Railway
Card required
P90 billUnavailable
- Typical
- —
- Worst
- —
- Over budget
- —
- Risk
- Blocked
Eligibility: Railway requires a payment card or prepaid credits before this scenario can keep running.
ShipCheap Decision Board
Billing risk simulator
Model the surprise-bill path before you deploy.
Pick a provider and workload shape. ShipCheap estimates how user growth, bot traffic, bandwidth, API calls, and storage can turn a cheap deploy into real spend.
Low simulated bill risk
1,000 runsMost simulated months stay near $4.02 or less.
CapRover
Free open-source software; real monthly cost is the VM or infrastructure you run
P90 simulated bill
$4.02
Card path
No-card path
Always-on
Supported
Simulated monthly bill
ShipCheap sampled 1,000 possible months from this scenario. The values are rough estimates for risk shape, not official pricing.
Typical month$4.01
Bad month$4.02
Worst sampled$4.03
Chance over your budget ($25)
0%
Chance over $25
0%
Chance over $100
0%
Modeled high-usage month
Around 638 users and 24,808 requests in the rough P90 usage sample.
Largest cost centers
base platform cost: $4.00
This is a rough Monte Carlo-style model, not an official provider quote. It uses ShipCheap's provider traits plus generic usage assumptions so you can test risk shape before checking live pricing docs.
Bill duel
Same workload. Different providers. Who blows up first?
Keep the scenario fixed and pit up to 3 providers against each other on P90 bill, over-budget odds, and the cost center that spikes first.
First to blow: CapRover
Base provider
CapRover
Provider details
CapRover
Blows up first
P90 bill$4.02
- Typical
- $4.01
- Worst
- $4.03
- Over budget
- 0%
- Risk
- Low
First cost center to spike: base platform cost ($4.00)
Render
Safest in this duel
P90 bill$0.00
- Typical
- $0.00
- Worst
- $0.00
- Over budget
- 0%
- Risk
- Low
First cost center to spike: no major overage center
Under this scenario, CapRover shows the highest P90 / over-budget pressure, while Render stays cheapest at the rough high-usage month. Verify live pricing before you trust the ranking.
Main risk drivers
- This scenario has a conservative no-card or capped path.
Controls to set first
- Keep checking provider limits before adding production traffic.
Billing incidents worth modeling
Public docs and major reporting show the risk pattern: spend controls, exposed keys, compromised credentials, bots, or metered usage growing faster than the owner notices.
Vercel
Spend controls need explicit actions
Vercel's docs say spend amounts can trigger notifications, webhooks, or pausing, but project pausing must be enabled as an action.
Firebase/Gemini
Client keys can expose AI usage
Firebase docs warn against adding Gemini API access to app-exposed keys and recommend App Check for serious apps.
Google/Gemini
$15.4k reported
TechRadar reported a case where an exposed Google API key was abused for Gemini requests before the owner could stop the spend.
AWS
$45k reported
Tom's Hardware reported a compromised AWS account used for crypto-mining workloads that generated a large cloud bill.
These examples are cautionary references from official docs or major reporting, not predictions or provider-specific price promises.
Lower-risk alternatives
Providers with no-card paths or lower listed billing risk.
Render
Medium riskA developer-friendly platform for web services, workers, cron jobs, static sites, and managed databases.
Koyeb
Low riskA serverless-style app platform for containers, Git deploys, APIs, workers, and global services.
Vercel
Medium riskA frontend-first platform with strong serverless and edge support, best known for Next.js deployments.
Supabase
Low riskA hosted Postgres platform with auth, storage, edge functions, and realtime features.