AWS continues to expand its cloud dominance with the general availability of M8azn instances, powered by fifth-generation AMD EPYC processors. These instances achieve a blistering 5 GHz clock speed, specifically designed for latency-sensitive tasks like real-time financial analytics and high-performance computing. By doubling compute performance compared to previous generations, Amazon is targeting industries where every millisecond counts, such as aerospace simulation and telecommunications.

In the generative AI space, Amazon Bedrock has significantly broadened its library by adding support for six prominent open-weight models. These include DeepSeek V3.2 and GLM 4.7, which cater to specialized workloads ranging from complex reasoning to autonomous coding. This move emphasizes a shift toward flexibility, allowing developers to deploy cost-effective models that are compatible with the widely used OpenAI API standards without being locked into a single provider.

Underpinning these updates is Project Mantle, a sophisticated distributed inference engine that manages how AI models are served to users (inference). By offering serverless inference with automated capacity management, AWS ensures that companies can scale their applications without managing underlying hardware. This simplifies the process of building agentic AI, which refers to systems capable of acting as autonomous agents to complete complex tasks independently.

Beyond hardware, the update introduces "Collection Groups" for Amazon OpenSearch Serverless, reducing costs by sharing computing resources across multiple data sets. These improvements highlight AWS's commitment to creating a more efficient, cohesive environment for developers building next-generation intelligent applications.

AWS continues to expand its cloud dominance with the general availability of M8azn instances, powered by fifth-generation AMD EPYC processors. These instances achieve a blistering 5 GHz clock speed, specifically designed for latency-sensitive tasks like real-time financial analytics and high-performance computing. By doubling compute performance compared to previous generations, Amazon is targeting industries where every millisecond counts, such as aerospace simulation and telecommunications.

In the generative AI space, Amazon Bedrock has significantly broadened its library by adding support for six prominent open-weight models. These include DeepSeek V3.2 and GLM 4.7, which cater to specialized workloads ranging from complex reasoning to autonomous coding. This move emphasizes a shift toward flexibility, allowing developers to deploy cost-effective models that are compatible with the widely used OpenAI API standards without being locked into a single provider.

Underpinning these updates is Project Mantle, a sophisticated distributed inference engine that manages how AI models are served to users (inference). By offering serverless inference with automated capacity management, AWS ensures that companies can scale their applications without managing underlying hardware. This simplifies the process of building agentic AI, which refers to systems capable of acting as autonomous agents to complete complex tasks independently.

Beyond hardware, the update introduces "Collection Groups" for Amazon OpenSearch Serverless, reducing costs by sharing computing resources across multiple data sets. These improvements highlight AWS's commitment to creating a more efficient, cohesive environment for developers building next-generation intelligent applications.