11/19/2026, 12:00:00 AM ~ 11/20/2026, 12:00:00 AM (UTC)

Recent Announcements

AWS Marketplace now supports agentic payments with AgentCore

AWS Marketplace now supports agentic payments over the x402 protocol, so AI agents can discover and buy select products and pay per request in US dollars through AgentCore payments, part of Amazon Bedrock AgentCore. Agents often need paid APIs, data, or content mid-task, but access usually requires a person to sign up, add a payment method, and create an API key. With agentic payments, customers skip that per-seller setup, and charges settle on their existing AWS bill with no crypto wallet required.\n Customers enable the AWS Marketplace payment connector in AgentCore payments, then they or their agents find listings through the AWS Marketplace connector in AgentCore Gateway. For example, a research agent can buy a single web search, or a coding agent can call a paid guardrail service. Customers can set spending limits, approved sellers, and guardrails before agents transact, Private Marketplace policies still apply, and transactions appear in AWS CloudTrail and AWS Cost Explorer.

AWS Marketplace sellers can charge AI agents per request without building a sign-up flow or billing system. They register with AWS Marketplace, configure agentic payments for their listings, and integrate the AWS Marketplace x402 SDK. AWS Marketplace collects customer payments and pays sellers through their AWS Marketplace seller disbursements, like other AWS Marketplace transactions.

The capability is available in all AWS Regions where AgentCore payments is available. To learn more, see AWS Marketplace and the Amazon Bedrock AgentCore Developer Guide. To list as a seller, see the AWS Marketplace Seller Guide.

Amazon Aurora DSQL now delivers up to 50% higher write throughput

With upgrades to its storage layer, Amazon Aurora DSQL now delivers up to 50% higher write throughput for many transactional workloads. Data loads finish sooner, and tables whose primary key always increases, such as tables keyed by a timestamp, an identity column, or a time-ordered ID like ULID or UUIDv7, see even greater benefits.\n Applications such as audit logs, ledgers, and event feeds are read in time order, and keying them by time keeps those reads fast. You can now use that design for write-heavy tables too. In our tests, an audit log table keyed by event time accepted up to 200% more inserts per second.

The improvement is available in all AWS Regions where Aurora DSQL is available, for new and existing clusters, at no additional cost. To learn more, see the Aurora DSQL User Guide.

Amazon CloudFront adds IP based routing and connection logs to SaaS Manager

Amazon CloudFront now allows web development and SaaS platforms to serve different tiers of customers on the same wildcard domain by using IP addresses to route them to the right configuration. For example, these platforms can serve premium and pro customers from dedicated Anycast static IP address lists, each with its own configuration, and serve free-tier customers on the shared CloudFront IP addresses, all under *.example.com. Additionally, these customers also have visibility into viewer connections with connection logging.\n Previously, CloudFront chose the configuration from the domain name alone, so the same domain always went to the same configuration. With address-based selection, you can register the same wildcard domain on several IP lists and send each list to its own tenant. Because you control which clients resolve to which IP list, you can shift traffic between your service tiers without your customers needing to make any changes. You can also serve IoT devices, embedded SDKs, and older TLS clients that connect by IP address without sending a domain name in the TLS handshake (Server Name Indication, or SNI). With viewer fallback, you choose a distribution to accept these connections, and CloudFront uses the Host header in each request to serve the right content. Connection logs now also record HTTPS connections to these tenants which helps you troubleshoot how clients connect.

These features are available at all CloudFront edge locations. To learn more, see Address-based selection and Connection Logs in the Amazon CloudFront Developer Guide.

Amazon CloudFront simplifies domain moves and onboarding for SaaS Manager

Amazon CloudFront now lets web development platforms and SaaS providers move a customer’s domain to their platform in a single request with no downtime, and serve requests to the domain while the TLS certificate is being configured. The CloudFront multi-tenant distributions allow these platforms to serve many customer domains from one shared configuration. Together, these features help move existing AWS customers to multi-tenant distributions without impacting their traffic.\n Previously, moving a domain that was already in use required disabling the original distribution, using a wildcard workaround, or contacting AWS Support. Now you can bring customer domains onto your multi-tenant setup as part of creating or updating a tenant. With the new OnDomainTakeover option, CloudFront checks domain ownership through DNS records and then moves one or more domains in a single request. This works within an AWS account or across accounts. If you add a domain before its certificate is issued, CloudFront serves it over HTTP until you attach a certificate. That gives you time to validate the domain with AWS Certificate Manager (ACM) or another certificate authority and attach the certificate once it’s issued.

These features are available in all edge locations at no additional cost. To learn more, see Move domains and Domains without a certificate in the Amazon CloudFront Developer Guide.

Amazon Connect launches Resolutions, an agentic solution for high-volume operational work

Amazon Connect Customer announces the general availability of Resolutions, an agentic operations solution that resolves high-volume work end-to-end, including refund requests, transaction disputes, fraud investigations, and insurance claims. Resolutions helps businesses reduce resolution times and cost to serve by combining the reasoning of AI agents with the reliability of deterministic workflows. Every request follows business policy, gets the oversight it requires, and ends in a verifiable outcome.\n Managers describe a process in natural language or upload standard operating procedures (SOPs), and Resolutions generates a Work plan in minutes. As it builds the Work plan, Resolutions asks clarifying questions that surface exceptions, set decision rules, and define approval thresholds, closing the gaps an SOP leaves open. Resolutions then opens a Resolutions Case for each incoming request and applies the right Work plan. AI agents gather information, apply policy, act across your business applications, and ask customers for missing details to drive every Case to completion. Each Step can run on prebuilt AI agents, AI agents configured in Amazon Connect Customer, or your own, while human specialists handle approvals and exceptions with the context they need to act.

For example, a bank resolving a transaction dispute can have AI agents pull the transaction history, check the claim against chargeback rules, email the cardholder for a missing receipt, and issue a provisional credit, so a specialist reviews only the disputes that need judgment. Resolutions verifies each outcome and records every action and decision, giving operations leaders an auditable, real-time view of progress, workload, and service level risk.

To learn more, visit our documentation, webpage, and pricing page.

AWS Marketplace agent mode now helps you assess, buy, and deploy solutions

You can now assess a seller’s security, buy, and deploy supported products in one conversation with AWS Marketplace agent mode, the conversational AI experience for procuring cloud solutions. You share your requirement once, and agent mode applies them as it guides you through discovery, purchase, and deployment.\n Describe your use case in natural language to get recommended software, services, AI agents and tools, and multi-product solutions that fit your business needs. To assess a seller’s security, request documents such as System and Organization Controls (SOC) 2 reports, and get a criterion-by-criterion assessment against a standard cloud security questionnaire, the Consensus Assessments Initiative Questionnaire (CAIQ).

When you’re ready to buy, sign in with your AWS account. Agent mode explains pricing and terms so you know what you’re committing to, shows your organizations’ purchase requirements, and completes the purchase after you review the terms and confirm. For supported products, agent mode reviews your AWS environment, proposes a deployment plan with the estimated monthly cost, and deploys after you approve. It then shows you how to start using the product and helps you troubleshoot if a step fails.

Agent mode is available in all AWS Regions where AWS Marketplace is available.

To get started, open agent mode on the AWS Marketplace website. To learn more, see Agent mode for AWS Marketplace in the AWS Marketplace Buyer Guide.

AWS Lambda durable functions now supports OpenTelemetry tracing

AWS Lambda announces OpenTelemetry (OTel) support for Lambda durable functions. You can now trace your durable executions and emit traces to Amazon CloudWatch and other industry-standard observability tools. This helps you find bottlenecks in your system, root cause failures faster, and understand how your workflows behave in production.\n With Lambda durable functions, you can build resilient, multi-step applications and AI workflows that run for up to one year, with built-in checkpointing and automatic retries. Durable functions can run for hours, or days, suspending and resuming across many Lambda invocations. The new OpenTelemetry plugin emits a unified trace that spans a full durable execution, including steps, waits, callbacks, and chained function invocations. You can use the trace to gain operational insights across your executions including step durations, retries, and failures across executions.  

The OpenTelemetry plugin is available for the TypeScript, Python, and Java SDKs. You can enable OpenTelemetry support by adding the OpenTelemetry plugin package alongside the Durable Execution SDK and register it on your durable handler or using an environment variable. OpenTelemetry support for Lambda durable functions is available in all AWS Regions that support Lambda durable functions.

To get started, see the  OpenTelemetry plugin documentation, the plugin interface documentation, and the sample application. To learn more, read the blog post.

Amazon Route 53 now supports health checks for private endpoints

Amazon Route 53 now offers Private Endpoint Health Checks, enabling business-critical workloads that never touch the public internet, such as internal APIs, databases, and self-hosted nodes behind internal load balancers or in private subnets, to use native health monitoring and automatic DNS failover. Customers can health-check any reachable private target by IP address, domain, or port over HTTP, HTTPS, or TCP. Customers can associate these health checks with Route 53 DNS records across routing policies in private hosted zones, so Route 53 routes traffic only to healthy endpoints and manages failover automatically. \n

Previously, teams built custom control loops using Amazon CloudWatch alarms, AWS Lambda functions, and the Route 53 API to achieve DNS failover for private resources. With this approach, failover took several minutes and added operational burden. Detection and DNS failover with Private Endpoint Health Checks are designed to complete in seconds rather than minutes, the service is fully managed, and nothing is exposed to the public internet. 

The feature supports on-premises and hybrid endpoints reachable from the VPC over AWS Direct Connect or AWS Site-to-Site VPN, enabling resilience patterns such as active-passive disaster recovery and cross-Region DR over AWS Transit Gateway or AWS Cloud WAN. Amazon Route 53 Private Endpoint Health Checks is available today in all AWS commercial regions. To learn more, visit the Amazon Route 53 Developer Guide and the Route 53 pricing page.

AWS Revenue Measurement now supports deal-level revenue attribution for AWS Marketplace offers and opportunities

Today, AWS announces deal-level revenue attribution for AWS Revenue Measurement (previously Partner Revenue Measurement) capabilities. AWS Partners can now see how the AWS revenue measured for their products maps to specific AWS Marketplace offers and/or APN Customer Engagements (ACE) opportunities. Until now, partners who implemented AWS Revenue Measurement via Resource Tagging, User Agent String or AWS Marketplace Metering could view aggregated, product-level attributed revenue. With this launch, partners can view attributed revenue by offers and opportunities on the Attributed Revenue dashboard in AWS Partner Central for components hosted in partners’ own AWS accounts. Two new capabilities make this possible: Revenue Attribution IDs and Marketplace Revenue Share.\n A Revenue Attribution ID is a unique identifier that allows partners to allocate a product’s measured AWS revenue to specific offers and opportunities. Partners create a Revenue Attribution ID, associate it with their deals, and enter a monthly attributed revenue allocation percentage for each deal. Partners with SaaS products on AWS Marketplace can automate allocation to offers with Marketplace Revenue Share. Each month, partners enter the percentage of the product’s attributed revenue contributed by AWS Marketplace deals in the Revenue Attribution page, and AWS then calculates the attributed revenue for each active offer. Partners can use Revenue Attribution IDs alongside Marketplace Revenue Share to adjust a calculated offer amount or attribute revenue to opportunities for future months. Both capabilities build on AWS Revenue Measurement, so partners can extend their current product-level implementations to gain deal-level insights.

Partners can manage Revenue Attribution IDs and Marketplace Revenue Share in AWS Partner Central or through the AWS Revenue Measurement and Marketplace Revenue Share APIs. AWS Revenue Measurement is generally available in all commercial Regions. To learn more, review the onboarding guide.

AWS Lambda now supports a 30-minute timeout for on-demand Event Functions

AWS Lambda now supports a 30-minute function timeout for Event Functions that run on demand, doubling the 15-minute limit for synchronous, asynchronous, and event source mapping (ESM) invocations. On Lambda Managed Instances, synchronous invocations also support up to 30 minutes, while asynchronous and ESM invocations keep the 90-minute timeout launched in September 2026. You can now run long-running workloads such as AI inference, media transcoding, and financial calculations for spiky or steady-state traffic without re-architecting your applications.\n Customers use Event Functions to build event-driven processors, API backends, and data pipelines. By default, Event Functions run on demand, and Lambda scales them automatically to handle spiky traffic. For steady-state traffic, you can also run them on Lambda Managed Instances, which runs functions on Amazon EC2 instances and maintains the operational simplicity of Lambda. For data-intensive workloads such as AI inference and media processing that run on demand and need longer continuous execution, the 15-minute function timeout had required customers to adopt architectural workarounds. With today’s launch, these workloads can run for up to 30 minutes, including when you use Lambda durable functions or invoke your functions from AWS Step Functions Standard workflows.

For Event Functions that run on demand, the 30-minute timeout is available in all commercial AWS Regions. On Lambda Managed Instances, it is available for synchronous invocations in all Regions where Lambda Managed Instances is available. You can set the function timeout by using the Lambda console, AWS CLI, AWS SDKs, AWS CloudFormation, AWS CDK, or AWS SAM. You can also use coding assistants such as Kiro, Claude Code, or Codex with the Agent Toolkit for AWS. Standard AWS Lambda pricing applies. To learn more, visit the AWS Lambda Developer Guide.

AWS Transform combines migration assessment and planning in one experience

AWS Transform for migrations now combines migration assessment and migration planning into a single migration job. Previously, you ran separate jobs to build a total cost of ownership (TCO) business case and to plan your migration. With this launch, you upload your on-premises inventory once and use it for both assessment and planning, in any order.\n After you upload your inventory data, you can estimate your AWS costs and build a business case, group your infrastructure into applications and plan migration waves, or do both. You can rerun either workflow as your data improves. For example, build a business case first and then plan waves from the same inventory, or plan your waves first and then estimate the cost of a single wave. Because assessment and planning share one inventory, your business case and migration plan stay consistent without moving data between jobs. After you finalize your plan and assessment, you can continue in the same job to migrate your networks, deploy landing zones, and migrate servers, storage, and containers.

This newest release is available in all AWS Regions where AWS Transform is offered. To learn more about migrating with AWS Transform, visit the product page and documentation.

AWS Transform extends application-centric migration across the lifecycle

AWS Transform now puts applications at the center of every stage of the migration journey, so you can discover, plan, migrate, validate, and track the applications your business runs on rather than just the servers underneath them. Application-level grouping has been part of AWS rehost workflows for years, and AWS Transform now carries that same view through your entire migration.\n Enterprise migrations typically involve thousands of servers, but the questions that drive decisions are about applications. Using discovery data, AWS Transform identifies application boundaries and dependencies, then surfaces logical groupings of applications, each with its own total cost of ownership (TCO) and migration tracking. AWS Transform also adds built-in surveys, letting application owners validate ownership, system boundaries, and business criticality. 

A new migration dashboard keeps everyone aligned as the migration moves forward. It gives your team, partners, and leadership one current view of your applications, inventory, assessments, wave plans, and migration status. The dashboard pulls in data from every job in a workspace and updates as work progresses, so no one has to build or maintain separate status reports. You can customize the dashboard and ask questions about migration progress in chat. 

With a more connected workflow and improved visibility across the migration, AWS Transform lets you move to the cloud with more confidence. This capability is available in all AWS Regions where AWS Transform is offered. To learn more about migrating with AWS Transform, visit the product page.

Amazon IVS Real-Time Streaming now provides a Linux broadcast SDK

The Amazon Interactive Video Service (Amazon IVS) Linux broadcast SDK is now available for IVS Real-Time Streaming. Applications running on Linux can join an IVS stage, receive audio from participants, and send audio back to the stage in real time. The SDK provides a C API and Go language bindings.\n The Linux broadcast SDK is ideal for AI applications that enrich live streams by processing audio, such as transcription, translation, and content moderation, as well as voice agents that converse with participants.

Amazon IVS is a managed live streaming solution designed to make low-latency or real-time video available to viewers around the world. Visit the AWS Region table for a full list of AWS Regions where the Amazon IVS console and APIs for control and creation of video streams are available.

To learn more, see IVS Broadcast SDK: Linux Guide.

AWS Elemental MediaTailor adds Dynamic Avail Sizing for live HLS

AWS Elemental MediaTailor now supports Dynamic Avail Sizing, which reduces the slate viewers see when ad breaks, also called avails, in live HLS streams go unfilled. MediaTailor is a server-side ad insertion and channel assembly service that personalizes ads in live and on-demand video streams.\n When the ad decision server returns fewer ads than a break allows, dynamic avail sizing omits the unfilled time and resumes programming early instead of showing slate. To keep playback continuous, MediaTailor holds back a configurable reserve of content behind the live edge and draws on it when a break ends early. When the reserve runs low, MediaTailor requests extra ad time in later breaks to refill it. Free ad-supported streaming TV (FAST) channel operators can use dynamic avail sizing to keep viewers watching programming instead of filler content.

Customers turn on dynamic avail sizing for each playback session by adding query parameters to the session initialization URL, with no change to their playback configuration. To learn more, see Dynamic avail sizing in the AWS Elemental MediaTailor User Guide.

Amazon Cognito now supports AWS Secrets Manager for app client secrets

Amazon Cognito now integrates with AWS Secrets Manager, so you can store and manage your user pool app client secrets in Secrets Manager. Amazon Cognito helps you implement secure sign-in and access control for users, AI agents, and microservices in minutes. An app client secret authenticates your application to Cognito. Previously, you had to manage and rotate these secrets in Cognito directly.\nWith this integration, you can leverage Secrets Manager for capabilities including rotation, access control with AWS Identity and Access Management (IAM) policies, and encryption with an AWS managed key or your own AWS Key Management Service (AWS KMS) customer managed key. This can help with periodic credential rotation or compliance requirements; this also helps with keeping secrets out of the console. You can enable this integration for either a new or an existing app client, and disable it at any time. This integration is available in all AWS Regions where Amazon Cognito and AWS Secrets Manager are available, except AWS European Sovereign Cloud (Germany). Standard AWS Secrets Manager charges apply. To learn more, see the Amazon Cognito Developer Guide for instructions.

AWS WAF launches new Security Analytics Dashboard for real-time investigations

AWS WAF now gives you one place to find and stop unwanted traffic across every application in your account. The new AWS WAF security analytics dashboard brings together traffic from all of your protection packs (web ACLs), whether they protect Regional resources or Amazon CloudFront distributions, into a single view. The dashboard also lets you search request logs right in the console, with no logging pipeline to build, and includes 21 days of full-fidelity request data at no additional cost. It is available alongside the existing dashboard for each protection pack (WebACL), which continues to work as it does today.\n Until now, spotting a trend across your applications, such as a bot campaign hitting several endpoints or a spike in blocked requests, meant opening each web ACL in turn or individually querying AWS WAF logs. With the AWS WAF security analytics dashboard, you can analyze traffic data from every protection pack (WebACL) at once. You can filter and group traffic by dimensions such as client IP address, country, ASN, user agent, JA4 fingerprint, host, path, rule triggered, and labels. You can see which endpoints are affected and how traffic is changing across all of them. The data is full-fidelity rather than sampled, and filters apply across the whole dashboard.

Investigating individual requests has also required delivering AWS WAF logs to Amazon CloudWatch Logs, Amazon S3, or Amazon Data Firehose and writing queries. AWS WAF Security analytics dashboard includes a logs widget that lets you search and explore raw request logs in the console, down to the full HTTP request, with no log delivery to configure. When you find unwanted traffic, you can block an IP address or a country directly from the page. You can also filter by the labels that managed rules attach to requests to see what a rule in Count mode would block before you enforce it.

The AWS WAF security analytics dashboard is available at no additional cost in all commercial AWS Regions where AWS WAF is available. This feature is not available in the AWS GovCloud (US) Regions and China Regions. To get started, open the AWS WAF console and choose Security analytics in the navigation pane. To learn more, see the AWS WAF Developer Guide and launch blog post.

AWS Partner Central now supports AI-assisted listing experience for product, services, and solutions

AWS Marketplace search, Agent Mode, and AWS sales teams use listings to recommend partner offerings to AWS customers. Partners who sell software and services can now create and improve their listings in AWS Partner Central with the new AI-assisted listing experience. It drafts content from a partner’s website and other materials, then shows which sections need clearer details, use cases, or evidence of customer outcomes. For single-AMI products, it also drafts the usage instructions buyers see after subscribing. Partners see, for each section, whether it meets what buyers look for, and choose which suggestions to use and review all content before publishing.\n Partners can position products and professional services as distinct solutions for different industries, use cases, or outcomes. Each solution gets its own AWS Marketplace page, bringing its case studies and AWS Specializations together with the listing’s pricing, reviews, and trust information, so buyers can evaluate and buy from the same page. Buyers can filter search results by Specialization. Partners not ready to share public pricing can publish without it using leads-only listings, through which interested customers request a private offer that reaches the partner as a lead in AWS Partner Central. Partners can test metering and billing setup before publishing, and ask the assistant metering questions while integrating.

These capabilities are available now to AWS Marketplace partners in all AWS Regions, except the AWS GovCloud (US) Regions and the China Regions. To get started, open AWS Partner Central and choose “Products and services” or “Solutions”. To learn more, see AI-assisted listing experience and Deciding what to list in the AWS Marketplace Seller Guide.

Aurora PostgreSQL with express configuration now supports Amazon VPC

Aurora PostgreSQL databases created with express configuration can now move into an Amazon Virtual Private Cloud (VPC) with a single API call or a modify action in the Amazon RDS console. This helps you take AI-built prototypes and other applications to production without migrating data, while meeting your security and compliance requirements. You can also create a new database cluster inside a VPC using express configuration.\n With express configuration, Aurora PostgreSQL lets you create and query a serverless database in seconds. Connectivity runs through an internet access gateway, which makes this a good fit for rapid prototyping and AI agent-driven development. When your application is ready for production, a single ModifyDBCluster API call moves the cluster into a VPC and turns off the internet access gateway. That gives you security groups, private subnets, and features such as Aurora Global Database, Amazon RDS Blue/Green Deployments, and Amazon RDS Proxy. The move requires no data migration, keeps your cluster endpoint the same, and interrupts connectivity only briefly. You can also reverse it with the same API call.

This feature is available in all AWS Regions where Aurora PostgreSQL with express configuration is available. To get started, use the Amazon RDS console, AWS CLI, or AWS SDKs. To learn more, see the Aurora User Guide.

Amazon Aurora now supports moving clusters to a different VPC

Aurora PostgreSQL and Aurora MySQL clusters can now move from one Amazon Virtual Private Cloud (VPC) to another without data migration. You can move a cluster with a single API call or a modify action in the Amazon RDS console. This lets you move a cluster to a VPC with more IP address space or a non-overlapping CIDR range, consolidate your network architecture, or meet changing security and compliance requirements.\n Previously, moving an Aurora cluster to a different VPC meant creating a new cluster in the target VPC, migrating your data, and updating your applications with new connection strings. Now, a single ModifyDBCluster API call moves your existing cluster to the new VPC. Your data stays in place, your cluster endpoint stays the same, and connectivity is interrupted only briefly during the move. Before you start, turn off features that are tied to the cluster’s current VPC, such as Aurora Global Database, Amazon RDS Proxy, and Amazon RDS Blue/Green Deployments. You can set them up again in the new VPC after the move.

This feature is available for Aurora PostgreSQL and Aurora MySQL in all AWS Regions where Aurora PostgreSQL with express configuration is available. To get started, use the Amazon RDS console, AWS CLI, or AWS SDKs. To learn more, see the Aurora User Guide.

Amazon MSK Connect now supports restarting existing connectors

Amazon MSK Connect (Amazon Managed Streaming for Apache Kafka Connect) now supports restarting both existing and newly created connectors and their tasks in place, enabling fast recovery for streaming data pipelines.\n Amazon MSK Connect is a fully managed service that allows you to deploy, monitor, and scale Apache Kafka Connect connectors with Amazon MSK (Managed Streaming for Apache Kafka). Previously, recovering from failures required deleting and recreating the connector, resulting in downtime and operational overhead. With this feature, you can restart your connector and all its tasks in place, or selectively restart only the failed tasks, making recovery seamless. You can also restart connectors that have no failed tasks to recover from transient issues or to pick up changes in external systems or dependencies.

This restart capability is available at no additional cost through the Amazon MSK Console, AWS CLI, AWS SDK, and AWS CDK in all AWS Regions where Amazon MSK Connect is available. To learn more, visit the Amazon MSK Connect documentation. To get started, visit the Amazon MSK Console.

Amazon Aurora DSQL now supports multi-Region peering across continents

Amazon Aurora DSQL now lets you create multi-Region clusters whose Regions are on different continents, such as US East (N. Virginia) and Europe (Ireland). Until now, the Regions in a multi-Region cluster had to be in the same geographic Region set.\n With intercontinental multi-Region clusters, applications on two continents read and write to one strongly consistent database. If a Region becomes unavailable, committed writes are preserved and applications keep reading and writing in the other Region. You choose the two peer Regions and the witness Region. Place your database close to users on each continent, or separate failure domains as widely as your resilience needs require.

Intercontinental multi-Region clusters are available in all AWS Regions where Aurora DSQL supports multi-Region clusters. To learn more, see Multi-Region cluster availability in the Aurora DSQL User Guide.

AWS Marketplace now supports buyer contact sharing at subscription

AWS Marketplace now supports buyer contact sharing. When customers subscribe to a third-party product, they can share their contact information with the seller. This helps sellers onboard new customers, send deployment guides and setup instructions, provide technical support, and share product updates from the start of the subscription.\n Customers share their first name, last name, email address, company name, and country or region in the deployment contact information section of the AWS Marketplace procurement page. AWS Marketplace pre-fills this section from the customer’s previous purchases, AWS Profile, or AWS account details where available, so customers don’t need to enter it again for each subscription. Customers can complete or edit any field before they subscribe, and view what they shared on the agreement details page in the AWS Marketplace console.

Sellers can view the customer’s contact information on the agreement details page in AWS Partner Central, or retrieve it with the new GetAgreementContacts API to bring it into their onboarding and support workflows. Sellers of software as a service (SaaS) products can call the API to pre-fill their registration page, so customers confirm their details instead of entering them again. This capability is available in all commercial AWS Regions where AWS Marketplace is available.

To learn more, see Sharing your contact information at subscription in the AWS Marketplace Buyer Guide and Receiving buyer contact information in the AWS Marketplace Seller Guide.

AWS Systems Manager Parameter Store raises API throughput limits

AWS Systems Manager Parameter Store has increased the default throughput limits for five API operations. For accounts that have turned on the higher throughput setting, the quotas have changed as follows, in transactions per second (TPS):\n

GetParametersByPath (from 100 to 500 TPS)

PutParameter (10 to 50)

DeleteParameter (5 to 50)

DescribeParameters (10 to 100)

GetParameterHistory (10 to 100)

Parameter Store provides powerful, hierarchical storage for configuration data. The higher quotas give more headroom to workloads that call these operations in bursts. Examples include applications that load configuration by path when many AWS Lambda functions start at once, infrastructure as code deployments that create or update many parameters, and cleanup jobs that bulk-delete stale parameters.

AWS Revenue Measurement now supports closed-source models in Amazon Bedrock

AWS Revenue Measurement (previously Partner Revenue Measurement) provides measurement of AWS consumption driven by Partner solutions. Today, AWS announces expanded Amazon Bedrock support in AWS Revenue Measurement. Partners can now gain attributed revenue visibility for their solutions’ use of closed-source models in Amazon Bedrock, giving them a more complete view of the revenue their solutions drive.\n Until now, only Amazon models (such as Amazon Nova) and open-weight models in Amazon Bedrock were included in AWS Revenue Measurement. With this update, attributed revenue for Amazon Bedrock reflects the total Bedrock charges, including closed-source model usage. Partners who already use Resource Tagging, User Agent string, Identity Tagging, or SaaS Account Tagging for Amazon Bedrock automatically benefit from the expanded coverage, with no additional implementation needed.

AWS Revenue Measurement is generally available in all commercial regions. To learn more, review the Amazon Bedrock best practices in the AWS Revenue Measurement onboarding guide.

Amazon Connect Customer enables Cases by default for new instances

Amazon Connect Customer enables Cases by default for new instances, so operations teams can start tracking and resolving customer issues on day one without any setup work. New instances come with Cases connected to rules for setting SLAs, Customer Profiles for personalizing customer experiences, and Amazon EventBridge for triggering actions in other business applications when a case is updated.\n Administrators of existing instances can enable these integrations with one click in the AWS Management Console. Newly added case fields sync automatically, so admins don’t have to update integrations each time they create fields.

Cases is available in the following AWS Regions: US East (N. Virginia), US West (Oregon), Canada (Central), Europe (Frankfurt), Europe (London), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Asia Pacific (Tokyo), and Africa (Cape Town). To learn more and get started, visit the Cases webpage and documentation.

AWS Transform supports optimized storage for server migration

AWS Transform for migrations now recommends the optimized storage type and configuration for every server you migrate, so you no longer do it server by server. Building on recent storage assessment capabilities, AWS Transform’s recommendations use each server’s observed performance and capacity data to reduce overprovisioning costs and avoid performance issues after cutover. You see the expected cost before you migrate, and AWS Transform applies the approved settings automatically at launch.\n Recommendations cover Amazon EBS and Amazon FSx for NetApp ONTAP. For FSx for NetApp ONTAP, they include the file system layout and sizing. You choose whether to optimize for cost or for performance, and recommendations stay consistent from assessment through launch.

This capability is available in all AWS Regions where AWS Transform for migrations is available.

To learn more, see the AWS Transform user guide.

AWS Revenue Measurement now supports Identity Tagging

AWS Revenue Measurement (previously Partner Revenue Measurement) provides measurement of AWS consumption driven by Partner solutions. Today, AWS announces Identity Tagging in AWS Revenue Measurement, a new way for Partners to get attributed revenue visibility for the AWS usage their solutions drive. Partners can now tag the IAM user or role their solution uses instead of tagging individual resources or decorating individual SDK/API calls. This makes attribution simpler and needs no code changes.\n Until now, Partners using Resource Tagging had to apply tags to each resource their solution created or managed. That could be hard to keep consistent across customer environments. With Identity Tagging, Partners apply a tag to the IAM role or user their solution uses in the AWS account. AWS Revenue Measurement then attributes usage from resources created by that identity to the Partner’s product. Identity Tagging works alongside existing methods such as Resource Tagging, User Agent string, and SaaS Account Tagging, so Partners can pick the method that best fits their architecture.

AWS Revenue Measurement is generally available in all commercial regions. To learn more, review the AWS Revenue Measurement onboarding guide.

Amazon EC2 launches performance status checks and new utilization metrics for EC2 Detailed Monitoring

Customers can now use Amazon EC2 performance status checks and new utilization metrics for EC2 Detailed Monitoring to observe and act on the performance of their EC2 workloads without setup within the instance, complementing their in-guest tooling. Customers running performance-sensitive EC2 workloads can use these signals to monitor their CPU and GPU resources, tune capacity needs, and troubleshoot application issues.\n EC2 performance status checks identify nonterminal hardware conditions that impact performance on accelerated computing instances. These checks also recommend an action, such as rebooting to complete GPU memory row remapping without replacing the instance. The additions to Detailed Monitoring include support for OpenTelemetry metrics, one-minute GPU metrics that help teams tune accelerated workloads, and a new, higher resolution 10-second CPU utilization metric that helps customers size and scale compute capacity using EC2 Application Auto Scaling. Customers can selectively opt into these new metrics using categories that they can easily apply to their instances across their fleet. Together, these new capabilities make it easier for customers to manage their compute resources.

EC2 performance status checks are available at no additional charge. EC2 Detailed Monitoring metrics are opt-in and incur standard CloudWatch charges for EC2 Detailed Monitoring. These capabilities are available in all commercial AWS Regions and the AWS GovCloud (US) Regions.

To learn more, see EC2 Detailed Monitoring documentation and EC2 performance status checks documentation.

AWS Marketplace now supports buyer approval workflows

Today, AWS Marketplace announces the general availability of buyer approvals. This lets you decide which AWS Marketplace purchases need sign-off, who approves them, and where approval happens. You can define buying policies for the conditions that matter to you, such as purchase amount, product type, pricing, and whether an offer is public or private. Additionally, you can choose to route each purchase request to the right place: the AWS Marketplace console, a chain of approvers by email, or your procurement system in Coupa or SAP Ariba.\nWhen you route approvals by email, procurement, finance, legal, and cloud governance teams can approve requests without needing to sign into the AWS console. They see the product, price, terms, and can approve right from their email. For example, you can route large software commitments to Coupa or SAP Ariba for review, send mid-size purchases to multiple approvers by email, and let low-cost or pay-as-you-go purchases go through without approval. When a purchase is routed to your procurement system, it follows your existing approval flows and gets a dedicated purchase order for cleaner spend tracking and reconciliation. With AWS Marketplace, customers can already choose which subset of available products their teams can buy, track each purchase with purchase orders, and require their employees to enter a purchase order when buying. Now, with buyer approvals, you can also control how each purchase is approved before making a financial commitment. Customers can now create buying policies and approvals in all AWS Marketplace commercial Regions. To get started, see the AWS Marketplace Buyer Guide.

Amazon WorkSpaces now lets developers enable Nested Virtualization

Developers on Amazon WorkSpaces Personal can now turn on Nested Virtualization for their own WorkSpaces from the Windows WorkSpaces client, without asking an administrator to make each change. Once an administrator grants self-service permission for the directory, developers can run Docker Desktop, Windows Subsystem for Linux (WSL2), and Android emulators directly on their WorkSpaces.\n Administrators can now grant a directory-level self-service permission for Nested Virtualization. Once granted, end users can turn the setting on or off for their own Windows WorkSpace from the WorkSpaces client. Administrators keep control because the permission is set per directory.

Self-service is supported on WorkSpaces Personal, using the Desktop Cloud Visualization (DCV) protocol, on Standard bundles and higher. Power (4 vCPU) or higher is recommended. Value and Graphics bundles are not supported. Nested virtualization is available in all AWS Regions where Amazon WorkSpaces is available, excluding the China (Ningxia) Region. There is no additional charge for supported bundles. For more information, visit the Amazon WorkSpaces documentation on Nested Virtualization. For pricing, visit the Amazon WorkSpaces pricing page.

Amazon Aurora DSQL now supports PL/pgSQL functions and procedures

Amazon Aurora DSQL now supports PL/pgSQL, enabling developers to create and execute stored functions and procedures directly in the database. Amazon Aurora DSQL is a serverless, distributed SQL database designed for applications requiring active-active availability across multiple AWS Regions with fast reads and writes. You can now encapsulate procedural programming constructs such as IF/ELSE statements, loops, and variable assignments alongside SQL operations within the database.\nPL/pgSQL support reduces network round trips and improves application performance for data validation workflows, complex calculations, and transactional operations requiring multiple steps. Financial applications can implement account transfer procedures that validate balances, update multiple tables, and handle error conditions within a single database function. PL/pgSQL support is available in all AWS Regions where Aurora DSQL is available. To learn more, see PL/pgSQL in the Aurora DSQL User Guide.

Announcing Data as a Service for AWS Marketplace

Today, AWS Marketplace announces the general availability of Data as a Service, a new product type that lets customers buy third-party datasets and licensed content. Customers often have the models and compute they need, but buying third-party data means adding each provider as a new supplier, negotiating separate contracts, and waiting on procurement cycles. Customers can now find and subscribe to Data as a Service products from AWS Marketplace sellers, and use them to enrich artificial intelligence (AI) applications, analytics, and other workloads.\nCustomers purchase Data as a Service products through public or private offers, and charges appear on their existing AWS invoice. For example, an investment team evaluating private companies can connect its AI agents to private-market data through an API and to credit analysis through a Model Context Protocol (MCP) server. Sellers set their own prices and license terms, using a contract, pay-as-you-go, or contract with consumption pricing model for each product. They can add a free trial to any of these models, or offer a product for free. They also define the onboarding and delivery experience, from setting up the customer’s account to providing the data through an API, a web portal, a real-time feed, an Amazon Simple Storage Service (Amazon S3) bucket, or another approach. To learn more, explore Data as a Service products on AWS Marketplace, the AWS Marketplace Buyer Guide, and the AWS Marketplace Seller Guide.

AWS Partner Central Agents now guide higher-quality co-sell submissions and explain co-sell motions

Today, AWS announces that AWS Partner Central Agents now guide AWS Partners to create and enrich opportunities for a high-quality submission, and explain the assigned co-sell motion of each opportunity and why AWS assigned it. Since the launch of Agentic co-selling on June 16, 2026, AWS Partner Central Agents score every opportunity with an Opportunity Quality Score and assign co-sell motions including AWS Field-engaged, Agent-engaged, or Partner-led.\n When partners create an opportunity, the agents now guide them to a high-quality submission by asking for the details the Opportunity Quality Score measures, such as the customer’s business problem, dated next steps, the Partner Primary Need from AWS, and the AWS services the customer will adopt, based on the co-selling best-practice guidance. After submission, the agents explain the reason behind each opportunity’s co-sell motion and why AWS assigned it. Where the partner can act, the agents recommend the next step: for Agent-engaged opportunities, how to raise the Opportunity Quality Score; for Partner-led opportunities where AWS co-sell support was not requested, how to progress using agents. As partners act on these recommendations, the score and motion recalculate, and Agent-engaged opportunities move closer to AWS Field-engaged.

The new enhanced experience is available today to AWS Partners in the US East (N. Virginia) Region. Partners can also use the agentic experience in native AI tools like AWS Quick and Kiro, or through MCP in their own CRM. To get started, partners sign in to AWS Partner Central and open opportunity management. To learn more, see Agentic Co-sell engagement and the AWS Partner Central Agents MCP guide.

AWS Marketplace now offers Express Checkout for buyers without an AWS account

Today, AWS announces AWS Marketplace Express Checkout, a simpler way for businesses that don’t have an existing AWS account to buy software from established vendors in AWS Marketplace.\n Express Checkout is built for businesses that don’t use AWS today and just want to buy software. With no cloud setup required, buyers sign in with an email address or a social login, pay with a credit or debit card, and check out in a few steps. Businesses that buy multiple applications this way can manage all of their Express Checkout subscriptions in one place, without a separate billing login for each vendor. From there, they can update their card, download invoices, or cancel a subscription themselves.

For sellers, Express Checkout opens a new segment of self-serve business buyers, using the same listing and AWS Marketplace integration they use today. Sellers with an eligible public SaaS listing can request to add their product to the Express Checkout catalog.

Express Checkout is available for eligible SaaS products to buyers in the United States and Canada. Businesses that already use AWS can continue to buy in AWS Marketplace with their AWS account.

To get started, visit the AWS Marketplace Express Checkout page. To learn more, see the  AWS Marketplace Buyer Guide and the  AWS Marketplace Seller Guide.

AWS Transform automates source and migration target actions to accelerate cutover

AWS Transform now automates the tasks that prepare and validate your servers during a migration, before they move and after they launch on AWS. AWS Transform is the agentic AI service that speeds up cloud migration, application modernization, and continuous tech debt reduction. With the new source actions capability, you can create actions and run them securely on your source servers at the right point in the migration, from the early stages through cutover. Post-launch actions complement this by automatically configuring and validating each workload as soon as it launches on AWS.\n Cutover preparation and validation are often manual, time-consuming tasks in large-scale migrations. Traditionally, migration teams log in to each source server to stop services and freeze database writes before shutdown, and then check connectivity, application health, and monitoring on each server after it launches.

With source actions in AWS Transform, you define each action once, run it in bulk on the servers and waves you choose, and reuse it across accounts and migrations. AWS Transform manages these actions centrally, so you can reuse them across accounts and migrations instead of rebuilding them each time. You can monitor the status of each action and confirm your go/no-go decision before moving to the next step. Source preparation and post-launch actions are now part of the migration plan itself, so you no longer have to coordinate and run these activities by hand.

This capability is available in all AWS Regions where AWS Transform for migrations is available. To get started, see the AWS Transform documentation or visit the AWS Transform product page.

AWS Health now supports resource and account tags

AWS Health now supports resource and account tags in Health events. This new capability enables you to enrich Health events with resource tags and account tags, providing additional context to help you manage and respond to events more effectively.\n With this feature, you can now view resource tags directly within Health events for supported AWS services. This allows you to quickly identify which resources are affected and understand their business context without needing to navigate to individual service consoles. Additionally, account-level tags help you categorize and filter events across your organization, making it easier to route notifications and automate responses based on your organizational structure.

This enhancement is particularly valuable for organizations managing large-scale AWS deployments. By leveraging tags in Health events, you can streamline incident response workflows, improve cost allocation tracking for health-related events, and enable more granular event filtering in AWS Health Dashboard and AWS Health API responses.

The feature integrates seamlessly with Amazon EventBridge, allowing you to create event-driven automation that responds to tagged resources and accounts. This enables sophisticated routing and handling of Health events based on tag values, supporting use cases such as automatically creating tickets in your IT service management system or sending targeted notifications to specific teams.

To enable tag enrichment, visit the AWS Health Dashboard or view the AWS Health User Guide for more information.

AWS Revenue Measurement now supports Account Tagging for SaaS Products

AWS Revenue Measurement (previously Partner Revenue Measurement) provides measurement of AWS consumption driven by Partner solutions. Today, AWS announces Account Tagging for SaaS Products for AWS Revenue Measurement. Partners can now measure the AWS revenue impact of their SaaS solutions by tagging the AWS accounts that host them, instead of tagging individual resources.\n Until now, Partners with SaaS listings measured that impact through Resource Tagging, User Agent strings, or AWS Marketplace Metering. With Account Tagging for SaaS products, a Partner uses AWS Organizations to tag partner-owned member accounts dedicated to a single SaaS product. The tag key is aws-apn-id- and the tag value is pc:. AWS Revenue Measurement then attributes all eligible AWS consumption in each tagged account to that product.

Account Tagging for SaaS Products is available for products with a SaaS listing on AWS Marketplace. Each tagged account must be dedicated to one product. Partners whose products share an account can continue to use Resource Tagging, User Agent string, or Identity Tagging. AWS Revenue Measurement is generally available in all commercial regions. To learn more about implementing AWS Revenue Measurement, review the onboarding guide for more information.

AWS Partner Central Agents now generate Value Realization Plans

Today, AWS announces that AWS Partner Central Agents can now generate Value Realization Plans (VRPs) for the Business Value Realization (BVR) motion in AWS Partner Central. The VRP is the plan of record for a BVR engagement. The agentic experience generates a VRP that includes the customer’s business goals; plan for adoption, change management, and governance; and baseline key performance indicators (KPIs) for Partners to review and finalize.\n AWS Partner Central launched the BVR motion on June 16, 2026, to help Partners drive business outcomes for their customers by structuring AWS service adoption into defined stages, with funding tied to demonstrated value. With this launch, Partners in the BVR motion no longer start a VRP from a blank page. The agents recommend KPIs, ramp targets, and baseline measurements based on nomination data, opportunity context, and service benchmarks. They also flag gaps before submission, such as a lack of customer sponsorship, and identify next steps. Partners start from a complete draft and refine it with their customer.

BVR is available in AWS Partner Central for consulting, system integrator, and managed services partners with Advanced or Premier tier status and a qualifying domain competency. This capability is available in the US East (N. Virginia) Region.

To get started, partners sign in to AWS Partner Central and choose Funding benefits, Business Value Realization. After AWS approves a nomination, partners ask AWS Partner Central Agents, “Help me draft a Value Realization Plan for my BVR engagement.” To learn more, see Business Value Realization.

AWS License Manager now supports AWS Managed Microsoft AD (Hybrid Edition)

AWS License Manager now supports AWS Managed Microsoft AD (Hybrid Edition), so you can use your existing Microsoft Active Directory (AD) to manage user-based subscriptions for Microsoft Office, Visual Studio, and Remote Desktop Services Subscriber Access License (RDS SAL). Hybrid Edition, a feature of AWS Directory Service for Microsoft Active Directory, extends your existing directory into AWS. License Manager now works with that directory directly, so you no longer need to bear the cost of duplicating and maintaining the directory in AWS for licensing.\n Previously, onboarding Microsoft user-based subscriptions required a trust relationship between your directory and a second directory in AWS. If your security policies didn’t allow that trust, you couldn’t onboard without changing those policies. Now, license-included Microsoft Office and Visual Studio instances that you launch on Amazon Elastic Compute Cloud (Amazon EC2) join your existing directory and apply the same Group Policy settings, such as security baselines and audit policies, so you don’t need to recreate them in a second directory.

This feature is available in all AWS Regions where AWS License Manager user-based subscriptions and AWS Managed Microsoft AD (Hybrid Edition) are available. To learn more, see the AWS License Manager User Guide, or get started in the AWS License Manager console.

Amazon Connect Customer now provides AI chat summaries in the CCP

Amazon Connect Customer now provides agents with a generative AI-powered post-contact summary of each chat conversation inside the Contact Control Panel (CCP) within seconds after the chat ends, so agents can complete after contact work from the summary instead of rereading the transcript. The summary covers what the customer needed, the actions the agent took, and any open items, and it appears at the end of the chat transcript. Conversational analytics already generates these summaries for voice and email contacts in the CCP, and this launch adds chat.\n An airline agent who finishes a flight rebooking chat sees the new travel date, the fare difference charged, and the confirmation sent, then writes the wrap-up note from those three facts. Summaries follow your flow’s redaction settings and use the same security profile permissions as voice summaries, so administrators who already enabled voice summaries can turn on chat analytics with post-contact summary in the Set recording and analytics behavior block. To learn more, see View generative AI-powered post-contact summaries in Connect Customer.

Amazon Polly now supports voice cloning

Today, we are excited to announce the general availability of voice cloning in Amazon Polly, as a gated release. Using a few minutes of recordings from a consenting speaker, you can now create a cloned voice that sounds like that speaker. The cloned voices run on the Amazon Polly Generative engine, which delivers similar naturalness and accuracy as the rest of the generative voices. Voice cloning launches in seven languages: English, Spanish, German, French, Italian, Portuguese, and Hindi. Each cloned voice speaks the language it was recorded in. The cloned voices are automatically encrypted and access-controlled: only you can access your cloned voices.\nAmazon Polly is a managed service that turns text into natural-sounding speech. With voice cloning, you can give your brand one recognizable voice across IVR and contact centers, voice agents, mobile apps, advertising, and connected devices. You can also create multiple voices for content creation use cases such as audiobooks, e-learning, podcasts, videos, and game characters. Amazon Polly does not retain a copy of your voice recordings once the voice is created and doesn’t use them to train Polly models. Voice cloning is available in the following AWS Regions: US East (N. Virginia), US West (Oregon), Europe (Frankfurt), Europe (Zurich), Europe (London), Asia Pacific (Sydney), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Tokyo), Canada (Central), and AWS GovCloud (US-West). Voice cloning is already enabled for select Amazon Polly customers. To request access, contact your AWS account team or apply here. Building a cloned voice is free for Amazon Polly customers, and you pay for the speech you synthesize, on a pay-as-you-go basis. For more details, see the Amazon Polly documentation, and Amazon Polly Pricing page. For speech AI use cases, you can build with Amazon Polly, including voice cloning. Check out the Amazon Polly GitHub repo.

AWS Lambda Durable Execution SDK for Rust is now generally available

Today, AWS announces the general availability of the open source AWS Lambda Durable Execution SDK for Rust. Rust developers can use the SDK to build resilient, long-running workflows using Lambda durable functions, to create multi-step applications like infrastructure automation, deployment rollouts with approval gates, and pipelines that call AI models and pause for human review directly in their Rust code without implementing custom progress tracking.\nLambda durable functions extend Lambda’s event-driven programming model with operations that checkpoint progress automatically and pause execution for up to one year when waiting on external events. With the SDK, developers write a durable function as an ordinary async Rust function that awaits each durable operation like any other future and can run independent operations concurrently. The SDK includes steps that save their results, callbacks for human-in-the-loop and agent-in-the-loop workflows, invocations of other durable functions, and timed waits. The SDK works with the Rust tools developers already use. Developers add the aws-durable-execution-sdk crate with cargo add, build with Cargo Lambda, and deploy a single binary on the provided.al2023 runtime. Before deploying, they can test complete workflows with cargo test and the SDK’s LocalRunner without requiring network access or AWS credentials. To get started, see the Lambda durable functions developer guide, the Rust SDK guide, and the AWS Lambda Durable Execution SDK for Rust on GitHub. For Regional availability and pricing details for Lambda durable functions, see the AWS Capabilities by Region page and AWS Lambda Pricing.

Amazon VPC Flow Logs now supports Amazon EKS metadata

Amazon VPC Flow Logs now adds Amazon Elastic Kubernetes Service (Amazon EKS) metadata to flow log records. Each record identifies the pod that sent or received the traffic, along with its cluster, namespace and labels. This makes it faster to investigate security events, understand traffic patterns, and troubleshoot network issues in your EKS clusters, without running monitoring software on your nodes.\nAmazon EKS runs pods on Amazon EC2 instances, called nodes, and many pods can share one node’s network interfaces. Previously, flow log records showed the network interface and IP address for each flow, not the pod behind it. Pods are short-lived, so by the time you investigated unexpected traffic, that IP address could belong to a different pod. To trace traffic back to the right pod, teams built custom tools or ran monitoring software on every node. Now, each record names the pod that sent or received the traffic, so one log gives you both VPC and Kubernetes context. For example, if a pod sends unexpected traffic to a database in your VPC, the record shows the pod and its namespace, even after that pod is replaced. To get started, create a flow log for your cluster’s VPC or subnets, and add the EKS fields to the log format. You don’t need to change your cluster or applications. The fields are available for pods on Amazon EC2 instances, including EKS Auto Mode, managed node groups and self-managed nodes, that use the Amazon VPC CNI plugin for Kubernetes. Logs go to the same destinations as before, so your existing analytics tools keep working. EKS metadata is available at no additional charge. Standard VPC Flow Logs charges for log delivery and storage still apply. This feature is supported in all AWS Regions where Amazon EKS is available. For more information, visit the Amazon VPC Flow Logs.

Amazon SageMaker AI now detects rising response times, idle GPUs, and missed optimizations on your generative AI inference endpoints

Amazon SageMaker AI now monitors the response time and GPU use of generative AI inference endpoints by default and notifies customers when responses slow down, GPUs sit mostly idle, or a configuration falls behind. Customers run GenAI Inference Recommendations to test configurations designed to restore response time or lower cost on their recent traffic, and they see measured results before anything changes in production.\n The best configuration for generative AI inference endpoints changes after deployment, as traffic shifts and new serving techniques become available. An endpoint whose latency rises or whose GPUs sit idle keeps running until users complain or the bill arrives.

With this launch, SageMaker AI tracks response time and GPU capacity use and checks each endpoint’s configuration. It does not read customer requests or responses. Findings include slower response times, unused GPU capacity, and configuration gaps such as an older serving container, missing speculative decoding, or missing GPU kernel tuning. When a change holds across the past week or a configuration gap is found, customers receive one AWS Health notification per endpoint in the AWS Management Console.

Each notification links to the SageMaker AI Agentic Experience in SageMaker Studio with the endpoint and recent traffic already filled in. Customers run Generative AI Inference Recommendations, which compares the current configuration and options on response time, traffic handled, and cost per million tokens. Tests leave the production endpoint untouched, and nothing changes until the customer applies a result. Customers can turn off monitoring anytime.

This capability is available in US East (N. Virginia, Ohio), US West (Oregon), Europe (Ireland, Frankfurt), Asia Pacific (Singapore, Tokyo) with no additional charge for monitoring. Standard compute costs apply for recommendation jobs. To learn more, visit the blog post or the documentation.

Amazon Inspector Code Security adds GitHub Enterprise Server and private connectivity

Amazon Inspector Code Security now supports GitHub Enterprise Server (GHES) integration and private network connectivity for GHES and GitLab Self-Managed. Code Security is also launching automatic fix pull requests in public preview. Organizations that host GHES or GitLab Self-Managed in private networks can now connect through Amazon Virtual Private Cloud (Amazon VPC) configurations or AWS CodeConnections, so they can scan repositories without opening their firewalls to the entire AWS IP address range.\n Amazon Inspector Code Security scans first-party application source code, third-party dependencies, and Infrastructure as Code for vulnerabilities in repositories on GitHub, GHES, and GitLab Self-Managed. With automatic fix pull requests, Amazon Inspector turns its suggested fix into a pull request that developers review and choose whether to merge, instead of applying the fix by hand. It opens pull requests only for high-confidence fixes to static application security testing (SAST) findings at or above a severity you set.

These features are available in all 10 AWS Regions where Amazon Inspector Code Security is available: US East (N. Virginia), US East (Ohio), US West (Oregon), Europe (Frankfurt), Europe (Ireland), Europe (London), Europe (Stockholm), Asia Pacific (Tokyo), Asia Pacific (Sydney), and Asia Pacific (Singapore). To learn more, see the Amazon Inspector User Guide and the Amazon Inspector product page.

Amazon S3 Tables now support promoting replicas to writable tables

Amazon S3 Tables replication now lets you make a read-only replica of your Apache Iceberg table a standalone, writable table with a single API call. S3 Tables replication keeps read-only replica tables in sync with your source tables across table buckets, AWS Regions, and accounts. Instead of copying a replica into a new table, you can promote it in place to move writes to another Region.\n When you disassociate a replica from replication, the standalone table keeps its namespace and table name, so your queries and pipeline code carry over when you point your applications at it. The promoted table then works like an S3 table you created, and you can use it as a source for new replicas.

This capability is available in all AWS Regions where S3 Tables replication is supported. You can use the AWS Management Console, AWS CLI, AWS SDKs, or S3 Tables API. For pricing details, visit the Amazon S3 pricing page. To learn more, see the S3 Tables overview page and documentation.

AWS AppConfig launches support for managing native AI model and AI prompt configuration

Today, AWS AppConfig introduces two new configuration profile types to help you manage the settings for your generative AI applications: AI model invocation configurations, and AI prompt configurations. You can now change model settings and prompts while your application is running, without redeploying code. These updates use AWS AppConfig’s safe deployment practices, including gradual rollouts and automatic rollback. This can allow you to move faster and safer, and avoid unexpected behaviors and tokenomics results with rapidly evolving AI solutions.\n AI model invocation configurations store the settings your application uses to call an AI model. These include the model identifier, inference parameters such as temperature and maximum tokens, and messages such as system prompts and example conversation turns. Provider-specific options are managed in custom parameters, so the same schema works across AI providers. Similarly, AI prompt configurations store prompts as separately versioned artifacts. You can update  and optimize a prompt on its own,without changing the model configuration, and reuse it across multiple applications. This allows for powerful prompt engineering.

You can roll out a change to these gradually, for example to 10%, then 20%, then 30% of your traffic. Along the way, you can measure the change’s effect on operational metrics, customer experience, and token usage. If an alarm in Amazon CloudWatch, Datadog, New Relic, or another observability solution goes off during the rollout, AWS AppConfig can automatically roll the change back to the previous state.

AWS CloudFormation StackSets adds progress visibility to simplify tracking deployments at scale

AWS CloudFormation StackSets now shows stage-by-stage progress for deployments across accounts and AWS Regions, whether you target tens of accounts or thousands. You can follow each deployment as it progresses and take action when needed, without waiting for it to finish. StackSets lets you consistently apply the same infrastructure or application components, such as networking and security settings, across many accounts and Regions in one operation. With this launch, you can see how many stack instances are deployed, how many remain, and which ones need attention. This gives you confidence that a rollout is on track at every stage.\n Stage-level progress gives you a clear, detailed view of where each deployment stands. Previously, a StackSets deployment showed an overall status while it ran. Now, StackSets also reports progress for two stages. The first is role provisioning, which prepares access in each target account of a service-managed stack set. The second is stack deployment, which deploys your template as a stack instance in each account and Region. For example, in a deployment to 2,000 accounts in 3 Regions, you can see that role provisioning is complete and stack deployment has reached 40 percent, with 2,400 stack instances completed and 3,600 pending. Progress for each Region and the result of each stage remain available after the deployment completes.

To get started, open the Progress view for a stack set operation in the CloudFormation console, or use the AWS CLI or AWS SDKs. Existing integrations continue to work without changes. Visit the StackSets documentation to learn more.

Deployment progress is available at no additional charge in all AWS Regions where CloudFormation StackSets is supported. Refer to the AWS Region table for service availability details.

AWS Blogs

AWS Japan Blog (Japanese)

AWS Architecture Blog

AWS Cloud Operations Blog

AWS Big Data Blog

AWS Compute Blog

Containers

AWS Database Blog

AWS Developer Tools Blog

AWS DevOps & Developer Productivity Blog

AWS for Industries

Artificial Intelligence

AWS for M&E Blog

Networking & Content Delivery

AWS Security Blog

Open Source Project

AWS CLI

Karpenter

Amazon EKS Anywhere