All news in category “Vendor and Hyperscaler Watch”

🚀 Amazon SageMaker HyperPod now supports model caching to speed up LLM inference by pre-loading model weights and container images onto cluster nodes. This reduces cold-start delays for scale-out events, allowing pods to start in seconds rather than minutes. The feature includes a weights cache on local NVMe and an image cache that pre-pulls container images, with automatic fallback to the original source if a cache is unavailable. Model caching is GA across all SageMaker HyperPod regions and is enabled via the HyperPod Inference Operator.

🔔 Microsoft is changing the destination addresses for two widely used services: Teams web users are being redirected to teams.cloud.microsoft and M365/Copilot web users to copilot.cloud.microsoft. Organizations should update network controls—client devices, proxies, firewalls, and secure web gateways—to ensure continued access and follow Microsoft’s recommended network requirements. Redirects are expected to complete by early October, with limited Teams exceptions until Dec. 31, 2026.

📢 AWS Lambda now supports configurable direct reads for Amazon S3 Files, letting you choose whether functions read from S3 Files high-performance storage or directly from your S3 bucket. This setting enables optimizing throughput and latency per application by routing large reads (≥1 MB) directly from the bucket for maximum throughput when enabled, or serving all reads from high-performance storage for lowest latency when disabled. The feature is available across most AWS commercial and GovCloud regions and can be configured via Console, CLI, SDKs, or CloudFormation with no additional charge beyond standard pricing.

🛠️ AWS DevOps Agent now enables engineers to manage production operations and the full investigation lifecycle directly within Slack. Teams can @mention the agent in connected private channels to initiate and steer investigations, with findings, recommended actions, and team context captured in a single thread. This reduces context switching during high-severity incidents and is available in all commercial AWS Regions where the agent is supported.

📊 AWS HealthOmics now publishes real-time run metrics to Amazon CloudWatch, providing live visibility into workflow resource utilization as runs execute. The 14 new metrics cover CPU and GPU usage, memory, file system usage and I/O, network throughput, and ephemeral storage. Emitted using the Amazon CloudWatch OpenTelemetry standard, these metrics support native dashboards, alarms, and integration with third-party observability tools. Real-time metrics are available in multiple US, Europe, and Asia Pacific Regions, and CloudWatch ingestion charges apply.

🔒 Cloudflare introduces automatic remediation policies for Cloudflare CASB, enabling security teams to define event-driven responses that revoke risky file shares or dispatch custom webhooks without manual steps. CASB previously provided visibility into SaaS misconfigurations; the new feature extends that visibility with native, automated remediation executed via Cloudflare One. Policies run on the Cloudflare developer platform and Workflows to ensure durable, rate-limit aware execution and deliver completion logs for audit and compliance.

📰 ServiceNow faces major shifts as AI and cybersecurity converge, altering its traditional seat-based SaaS model. The company’s Armis acquisition and expansion into consumed services signal a pivot toward consumption pricing, infrastructure, and AI token usage. This strategy aims to defend its dominant position in large enterprises while responding to threats like seat compression and competition from cloud platforms. Buyers should watch for complex contracting and potential cost surprises.

🛠️ Microsoft patched a bug that blocked Teams and the new Outlook from launching on ARM-based Windows 11 devices after August 2026 updates. The issue affected devices such as Surface Pro 11 and Surface Laptop 7 that had not installed Microsoft Store updates, while classic Office apps were reportedly unaffected. Microsoft resolved the problem in the September 8, 2026 cumulative update (KB5124012) and advised users to install the latest updates.

🔗 Amazon SageMaker Unified Studio now supports ODBC connections, enabling Microsoft Power BI and other ODBC-compatible tools to connect directly to governed data using the Amazon Athena ODBC driver. The project overview page adds an ODBC connection details view alongside the existing JDBC view, providing DSN-less connection strings or parameters for named DSNs. The Athena ODBC driver (v2.2.0.1+) supports SageMakerBrowserIdc and SageMakerIam authentication modes. ODBC access is available today in all Regions where SageMaker Unified Studio is offered, at no additional cost.

🔧 AWS Parallel Computing Service (AWS PCS) now supports custom Generic Resource (GRES) settings, hardware topology configuration, and NVIDIA Multi-Instance GPU (MIG), and upgrades to Slurm 26.05. These additions let Slurm autodetect NUMA domains as sockets, configure GPU core affinity and GPU interconnects, and enable finer control over task placement for HPC and ML workloads. Custom GRES and topology settings let administrators declare accelerators, partition GPUs with MIG, or override defaults for specialized scheduling.

🔔 Amazon RDS for Oracle now supports the Supplemental Patch Bundle (SPB) for the July 2026 Release Update (RU) for Oracle Database 19c and 26ai. Starting April 2026, Oracle renamed the Oracle Spatial Patch Bundle to Supplemental Patch Bundle, which includes targeted fixes for features like Oracle Spatial, Data Pump, and GoldenGate. You can apply the SPB to new or existing instances by selecting the "Supplemental Patch Bundle Engine Versions" checkbox in the AWS Console, and use AWS Organizations rollout policy to stagger automatic minor upgrades across environments.

🚀 Microsoft reports being named a Leader in the 2026 Gartner® Magic Quadrant™ for Container Management, positioned furthest on Completeness of Vision. The post outlines how Azure's container portfolio—including AKS, Azure Container Apps, Azure Arc, and Azure Kubernetes Fleet Manager—supports diverse AI and application workloads across cloud, edge, and hybrid environments. It emphasizes open-source alignment with upstream Kubernetes and investments in operational automation and agentic operations to address cluster sprawl and governance challenges.

🔒 AWS Lambda recursive loop detection is now supported in Europe Sovereign Cloud, helping prevent unintended billing by automatically detecting and stopping recursive invocations between Lambda functions and supported services. The feature covers common event sources like Amazon S3, Amazon SQS, and Amazon SNS, and halts processing when a loop is detected while sending an AWS Health Dashboard notification with troubleshooting guidance. Recursive loop detection is enabled by default for Lambda functions using a supported SDK version, and can be disabled per function using the PutFunctionRecursionConfig API for intentional recursion.

🔔 Amazon API Gateway now supports configurable delivery destinations and larger execution log events for REST APIs. Previously limited to a single API Gateway-managed CloudWatch Logs group with 1 KB truncation, execution logs can now be routed up to 1 MB to your own CloudWatch Logs groups, Amazon S3 buckets, or Amazon Data Firehose streams. You can deliver to multiple destinations simultaneously and choose structured formats like Apache Parquet for S3 or JSON for CloudWatch. This capability is available in all Regions where API Gateway REST APIs run, including AWS GovCloud (US), and log delivery is charged at vended logs rates.

🛠️ AWS announced that AWS Transform for .NET can automatically generate unit tests for modernized code. When enabled, the feature targets testable classes such as business logic and controllers and produces unit test code as part of the migration job. The generation is opt-in and available via the AWS Toolkit for Visual Studio in all Regions that support AWS Transform for .NET. To use it, run a .NET transformation in Visual Studio and choose to generate unit tests.

🚀 Today Google Cloud announced a new installable plugin for AI coding agents that packages related agent skills and tools to simplify working with Google Cloud. The plugin follows the open Agent Plugins specification to provide consistent manifests and interoperability across agent environments. The flagship google-cloud-developer plugin helps agents with authentication, authorization, project management, gcloud guardrails, and integrates the Developer Knowledge MCP server for up-to-date documentation grounding.

🛠️ AWS Lambda durable functions now integrates with Pydantic AI, enabling Python AI agents to persist progress and resume after interruptions. Each model and tool call becomes a durable execution step, preventing repeated work and avoiding duplicate costs or side effects. The integration is available in all Regions where Lambda durable functions run and can be used in any Python Lambda durable function.

🐧 Amazon MQ now supports RabbitMQ 4.3, introducing quorum queue enhancements including compaction, expanded strict priority support, native delayed retries, and configurable consumer timeouts. The release also brings bug fixes and memory management improvements. Some legacy features—Transient non-exclusive queues, Global QoS, and Classic queues v1 storage—are removed. The new broker version is available on m7g instances and Amazon MQ will manage patch upgrades automatically.

🔔 AWS has announced general availability of second-generation single-rack AWS Outposts, a self-contained 42U rack integrating compute, storage, and networking for low-latency and space-constrained environments. The rack delivers up to 2,688 vCPU and 100 TB of Amazon EBS storage and supports the latest x86-powered EC2 instances, including M7i, M8i, C7i, C8i, R7i, and R8i, as well as Outposts accelerated networking instances. It provides a consistent experience with multi-rack Outposts using the same APIs, console, automation, governance, and security controls across AWS Regions and on-premises locations.

🔍 This post, the final entry in The Economics of Agent Optimization series, explains how AI agent governance makes consumption visible, bounded, and attributable to control costs. It outlines request-time token limits, project quotas, and Azure API Management’s AI Gateway observability to enforce limits and trace usage. The article also describes Microsoft Foundry’s emerging ROI features that map agent costs to business outcomes, enabling teams to measure value and prioritize optimizations.