Architecting AI Infrastructure

A practical series on GPU consumption models, vSphere and VCF placement decisions, and efficiency trade-offs for production AI platforms.

This series explains how AI infrastructure behaves under real constraints: GPU non-fungibility, placement decisions, and practical sizing trade-offs.

Browse all parts below.

All parts

  1. Why GPU Placement Becomes the Defining Problem (Part 1)
    Feb 9, 2026
  2. GPU Consumption Models as the First Architectural Choice in Production AI (Part 2)
    Feb 11, 2026
  3. How vSphere DRS Makes GPU Placement Decisions (Part 3)
    Feb 13, 2026
  4. How vSphere GPU Modes and Assignment Policies Determine Host Level Placement (Part 4)
    Feb 17, 2026
  5. How Same Size vGPU Mode and Right-sizing Shape GPU Placement Efficiency (Part 5)
    Feb 19, 2026
  6. Mixed Size vGPU Mode in Practice (Part 6)
    Feb 24, 2026
  7. Same Size vs Mixed Size Placement at Cluster Scale (Part 7)
    Mar 1, 2026
  8. MIG Partitioning, Placement Geometry, and Stranded Capacity (Part 8)
    Mar 6, 2026
  9. Why Multi GPU Requires Topology Awareness (Part 9)
    Mar 16, 2026
  10. Understanding Multi-GPU Topologies Within a Single Host (Part 10)
    Mar 27, 2026
  11. Topology-Aware Multi-GPU VM Placement (Part 11)
    Mar 31, 2026