Claude Platform on AWS: Native Anthropic APIs Inside Your AWS Account
AWS has added Claude Platform on AWS, giving developers direct access to Anthropic's native Claude Platform APIs and the Anthropic Console without leaving the AWS account boundary. Previously, accessing Claude through AWS meant using Amazon Bedrock's wrapped API surface; Claude Platform on AWS exposes the full Anthropic API directly — including all Claude model variants, the same tooling available through Anthropic's own console, and the same pricing — while billing flows through the AWS account.
For enterprise developers already managing cloud costs through AWS consolidated billing, this removes a friction point: Claude API usage joins the same invoice as EC2, S3, and Lambda rather than requiring a separate Anthropic account and payment method. It also means Claude API access can be governed by the same IAM policies, AWS Organizations controls, and CloudTrail audit logging that govern other AWS services, addressing security and compliance requirements that previously made direct Anthropic API use difficult in regulated industries. Developers building on Claude Code, the Anthropic SDK, or any Claude-based tooling retain the exact same API behaviour — the change is in the billing and access pathway, not the model or API surface.
Read more — AWS Blog
EC2 M3 Ultra Mac Instances: Double the Memory for macOS Workloads
AWS has launched EC2 M3 Ultra Mac instances, the first Apple Silicon Mac instances in the M3 generation to bring the full 192GB unified memory configuration of the Mac Studio M3 Ultra to the cloud. The M3 Ultra chip's 28-core CPU and 60-core GPU make these instances relevant for iOS and macOS build workloads that previously required multiple smaller instances in parallel, and for ML inference workloads on macOS that benefit from Apple's unified memory architecture.
For development teams maintaining iOS or macOS apps, the practical benefit is faster CI/CD pipelines for large Xcode builds. A 192GB unified memory pool allows significantly larger build graphs to remain in memory simultaneously, reducing the cache misses and swap overhead that slow down builds on smaller Mac instances. The M3 Ultra generation also brings Metal 3 GPU improvements that benefit tools like Core ML and Turi Create that developers use for on-device model training and inference testing. Pricing follows the existing Mac instance model: instances are provisioned on dedicated Mac mini hardware and billed by the hour with a 24-hour minimum allocation period.
Read more — AWS Blog
AWS Transform at One Year: 4.5 Billion Lines of Code Migrated
AWS Transform, the agentic migration service that automates large-scale code modernisation and cloud migration projects, has reached its one-year anniversary with numbers that demonstrate meaningful adoption. The service has processed over 4.5 billion lines of code and migrated hundreds of thousands of servers, primarily across Java-to-cloud migrations, .NET modernisation, and VMware-to-EC2 transitions.
The milestone update introduces two notable integrations. AWS Kiro, the replacement for Amazon Q Developer IDE plugins announced earlier this month, is now integrated with Transform to provide spec-driven guidance during migration projects — agents can use Kiro's steering file format to maintain migration conventions across large codebases. Claude (via the Claude Platform on AWS integration) is now the primary model for Transform's analysis and code generation tasks, replacing the previous mix of models. The move to Claude for core Transform functionality is consistent with the AWS-Anthropic strategic partnership and aligns with the Bedrock team's broader direction of positioning Claude as the default model for agentic developer tooling on AWS.
Read more — AWS Blog
Links & Sources
You May Also Enjoy
