UK puts AWS, Azure, Google Cloud and Oracle under direct financial regulation as Critical Third Parties On 13 July 2026 the UK's Critical Third Parties regime went live, with HM Treasury designating AWS EMEA, Google Cloud EMEA, Microsoft Ireland, and Oracle UK as the first providers under direct regulatory oversight. For the first time, the Bank of England, the PRA, and the FCA will
CloudNativePG 1.30 brings declarative Postgres roles and safer failover to Kubernetes The CloudNativePG community has released version 1.30.0 of its PostgreSQL operator, the de facto standard for running production Postgres on Kubernetes. The headline feature is a new DatabaseRole custom resource that turns database roles into standalone Kubernetes objects with their own lifecycle and RBAC — a genuinely GitOps-friendly model
Amazon EKS makes Kubernetes upgrades reversible with version rollbacks Kubernetes upgrades have long been a one-way door: once a cluster moved to a new minor version, the only way back was a rebuild and migration. AWS has now closed that gap with version rollbacks for Amazon EKS, letting administrators reverse a control-plane upgrade within seven days of completing it.
Claude Sonnet 5 brings Opus-class agents down to mid-tier prices Anthropic released Claude Sonnet 5 on June 30, and the significant part is not another benchmark race — it is the economics of running AI agents. The company positions Sonnet 5 as its most agentic mid-tier model yet: it plans, drives browsers and terminals, and completes multi-step work that until recently
VS Code 1.128 brings multi-chat agent sessions and general availability for Copilot Vision Microsoft shipped Visual Studio Code 1.128 on July 8, 2026, and the headline feature is multi-chat agent sessions: a single Claude-powered agent session in the Agents window can now hold several related chats at once, so a developer can run a main conversation, fork an alternate approach, and track
Git 2.55 ships Rust support by default and a new history fixup command Git 2.55 landed this week, and the headline change is structural: Rust support is now enabled by default in the build, a first step toward hardening parts of Git’s C core with a memory-safer language without changing anything for ordinary users. The release also ships a new git
AWS DevOps Agent learns to gate releases before they ship AWS added a release management capability, now in preview, to its DevOps Agent, the same assistant that already handles post-deployment incident response. The new features target a different bottleneck: pull requests piling up faster than teams can review them as AI coding tools accelerate code output. A release readiness review
AWS Lambda MicroVMs: Firecracker-isolated sandboxes for running untrusted code At its New York Summit, AWS introduced Lambda MicroVMs, a new serverless compute primitive aimed at a problem that has been awkward to solve: giving every end user or session its own isolated environment to safely run code the application developer didn't write. Think AI coding assistants, interactive
IBM Sovereign Core: digital sovereignty becomes an operational product for AI At Think 2026, IBM moved digital sovereignty from policy to operations with the general availability of IBM Sovereign Core, a self-managed software platform for deploying and running AI and cloud workloads in environments governed entirely by the customer. The pitch is that sovereignty now reaches well beyond data residency: it
IBM and Red Hat's $5B Project Lightwell bets on securing open source in the AI era IBM and Red Hat have committed $5 billion and more than 20,000 engineers to Project Lightwell, an initiative to secure the open-source software that underpins enterprise IT against AI-accelerated threats. The premise: frontier models are collapsing the exploit window from weeks to hours, with one preview model reportedly flagging
Google Cloud's eighth-generation TPUs and the agentic-era infrastructure race At Google Cloud Next '26, Google expanded its AI Hypercomputer platform with two new eighth-generation TPUs: the TPU 8t for training, packing 9,600 chips in a single superpod for nearly 3x the compute of the prior generation, and the TPU 8i tuned for low-latency inference and reinforcement learning,
