[[Key Insights from 80 Essays]] ### Fundamentals & Market - [[Next Generation Data Center]] - [[Kinds of Data Centers]] - [[Industries Benefiting from Data Center Set Ups]] - [[Demand Drivers for Data Centers]] - [[Data Center Opportunities]] - [[India Data Center Revolution - A USD 600 Billion Opportunity]] - [[The Unsustainable Data Center]] ### Reliability & Standards - [[Data Centre Redundancy]] - [[Data Center Tiers]] - [[CE EU Standards]] - [[ISO Standards Security x DC]] - [[Site Set Up]] - [[Site Acceptance Testing - SAT]] - [[SAFE Modules]] ### Power Chain - [[Power Infrastructure]] - [[Single Line Diagram]] - [[UPS Batteries]] - [[Peak Power Output]] - [[RMS]] ### Metrics & Measurement - [[Data Centre Metrics]] - [[Data Center Master Metrics]] - [[Rack power Density]] - [[Power Usage Effectiveness - PUE]] - [[Thermal Design Power - TDP]] - [[IT Load]] - [[Power Consumption x Scalability]] ### Cooling Hierarchy - [[Heat is the new constraint]] - [[Cooling and Energy Efficiency Technologies]] - [[Direct-to-Chip Cooling]] - [[Immersion Cooling]] - [[Two-Phase Cooling]] - [[How AI Will Impact Energy Demand]] - [[Data Centre Energy Demand]] ### Networking & Connectivity - [[The Data Center is the Computer - How the Network Shapes Performance]] - [[NVMe Fabric]] - [[Active Optical Cable]] - [[NIC - Network Interface Cards]] ### Compute Abstraction Stack - [[VLSI]] - [[Bare Metal]] - [[VMs]] - [[Docker Containers]] - [[MIGs]] - [[OpenCL]] - [[Clustering]] - [[Scheduling]] ### Multi-Tenancy & Utilisation - [[multi-tenancy]] - [[Instance]] - [[Clusters vs Instances]] - [[Racks]] - [[10 MW DLC CW UPS T3]] ### The AI Stack - [[The AI Stack - Building Blocks]] - [[The AI Timeline - Navigating the Road Ahead]] - [[Scaling Edge Physical AI - Bridging Cloud Intelligence with Real-World Systems]] ### Inference Architecture - [[Inference]] - [[AI Inference Infrastructure]] - [[Nvidia-Groq - Inference Disaggregation Play]] - [[Inference is Eating AI Compute]] - [[Noisy Neighbour Problem]] - [[Convenience-Control Tradeoff]] ### Modular DCs - [[Modular Data Center Design Principles]] - [[Elements of Modular DC Architecture]] - [[Defining Modular DC Architecture]] - [[Challenges in defining Modular DC Architecture]] - [[Documenting Modular DC Architecture]] --- ### Directional Arrows of Progress **Power Density per Rack** `3-6 kW/rack → 20 kW/rack → 50 kW+ /rack` Driven by GPU-dense AI workloads demanding more watts per square metre. **Cooling Phase Shift** `Air (CRAC/CRAH) → Rear-Door Heat Exchangers → Direct-to-Chip Liquid → Full Immersion → Two-Phase` Each step handles roughly 3-5x more heat per rack unit than the last. **Compute Disaggregation** `Monolithic Server → Composable Infrastructure → Disaggregated Compute/Storage/Network` Resources become independently scalable pools rather than fixed server bundles. **Workload Slicing** `Whole Server → VMs → Containers → MIGs (GPU Partitions)` Granularity of allocation keeps shrinking, improving utilisation and multi-tenancy. **Inference vs Training** `Training-dominated capex → Inference-dominated opex` As models mature, inference workloads overtake training in total compute demand.