Infrastructure news, including cloud architecture, backend systems, performance optimization, and scalable deployment strategies
185 stories
AMD unveiled Instella‑MoE‑16B‑A3B, a 16B parameter open‑source Mixture‑of‑Experts language model that activates only 2.8B parameters per token, trained from scratch on Instinct MI300X and MI325X GPUs.
Google Cloud’s monthly blog highlights new AI infrastructure releases, including managed Lustre, C4N VMs, and GKE Dataplane V2, all announced on July 31, 2026.
A Katalyst Space robot tasked with raising the Swift Observatory failed to stabilize, raising questions about private‑sector space operations.
Microsoft announced a new AI model that reduces cybersecurity expenses by half, leveraging specialized models for threat detection and response.
Enterprise spending on cloud infrastructure services hit a record $143.4 billion in Q2 2026, driven by generative‑AI workloads and expanding hyperscale and neocloud providers.
A Currence report warns that 30‑50% of announced 2026 data‑center capacity may be delayed due to power constraints and construction bottlenecks.
Amazon Web Services launched the general availability of AWS Interconnect, a multicloud networking product that connects AWS with Oracle Cloud Infrastructure and Google Cloud.
The cloud‑security startup Act Security has secured $60 million to launch an action‑centric platform that protects enterprises from AI‑agent access risks in large‑scale cloud environments.
China unveiled a $1.03 trillion national infrastructure plan to connect AI computing, power grids, water systems and logistics, aiming to boost competitiveness amid tech restrictions.
Oracle introduced a new on‑premises appliance that lets enterprises run its flagship database and AI workloads in branch offices, blending cloud economics with local control.
A bug in Microsoft’s automated network maintenance system caused IP routes to be removed from devices, isolating a West US Azure datacenter and disabling Teams, SharePoint, OneDrive and other services for nearly five hours.
Microsoft disclosed an elevation‑of‑privilege flaw in Azure DNS that could allow unauthorized DNS modifications, urging customers to review RBAC and tighten identity controls.
An AWS networking failure cut connectivity between the us‑west‑2 region and Seattle, causing cascading outages for dozens of services.
A power line outage outside Washington, DC, caused over 3 GW of data centers to shut down, highlighting the grid impact of AI workloads and the need for better power management.
Atos announced its new sovereign cloud offering, aimed at governments, defense and other regulated sectors, providing full data control and cyber‑resilience within the EU.
Runway introduced its Media Router, an infrastructure layer that automatically selects the best generative media model based on quality, speed, or cost.
AI companies’ strict guardrails and vetted programs are limiting legitimate security researchers’ ability to test and defend systems, prompting calls for more open access.
Google unveils a new data‑center blueprint to support large‑scale AI agents, introducing updated hardware, software, and orchestration layers.
AWS announced that EKS Auto Mode and the open‑source Karpenter now support Elastic Fabric Adapter and EC2 placement groups, enabling higher performance for distributed training and inference workloads.
Google Cloud’s revenue grew 82% year‑over‑year in the April‑June quarter, while Alphabet increased its 2026 AI-capital‑expenditure range to $195‑$205 billion.