AWS re:Invent 2025 - What's new with AWS Lambda (CNS376)

Overview of AWS Lambda Advancements

• AWS Lambda, the serverless computing service, has undergone significant transformations in 2025, enabling entirely new classes of workloads.
• The primary focus is on providing speed and agility to developers, allowing them to build engaging applications and iterate quickly based on user feedback.
• AWS aims to make Lambda applications "well-architected by default" by handling scalability, availability, reliability, and other operational concerns.

New Primitives for Expanded Workloads

Managed Instances for Steady-State Workloads

• Customers requested the ability to optimize performance for applications with steady-state traffic, beyond the initial burst and scale-to-zero use cases.
• Managed Instances allow developers to choose the specific compute, memory, or network instances to run their Lambda functions, while still benefiting from serverless operations.
• This enables a wider range of workloads to be migrated to serverless, including media processing, data processing, and web applications.
• Managed Instances also provide the ability to comply with regulatory requirements by specifying the compute zones or availability zones to use.

Tenant Isolation for Multi-Tenant Applications

• For SaaS builders and other multi-tenant applications, the new Tenant Isolation feature provides clean, isolated execution environments for each tenant.
• This eliminates the need for custom tooling and processes to manage tenant-specific state, improving CI/CD cycles and developer velocity.
• Developers only need to pass a unique tenant identifier, and AWS Lambda will provide the isolated execution environment.

Improving Developer Productivity and Application Resilience

New Runtimes and Automatic Upgrades

• AWS Lambda is introducing new runtimes, including Python 3.14, Java 25, and Node.js 24, with performance optimizations and security fixes.
• The goal is to make these new runtimes available within 90 days of their community release.
• To address the challenge of runtime upgrades, AWS is leveraging generative AI to automate the process, reducing technical debt by up to 85%.

Snap Start for Faster Cold Starts

• Snap Start, previously available for Java, is now expanded to Python and .NET, significantly reducing cold start times for Lambda functions.
• This feature takes a snapshot of the entire execution environment during function deployment, allowing for faster restoration on subsequent invocations.

Fault Injection Testing for Resilience

• AWS Fault Injection Service (FIS) integration allows developers to test the resilience of their Lambda-based applications by simulating various failure scenarios, such as increased latency or unavailable downstream dependencies.
• This helps developers identify and address potential issues, improving the overall reliability and stability of their applications.

Enhanced Observability for Event-Driven Patterns

• To address the observability challenges of event-driven architectures, AWS Lambda now provides additional CloudWatch metrics, including poller count, lag, and throttles, for event sources like Amazon SQS and Apache Kafka.
• This allows developers to more easily detect and resolve issues related to the event processing pipeline.

Avro Schema Registry Support

• For customers using Apache Kafka and the Avro data format, AWS Lambda now supports automatic serialization and deserialization of Avro events, eliminating the need for custom boilerplate code.
• This feature also enables schema evolution, reducing the effort required to manage schema changes.

Provisioned Concurrency for Faster Scaling

• The new Provisioned Concurrency feature for Amazon SQS triggers allows developers to pre-warm capacity to handle spikes in traffic, ensuring faster scaling and better SLA compliance.
• This builds on the previously released Provisioned Concurrency for Apache Kafka triggers.

Accelerating Developer Workflows

Seamless Console-to-IDE Integration

• Developers can now create a basic application directly in the AWS Lambda console and, with a single click, have the necessary dependencies packaged and ready to continue development in their local IDE.

Local Testing with Local Stack

• The integration with Local Stack allows developers to emulate and simulate various AWS services, enabling them to test their Lambda-based applications entirely on their local machines.
• This facilitates faster iterations and testing of business logic without the need for additional custom tooling.

Remote Debugging in Production

• The new Remote Debugging feature enables developers to set breakpoints and analyze variables in their Lambda functions running in production, similar to the experience of local debugging.
• This helps developers identify and address issues that may only manifest in the production environment, without the need for custom instrumentation.

Managed Code Pipeline (MCP) Server

• The MCP Server, released in Q2, bakes in best practices for code generation, including input validation, error handling, and consistent status codes.
• This helps ensure that the code generated by AI-powered coding assistants is production-ready, reducing the need for extensive code reviews and further accelerating developer velocity.

Conclusion and Next Steps

• The presentation covered a wide range of new capabilities and features for AWS Lambda, enabling developers to tackle a broader set of workloads and improve their overall development and deployment workflows.
• Attendees are encouraged to explore the additional sessions on serverless and the AWS Lambda roadmap, which is now available on GitHub for community feedback and participation.