AWS Weekly Roundup: Anthropic & Meta partnership, AWS Lambda S3 Files, Amazon Bedrock AgentCore CLI, and more (April 27, 2026) | Amazon Web Services

This week’s AWS news centers on deeper AI infrastructure partnerships and tools for building production-ready agents. AWS and Anthropic are expanding Claude’s integration with AWS hardware and Amazon Bedrock, while Meta is adopting Graviton for large-scale agentic AI workloads. New services for Lambda, EKS, Aurora, and Bedrock also emphasize simpler data access, hybrid networking, serverless scaling, and faster agent development.

Anthropic and Meta Expand AWS AI Partnerships

• Anthropic is training advanced foundation models on AWS Trainium and Graviton processors.
• Anthropic and AWS’s Annapurna Labs are co-engineering at the silicon level to improve efficiency across the stack.
• Claude Cowork is now available through Amazon Bedrock, allowing enterprise teams to collaborate with Claude while keeping data within AWS.
• A unified Claude Platform on AWS is planned, offering a single experience for building, deploying, and scaling Claude applications.
• Meta signed an agreement to deploy tens of millions of AWS Graviton cores for CPU-intensive agentic AI tasks, including reasoning, code generation, search, and orchestration.

New Lambda and Kubernetes Infrastructure

• AWS Lambda can mount Amazon S3 buckets as file systems using S3 Files.
• Functions can perform standard file operations without downloading data first.
• Built on Amazon EFS, S3 Files combines file-system access with S3’s scalability, durability, and cost model.
• Multiple Lambda functions can share the same workspace, supporting AI agents that need persistent memory or shared state.
• The Amazon EKS Hybrid Nodes gateway simplifies networking between cloud-based EKS resources and on-premises Kubernetes Pods.
• It enables pod-to-pod traffic, control-plane webhook communication, and access to AWS services without making on-premises pod networks routable.
• The gateway is available at no additional charge.

Aurora Serverless and Bedrock Agent Development

• Aurora Serverless now offers up to 30% better performance on platform version 4.
• Its scaling algorithm better handles competing workloads, including busy APIs and bursty agentic AI applications.
• The service continues to scale to zero during idle periods, with no additional charge for the improvements.
• Amazon Bedrock AgentCore adds a managed harness in preview, allowing developers to define a model, system prompt, and tools without writing orchestration code.
• Harnesses can later be exported as Strands-based code for greater control.
• The AgentCore CLI supports governed, auditable deployments through AWS CDK, with Terraform support planned.
• The CLI is available in 14 AWS Regions at no additional charge, and AgentCore skills support coding assistants.

Cost Management, Operations, and Machine Learning

• Granular cost attribution for Amazon Bedrock enables teams to track usage by project or organization and support detailed chargeback.
• AWS DevOps Agent can work with the Salesforce MCP Server to investigate incidents, diagnose causes, and notify customers through Salesforce Service Cloud.
• AWS microcredentials are now free through AWS Skill Builder in supported countries.
• These hands-on assessments use simulated business scenarios and live AWS environments rather than traditional multiple-choice testing.
• Amazon SageMaker AI can recommend optimized generative AI inference configurations, including instance types, containers, and inference parameters, helping reduce latency and deployment costs.

Upcoming AWS Events

• “What’s Next with AWS” is scheduled as a virtual event on April 28.
• AWS Summits continue in May across cities including Singapore, Tel Aviv, Warsaw, Stockholm, Sydney, Hamburg, Seoul, Amsterdam, Bangkok, and Milan.

AWS’s latest releases point toward a more integrated AI platform: specialized hardware for model execution, managed agent tooling, shared state through serverless storage, and stronger cost and operational controls. Builders should evaluate S3 Files and AgentCore for AI workflows, while teams running production inference can benefit from SageMaker recommendations and Bedrock’s improved cost attribution.