All news in category “Vendor and Hyperscaler Watch”

🧠 Cloudflare announces Agent Memory, a private beta managed service that extracts information from agent conversations and makes it available without filling model context windows. The service offers persistent profiles with operations to ingest conversations, explicitly remember or forget items, and recall synthesized answers, integrating with Cloudflare Workers and a REST API. Agent Memory uses a retrieval-based architecture with deterministic ingestion, multi-stage verification, vector and full-text retrieval channels, and Reciprocal Rank Fusion to synthesize concise, contextual responses. Memories are classified, versioned or superseded as appropriate, and fully exportable so organizations retain ownership.

🔁 Cloudflare introduces Redirects for AI Training, a toggle that turns existing rel="canonical" tags into HTTP 301 redirects for verified AI training crawlers. On paid Cloudflare plans this enforcement redirects AI crawler traffic (examples include GPTBot, ClaudeBot, Bytespider) to canonical URLs, preventing ingestion of deprecated content. Human visitors and other automated classes are unaffected.

🚀Using Real User Measurements that capture browser‑side timing via a small background speed test, Cloudflare reports it became the fastest provider in 60% of the top 1,000 networks by December 2025, up from 40% in September. Rankings rely on the trimean of TCP connection time to smooth outliers and reflect real user experience. Improvements came from new points of presence (Wroclaw, Malang, Constantine) and software optimizations such as HTTP/3 support and tighter congestion handling, producing an average 6ms lead over the next provider in December.

⚙️ AWS Clean Rooms now supports configurable Spark properties for PySpark analyses, allowing customers to tune memory overhead, task concurrency, and network timeouts on a per-job basis. This capability helps teams adapt resource allocation to specific performance and scale requirements, improving throughput and cost efficiency. For example, pharmaceutical researchers collaborating with healthcare partners can set tailored memory and concurrency settings for large real-world clinical datasets to optimize runtime and expenses.

🔒 Google announced Play policy updates to restrict contact and location permissions and to strengthen app ownership protections, while reporting it blocked or removed over 8.3 billion ads and suspended 24.9 million accounts in 2025. The update introduces a standardized Contact Picker and a one‑time precise location button in Android 17, and urges developers to remove broad READ_CONTACTS usage. Google also added a native account transfer feature and said its Gemini AI is detecting and preemptively blocking malvertising at scale.

🔒 Palo Alto Networks is participating in Anthropic’s Project Glasswing to test the Claude Mythos model for vulnerability discovery. EMEA CEO Helmut Reisinger says Mythos has identified unprecedented zero-day flaws across multiple operating systems and browsers and can often generate working exploits. Palo Alto is integrating Protect AI, Chronosphere, CyberArk, and soon Koi into its modular platform to secure AI, identity, observability, and agentic endpoints. Reisinger highlighted BYOK, European AI Act compliance, and preparations for the post-quantum era.

📣 This article describes how the Dev Signal team transitioned a multi-agent prototype into production on Google Cloud by combining a FastAPI service, a Vertex AI memory bank, and the Agent Developer Kit. It highlights production-ready concerns including OpenTelemetry traces exported to Cloud Trace for visibility into agent reasoning, and secure secret handling via Secret Manager so credentials never appear in environment variables. The guide also demonstrates reproducible infrastructure using Terraform to provision Artifact Registry, service accounts, Cloud Run, and related APIs, and outlines containerization and Cloud Build steps to deploy new revisions.

🚀 SageMaker JumpStart now offers optimized deployments for foundation models, providing pre-configured, task-aware settings tailored to specific use cases and performance goals. Customers can choose cost-, throughput-, latency-optimized, or balanced configurations and preview P50 latency, time-to-first-token, and throughput metrics before deployment. Supported models include variants from Meta, Microsoft, Mistral AI, Qwen, Google, and TII, and deployments target SageMaker AI Managed Inference endpoints or HyperPod clusters. The feature leverages VPC deployment capabilities and is available in all regions where JumpStart is supported.

🌍 Amazon has expanded CloudWatch RUM to the AWS European Sovereign Cloud, enabling customers subject to strict data residency and sovereignty rules to collect client-side performance data within the sovereign boundary. The feature captures page load times, JavaScript errors, HTTP failures, and mobile signals to help teams triage issues faster. It is intended for enterprises, public sector organizations, and regulated industries operating in Europe.

📡 Amazon CloudWatch now lets customers audit and enable telemetry from AWS services such as Amazon EC2, Amazon VPC, and AWS CloudTrail across multiple Regions from a single region. Administrators can create organization-wide enablement rules scoped to specific regions or all supported regions, and rules targeting all regions automatically expand to include newly launched regions. The feature is available in all AWS commercial regions and standard CloudWatch ingestion pricing applies.

🛡️ AWS Elastic Disaster Recovery (AWS DRS) is now available in the AWS European Sovereign Cloud (Germany). It enables organizations with data sovereignty requirements to protect mission-critical workloads and supports recovery from physical infrastructure, VMware vSphere, Microsoft Hyper‑V, and cloud environments. AWS DRS delivers point-in-time recovery with RPOs measured in seconds and RTOs typically in minutes, and provides a unified process for testing, recovery, and failback.

🖥️ Amazon WorkSpaces Personal and Amazon WorkSpaces Core are now available in US East (Ohio) and Asia Pacific (Malaysia), allowing organizations to provision virtual desktops closer to users for improved responsiveness and potential in‑country data residency. In Ohio, the availability also supports disaster recovery implementations and regional compliance requirements. Administrators can begin by selecting the new Regions in the WorkSpaces management console.

🚀 Amazon has expanded FSx for Lustre Persistent-2 to four more Regions—Asia Pacific (Hyderabad, Jakarta), Europe (Zurich), and South America (São Paulo). These file systems run on AWS Graviton processors and deliver up to 1 GB/s per terabyte throughput with a lower cost of throughput than previous generations. The change helps accelerate machine learning, high-performance computing, media & entertainment, and financial simulation workloads while reducing storage costs. To get started, create a file system through the AWS Management Console.

🤖 WPP leveraged Google Cloud G4 VM instances powered by the NVIDIA RTX PRO 6000 Blackwell and the NVIDIA Isaac Sim image to cut humanoid robot training from hours or days to under an hour, achieving more than 10x speedups. Their pipeline combines OptiTrack motion capture, OpenUSD digital twins, and MuJoCo-based validation to retarget complex human motion to constrained robot kinematics. Training ran at scale using GPU P2P topology and the AI Hypercomputer, condensing learned policies into ONNX for real-time deployment while preserving safety and robustness.

🔐 Post-quantum cryptography is imminent, and Microsoft emphasizes that the biggest challenge is locating every use of cryptography across applications, devices, networks, and services. Building a comprehensive, ongoing cryptographic inventory enables risk-based decisions, crypto agility, and regulatory compliance. The article outlines a practical Cryptography Posture Management lifecycle and recommends Microsoft tools—GitHub Advanced Security, Defender suites, Azure Key Vault—and partner integrations to discover, normalize, assess, prioritize, and remediate cryptographic risks.

🚀 The BigQuery Studio notebook gallery is now generally available, providing a curated collection of pre-built templates that help teams skip setup and start analysis faster. The gallery supports SQL, Python, and Spark workflows and includes templates for generative AI, ML development, and data pipelines. Templates demonstrate best practices for BigQuery DataFrames, serverless Spark, and multimodal analysis. Users can preview templates in read-only mode and add copies directly into their projects from the BigQuery Studio console.

🔔Amazon EC2 X8aedz instances are now available in the Europe (Stockholm) region, offering up to 5 GHz CPU frequency on 5th Gen AMD EPYC (Turin) processors. Built on sixth-generation AWS Nitro Cards, these instances target EDA workloads and relational databases that need high single-thread performance and large memory. They provide a 32:1 memory:vCPU ratio across eight sizes (2–96 vCPUs, 64–3,072 GiB), including two bare metal options and up to 8 TB of local NVMe storage. Purchase options include On-Demand, Spot, and Savings Plans.

🚀 AWS announced general availability of Amazon EC2 C8in and C8ib instances powered by custom sixth-generation Intel Xeon Scalable processors and the latest AWS Nitro cards. These instances deliver up to 43% higher performance versus the prior C6in generation and scale up to 384 vCPUs. C8in provides up to 600 Gbps networking for network-intensive workloads, while C8ib offers up to 300 Gbps EBS bandwidth for high-performance databases and file systems. Both families are available in select regions and via On-Demand, Savings Plans, and Spot.

🚀 Cloudflare details optimizations to run extra-large open-source LLMs on Workers AI, notably making Kimi K2.5 three times faster and adding more models. The post explains hardware tuning, prefill–decode disaggregation, token-aware load balancing, and prompt-caching via an x-session-affinity header to improve throughput and tail latency. It also covers KV-cache sharing with Mooncake, speculative decoding with NVIDIA EAGLE-3, and Cloudflare’s Rust-based inference engine Infire for multi-GPU, low-memory, fast cold-start inference.

🚀 Cloudflare announced a unified AI inference platform that lets developers call models from multiple providers through one API and a single billing plane. Using the same AI.run() binding for Workers, teams can switch providers with a one-line change; REST support is coming for other environments. The update adds default gateways, automatic retries, finer logging, and centralized cost monitoring to reduce latency, improve reliability, and control costs for agentic workflows.