom-built Arm-based processors, marking a significant escalation in the hyperscaler’s long-running strategy to differentiate its cloud infrastructure through in-house silicon. Designed to serve a wide variety of compute, memory, and network-intensive workloads, Graviton5 introduces substantial performance improvements.
The Graviton5 processor would bring up to 25% higher compute performance than its predecessor – while reinforcing AWS’s focus on energy efficiency, sustainability, and hardware-rooted security.
For enterprise buyers navigating intensifying cost pressures and sustainability mandates, the launch underscores the accelerating shift toward vertically integrated cloud architectures. AWS’s ability to design chips, servers, and systems software as a unified stack is becoming a strategic lever, enabling performance gains that translate directly into lower operational expenditure and smaller carbon footprints for customers.
At the center of the announcement is the new Graviton5-based Amazon EC2 M9g instance family. With up to 192 cores in a single package – the highest CPU core density available on EC2 – AWS positions M9g as a platform for high-throughput, latency-sensitive, and scale-out workloads. According to AWS, the denser design shortens the physical distance between cores and reduces inter-core communication latency by as much as 33%. For businesses operating real-time gaming platforms, large-scale application servers, high-performance databases, big-data analytics, or semiconductor design workloads, this architectural shift could materially improve throughput and consistency under load.
Graviton5 integrates a dramatically expanded L3 cache, offering five times the total capacity of the previous generation. Each core gains access to 2.6 times more L3 cache than Graviton4, helping reduce stalls associated with fetching data from memory and improving overall application responsiveness. Memory subsystem performance also rises as AWS increases memory speeds, providing stronger support for data-intensive and memory-bound workloads.
Both networking and storage throughput receive sizable upgrades as well. The new instances deliver an average of 15% higher network bandwidth and 20% higher Amazon Elastic Block Store (EBS) bandwidth across instance sizes. The largest instances may see up to double the network bandwidth of their predecessors, accelerating distributed application performance, snapshot operations, and large-scale data transfers.
While performance dominates the headline, AWS’s commitment to energy efficiency remains central to the Graviton roadmap. By adopting 3nm process technology, optimizing chip architecture for cloud-native patterns, and enabling advanced system-level features such as bare-die cooling, AWS aims to deliver more compute per watt than competing architectures. For companies balancing growth with emissions-reduction targets, this energy profile may become a decisive factor.
Security is another pillar where AWS is expanding its proprietary advantage. Graviton5 builds upon the AWS Nitro System, which uses dedicated hardware to offload virtualization, network, and storage operations – freeing nearly all server compute and memory resources for customer workloads. Nitro’s zero-operator access design ensures that neither AWS personnel nor other tenants can access host memory or customer data.
New in this generation is the Nitro Isolation Engine, which uses formal verification methods to mathematically prove that isolation properties function as intended. This approach reduces the trusted computing base and introduces a level of security assurance rarely seen in commercial cloud environments. AWS plans to make the implementation and proof artifacts available to customers for evaluation, signaling a push toward greater transparency and verifiable guarantees.
Enterprises across industries are already validating the performance impact. Adobe reports faster real-time video stream processing, powering more personalized broadcast experiences. Epic Games notes improvements in latency and bandwidth that support smooth gameplay for millions of daily players. Formula 1 relies on Graviton to analyze race telemetry at speeds approaching 350 km/h, delivering instantaneous insights to global audiences.
Pinterest, which serves over 500 million monthly active users, highlights the price-performance advantages of running at scale on Graviton-based infrastructure. Airbnb reports that Graviton5-based M9g instances are among the fastest EC2 instances it has tested; production search workloads saw performance improvements of up to 25% over same-generation alternatives and up to 20% over Graviton4, with notable gains in P95 latency – critical for user experience consistency.
Atlassian, which has already migrated more than 3,000 EC2 instances for Jira and Confluence to Graviton4, reports 30% higher performance and 20% lower latency when testing Jira on Graviton5-powered M9g instances. Siemens Digital Industries Software cites up to 30% additional performance improvements for semiconductor design verification workloads, building on the 20% improvements and 30% cost reductions seen with Graviton4.
SAP reports some of the most dramatic gains: a 35% to 60% boost in OLTP query performance for SAP HANA Cloud on M9g instances. Synopsys, which collaborates with AWS on custom silicon development, notes up to 35% faster runtimes across tools such as PrimeTime and Fusion Compiler, while Arm observes up to 40% faster runtimes for Synopsys VCS compared with previous generations.
Graviton5-based EC2 M9g instances are available in preview, with C9g (compute-optimized) and R9g (memory-optimized) families scheduled for release in 2026. For enterprises evaluating long-term modernization strategies, these milestones indicate that AWS’s silicon roadmap is not only accelerating but becoming integral to its broader cloud differentiation.
Executive Insights FAQ
What business value does Graviton5 offer compared with previous AWS chip generations?
Graviton5 delivers up to 25% more compute performance, larger cache capacity, higher bandwidth, and improved energy efficiency. For enterprises, this translates into faster application throughput, cost savings from better price-performance ratios, and lower carbon impact from more efficient infrastructure utilization.
How does Graviton5 improve performance for data-intensive and latency-sensitive workloads?
With a fivefold increase in total L3 cache, faster memory speeds, and reduced inter-core latency, Graviton5 accelerates workloads that rely on rapid data access and high concurrency. Applications such as gaming backends, real-time analytics, and HPC simulations benefit from quicker response times and smoother scaling.
What distinguishes AWS’s security model with Graviton5 from other cloud processors?
Graviton5 leverages the AWS Nitro System’s hardware offload architecture and zero-operator access model, combined with the new Nitro Isolation Engine, which uses formal verification to mathematically validate isolation. This reduces the attack surface and offers security assurances uncommon in general-purpose cloud environments.
Which industries are seeing the strongest performance gains from migrating to Graviton5?
Media, gaming, enterprise SaaS, semiconductor EDA, industrial engineering, and ERP platforms have reported significant improvements. Use cases span real-time video processing, large-scale search, collaborative productivity software, chip design verification, and transactional database workloads.
When will enterprises be able to fully adopt the broader Graviton5 instance portfolio?
The general-purpose M9g instances are available in preview now. Compute-optimized C9g and memory-optimized R9g families are planned for 2026, giving organizations a roadmap to migrate diverse workload types as the ecosystem matures.
