Amazon Web Services expands EC2 instance lineup for AI workloads

Amazon Web Services announced that its new I8ge storage‑optimized EC2 instances are now generally available in the Europe (London) and Canada (Central) regions. Powered by Graviton4 processors, I8ge instances deliver up to 60 % higher compute performance than the previous Graviton2‑based generation and up to 55 % better real‑time storage performance per terabyte, with latency reductions of up to 75 %. The instances offer up to 120 TB of local NVMe storage and up to 180 Gbps of network bandwidth.

In addition, AWS made M8a instances available in the Asia Pacific (Hyderabad) region and expanded G7e instances for SageMaker AI inference across more regions. The announcements accompany a broader evolution of EC2 compute to meet growing demand from generative and agentic AI workloads, including the continued integration of AMD EPYC processors, high‑frequency instances with 5 GHz clocks, and expanded Nitro security features.

AWS executives highlighted that AI workloads are reshaping cost and infrastructure decisions for enterprises, driving a shift from purely cloud‑centric models to more nuanced strategies that balance performance, storage, and network requirements across a diversified instance portfolio.