# Cost prediction · ICE by Light Cloud

> Per-canvas monthly projection before you apply. See the bill while you design, not at the end of the month.

## How it works today

- **Per-block, not per-bill.** Every block contributes a line item. You see what each piece costs, not just a total.
- **Recomputed as you edit.** Bump the Postgres instance size, the projection updates immediately.
- **Tied to the provider.** Same canvas, different target, different estimate. Cost is one of the things a migration affects, and you see it before you commit.
- **Plan before apply.** The estimate is part of the plan output. The cost stays visible while the deploy runs.

## What goes into the projection

ICE uses each provider's public pricing data plus the parameters you set on each block: instance class, region, replicas, storage size, scaling profile. The number is a projection, not a guarantee. Variable costs (egress, per-request) are estimated against modelled traffic and surfaced as a separate line. Free tiers and committed-use discounts are subtracted where applicable.

## Coming up

- Projected vs actual dashboards on the canvas.
- Drift alerts when the bill diverges from the projection.
- Cost outliers surfaced as proactive suggestions.
- Reserved instance and committed use suggestions per cloud.
- Per-environment budgets with hard and soft caps.

Roadmap: https://github.com/light-cloud-com/ice/blob/main/ROADMAP.md#observability