Topic map
Start with the compute-cost problem you are actually facing.
The site is intentionally focused on compute placement decisions where cloud bills, GPU quotes, data movement, and operational tolerance make the answer unclear.
Use these pages when the GPU hourly rate is visible but the workload placement decision is still muddy. The useful question is usually about useful GPU-hours, data movement, provider variance, commitment, and operations.
AWS Bill Shock Decisions AWS Bill Shock Decisions A practical library for diagnosing high AWS bills before deciding whether to optimize, migrate, or re-place the workload.Use these pages when an AWS bill feels wrong and the team is not sure whether the answer is cleanup, architecture change, or leaving AWS.
Cloud Migration Decisions Cloud Migration Decisions A practical library for deciding whether to leave AWS, move a workload, or stay put and fix the expensive parts.Use these pages when the bill makes migration tempting but the real question is whether the workload is portable enough, expensive enough, and operationally worth moving.
AI Inference Cost Decisions AI Inference Cost Decisions A practical library for estimating AI inference cost, API versus self-hosted tradeoffs, batch versus realtime serving, managed inference, and GPU idle cost.Use these pages when an AI app is moving from prototype to production and the real question is what inference will cost per request, per month, and per completed workload.
Framework library Definitions and formulas Workload placement, useful GPU-hours, bill shock, exit paybackUse these when a teammate, article, or planning doc needs a clear concept instead of a single checklist.
Resource library Checklists and worksheets Quote review, triage, exit cost, and placement assetsUse these when a publication, teammate, or planning doc needs a practical worksheet rather than another opinion page.
Topic Coverage
Use this map when you are not sure whether the problem is inference cost, GPU pricing, bill shock, migration, or a broader placement decision.
| Cluster | Scope | Hub |
|---|---|---|
| AI inference cost | API vs self-hosted inference, managed serving, batch vs realtime, effective cost per request | /topics/ai-inference-cost |
| GPU cloud pricing | H100/A100 quotes, useful GPU-hours, hidden fees, inference and training cost | /topics/gpu-cloud-pricing |
| AWS bill shock | NAT Gateway, data transfer, CloudWatch, S3, idle compute, surprise monthly deltas | /topics/aws-bill-shock |
| Cloud migration | AWS exit cost, smaller cloud, bare metal, payback period, migration risk | /topics/cloud-migration |
| RunPlacement frameworks | Definitions and formulas for infrastructure decision concepts | /frameworks |
| Workload placement | Default cloud, GPU cloud, smaller cloud, bare metal, managed platform | /resources/workload-placement-worksheet |
| Authority resources | Reusable checklists and worksheets for sharing with teammates, vendors, and writers | /resources |