ComputeLabs Research

AI & Compute Infrastructure — September 23, 2026

Edition of · 19 stories

🌎 Macro, Policy & Capital

Microsoft plans over $10 billion in Middle Eastern capital and operating spending through 2030, expanding cloud and AI infrastructure.

Microsoft (微软) announced plans for more than $10 billion in capital and operating spending through 2030 (“超过 100 亿美元资本及运营支出”). The geographic focus comprises Kuwait, Qatar, Saudi Arabia, and the United Arab Emirates; this is a spending plan, not completed investment or exclusively data-center construction expenditure.

Read the full story · 3 sources →

🖥️ Chips, GPUs & Systems

AI-infrastructure provider WhiteFiber commercially launched Continuum, a cross-data-center networking solution supporting distributed GPU supercluster architecture for enterprise customers.

WhiteFiber announced the commercial availability of WhiteFiber Continuum for enterprise customers. The offering is a cross-data-center networking solution supporting a distributed graphics processing unit (GPU) supercluster architecture, rather than a new GPU or server product.

Read the full story · 1 source →

⚡ Power, Grid & Data Centers

Federal Energy Regulatory Commission rejected ComEd’s cancellation of a contract for PowerHouse Hillwood’s planned 1.8-gigawatt, $20 billion data center.

The Federal Energy Regulatory Commission (FERC) rejected ComEd’s cancellation of a contract associated with PowerHouse Hillwood’s planned data center. Utility Dive identifies the project as 1.8 gigawatts and $20 billion, both attached to the planned development rather than operating capacity or expenditure already incurred.

Read the full story · 1 source →

☁️ Cloud & Compute Infrastructure

Google Kubernetes Engine 1.37 adds native scaling to and from zero replicas, restarting workloads on capacity buffers when demand returns.

Google announced native scale-to-zero functionality in Google Kubernetes Engine (GKE) 1.37 for intermittently active workloads such as batch processors, event-driven workers, and development environments. Workloads can scale down to zero replicas and restart on GKE capacity buffers when demand returns; this does not mean the entire cluster or its reserved capacity scales to zero.

Read the full story · 2 sources →

🤖 AI Models, Software & Research

All ComputeLabs Research editions