The Edge-vs-Cloud TCO
Ada re-derives the 3-year edge-vs-cloud total cost of ownership, every line item
ADA · CALCULATION AUDIT
Foundations · optional mathematics
The Edge-vs-Cloud TCO
The chapter compares three-year total cost for 500 vibration sensors: cloud-only totals $11,858, while an edge-hybrid path with a 95% FFT-plus-anomaly cut totals $7,027 — a 41% saving. A 41% figure is just a subtraction of two totals, and bandwidth is 71% of the cloud bill. This audit lists every line item of the edge-vs-cloud TCO so the number survives a budget review.
Companion to the chapter Edge Cost and Trade-offs — every number here comes from that chapter.
Ada's Calculation Audit: the 3-year edge-vs-cloud TCO — 500 vibration sensors, every line item, ~5 minutes
A 41% saving is a subtraction of two totals. Here is each line item for both architectures, so the number survives a budget review rather than a slide.
The working
1. Size the raw pipe. Per sensor 1000 samples/s × 2 B = 2000 B/s = 16 kbps; fleet 500 × 16 = 8000 kbps = 8 Mbps (9.6 Mbps with 20% overhead).
2. Cloud-only over 36 months. Storage 2592 × $0.023 × 36 = $2,146; bandwidth 2592 × $0.09 × 36 = $8,398; compute $0.05 × 8760 × 3 = $1,314. Total = $11,858.
3. Edge-hybrid over 36 months. A 95% FFT-plus-anomaly cut leaves 129.6 GB/month cloud. Hardware $5,000; storage $107; bandwidth $420; periodic training $1,500. Total = $7,027.
| Line item | Cloud-only (3 y) | Edge-hybrid (3 y) |
|---|---|---|
| Edge hardware | — | $5,000 |
| Cloud storage | $2,146 | $107 |
| Bandwidth | $8,398 | $420 |
| Compute | $1,314 | $1,500 |
| Total | $11,858 | $7,027 |
| Saving | $11,858 − $7,027 = $4,831 = 40.7% ≈ 41% | |
What the audit buys you: bandwidth ($8,398) is 71% of the cloud-only bill, and the FFT step (a verifiable 60.9% cut before anomaly filtering, 95% after) is what collapses it — the gateway's $5,000 is repaid by bandwidth savings alone, which is why the 41% edge advantage holds even though the edge path buys hardware the cloud path never does.
Every number above is taken from the chapter's own edge-vs-cloud TCO example and re-derived step by step.