AI Infrastructure Economics Series · Tool 5

NeoCloud Sourcing & Fit Calculator

Two questions, one page. First: do you need dedicated NeoCloud GPU capacity at all — and if so, how much should sit on committed reservation versus burst? Second: which provider is the best fit, on a weighted matrix that prices counterparty risk alongside price and availability. Built for the operator deciding where GPU workloads should live before signing a multi-year commitment.

On the defaults. Rate and provider defaults are starting points, not claims — replace them with your own quotes and diligence. Compute runs server-side; the sourcing model, provider assessment, and counterparty scoring are proprietary to SVAG Labs. Default provider scores draw on the SVAG Labs NeoCloud Counterparty Risk dossier and move with the filings; verify before you commit.

Three reference workloads with distinct sourcing shapes. Select one as a starting point, then edit any value.

Stage 1 Do you need it, and how much? Size the demand, price it across sourcing modes, find the reserve-vs-burst breakeven.
Prepayment requiredRaises counterparty weight.

Rates for the selected GPU class, in $/GPU-hour. Defaults load for the class and term — editable, not API-fetched live. The hyperscaler rate is the "stay where you are" baseline.

Stage 2 Which provider is the best fit? Weighted matrix. Counterparty risk is weighted automatically by your Stage 1 commitment.

Relative importance of each criterion — normalized, so only ratios matter. Counterparty durability is set automatically from your Stage 1 commitment term and prepayment. Override it like any other weight.

Provider scores are computed server-side and not shipped to the browser. Leave a cell blank to use the SVAG default; enter 1–5 to override with your own diligence. The composite updates live.

Best Annual Cost
Monthly Cost
best strategy
$ / GPU-hr Effective
per busy hour
Breakeven Utilization
reserve above this
Top-Fit Provider
weighted

Annual cost by sourcing strategy

Same demand, priced five ways. Cheapest highlighted. The hyperscaler bar is the cost of doing nothing.

The reserve-vs-burst breakeven

Reserved cost is flat; on-demand scales with utilization. Where they cross is the decision. Your utilization is marked.

Weighted fit score by provider

Normalized 0–100. Watch what the counterparty weight does to the leveraged names at longer terms.

Stage 1 verdict

Download the current configuration and computed results for procurement or finance review.