Top AWS AI & Cloud Innovations: Claude, Graviton, and Key Updates

AWS has deepened its collaboration with leading AI companies and unveiled a series of infrastructure and tooling updates designed to accelerate enterprise adoption of generative AI. The announcements, highlighted in the latest AWS Weekly Roundup, reflect a strategic push to integrate advanced AI capabilities directly into AWS’s cloud ecosystem while optimizing performance, cost, and developer productivity.

Anthropic Partnership Expands AI Training on AWS Silicon

AWS and Anthropic have strengthened their product collaboration, with Anthropic now training its most advanced foundation models on AWS’s custom silicon—Trainium and Graviton—through co-engineering efforts with Amazon’s Annapurna Labs. This integration aims to maximize computational efficiency from the hardware layer up through the full software stack, enabling faster and more cost-effective AI model development.

The partnership also introduces Claude Cowork, a collaborative AI capability now available within Amazon Bedrock. Claude Cowork allows enterprise teams to deploy Anthropic’s AI as a true collaborator rather than just a tool, enabling secure, team-based AI workflows while keeping data within the AWS environment. This integration is particularly significant for organizations seeking to embed AI into complex, multi-step business processes without compromising data security or compliance.

Looking ahead, AWS plans to launch the Claude Platform on AWS, a unified developer experience for building, deploying, and scaling Claude-powered applications directly through Amazon Bedrock. The platform is positioned as a major step forward for developers working with generative AI on AWS, offering streamlined access to Anthropic’s models without requiring external integrations.

Meta Adopts AWS Graviton for Agentic AI Workloads

Meta has signed an agreement to deploy AWS Graviton processors at scale, marking one of the largest adoptions of AWS’s custom silicon to date. The deployment will begin with tens of millions of Graviton cores powering CPU-intensive agentic AI workloads, including real-time reasoning, code generation, search, and multi-step task orchestration. This move underscores Graviton’s growing role in handling high-performance AI workloads, particularly for applications requiring low-latency processing and cost efficiency.

Graviton5, AWS’s latest custom CPU announced at re:Invent 2025, delivers up to 25% higher performance than its predecessor, with 192 cores per chip and a 5x larger cache. The processor has gained traction among AWS’s largest customers, including Adobe, Airbnb, Epic Games, Formula 1, Pinterest, SAP, and Siemens, with over half of new CPU capacity added to AWS in the past three years powered by Graviton. Meta’s adoption further validates AWS’s strategy of offering custom silicon tailored for AI and cloud-native workloads.

New Tools for AI and Cloud Development

AWS has introduced several updates to its developer tools and services, aimed at simplifying AI and cloud workflows:

• AWS Lambda S3 Files: Lambda functions can now mount Amazon S3 buckets as file systems using S3 Files, enabling standard file operations without downloading data for processing. Built on Amazon EFS, this feature allows multiple Lambda functions to share data through a common workspace, which is particularly valuable for AI and machine learning pipelines where agents need to persist memory and share state across steps.
• Amazon EKS Hybrid Nodes Gateway: Amazon Elastic Kubernetes Service now includes the EKS Hybrid Nodes gateway, which automates networking between EKS cluster VPCs and Kubernetes Pods running on hybrid nodes. This eliminates the need for manual network configuration, simplifying hybrid Kubernetes environments by enabling seamless pod-to-pod traffic across cloud and on-premises environments.
• Amazon Aurora Serverless Improvements: The latest version of Amazon Aurora Serverless delivers up to 30% better performance and smarter scaling algorithms, designed to handle workloads with competing resource demands, such as busy APIs and agentic AI applications. The service now supports more demanding workloads while maintaining its pay-per-use model and automatic scaling to zero when idle.
• Amazon Bedrock AgentCore: AgentCore introduces new features to help developers build AI agents faster, including a managed harness (in preview), a CLI, and pre-built skills for coding assistants. The AgentCore CLI allows developers to deploy agents with governance and auditability through infrastructure-as-code, supporting AWS CDK today with Terraform support coming soon. These tools are available in 14 AWS Regions at no additional charge.

Cost Optimization and Developer Resources

AWS has also rolled out features to improve cost visibility and developer productivity:

• Granular Cost Attribution for Amazon Bedrock: Organizations can now tag and track Bedrock usage costs at a finer level of detail, enabling precise cost visibility and chargeback capabilities for teams or projects running on the platform. This feature is particularly useful for enterprises managing multiple AI initiatives with varying budget requirements.
• AWS DevOps Agent Integration with Salesforce: AWS DevOps Agent, integrated with Salesforce’s MCP Server, automates the full lifecycle of infrastructure incident investigation, from identifying issues to diagnosing root causes and notifying customers through Salesforce Service Cloud. This integration demonstrates how AI agents and MCP-based tool connectivity are reshaping DevOps workflows, reducing mean time to resolution in production environments.
• Free AWS Microcredentials: AWS has made its microcredentials available at no cost through AWS Skill Builder. Unlike traditional certifications, these hands-on assessments place builders in simulated business scenarios where they configure, troubleshoot, and optimize directly in a live AWS environment, mirroring real-world job requirements.
• Amazon SageMaker AI Inference Recommendations: SageMaker AI now supports automated identification of optimized deployment configurations for generative AI models, including instance type, container, and inference parameters. This capability helps reduce costs and improve latency for AI applications in production by eliminating the guesswork in tuning inference infrastructure.

Upcoming AWS Events

AWS has announced several upcoming events for developers, builders, and enterprise customers:

• What’s Next with AWS: A virtual event on April 28 featuring the latest product announcements and updates directly from AWS teams.
• AWS Summits: Free in-person events focused on cloud and AI innovation, with upcoming summits in Singapore (May 6), Tel Aviv (May 6), Warsaw (May 6), Stockholm (May 7), Sydney (May 13–14), Hamburg (May 20), Seoul (May 20), Amsterdam (May 27), Bangkok (May 28), Milan (May 28), and Mumbai (May 28). The Los Angeles Summit is scheduled for June 10.
• AWS Community Days: Community-led conferences featuring technical discussions, workshops, and hands-on labs. Upcoming events include Athens, Greece (April 28), Vancouver, Canada (May 1), İstanbul, Türkiye (May 9), and Panama City, Panama (May 23). The AWS Community Day Belo Horizonte in Brazil is set for August 22.

These updates reflect AWS’s broader strategy to integrate AI deeply into its cloud ecosystem while providing developers and enterprises with the tools and infrastructure needed to build, deploy, and scale AI applications efficiently. The focus on custom silicon, cost optimization, and developer productivity positions AWS as a key player in the evolving AI landscape, particularly for organizations seeking to balance performance, security, and cost.