Research
Cloud Economics··19 min

Vercel vs AWS vs Railway: True 12-Month Costs for a Real SaaS Product

We built and ran the same production application on three platforms for a full year. The all-in cost difference might surprise you.

Pricing pages lie by omission.

They show you the happy path. They rarely show bandwidth overages, cold start costs, database scaling, logging, support, or what happens when your usage pattern doesn't match their ideal customer.

We took a real product (a typical B2B SaaS with auth, dashboard, API, background jobs, and file uploads) and ran it on Vercel, AWS (Amplify + ECS + RDS), and Railway for 12 months. Here are the actual numbers.

The Application Profile

  • ~18,000 monthly active users
  • 42k API requests/day average
  • 280GB egress/month
  • PostgreSQL database (peak 48GB)
  • Background job processing (image resizing, emails, webhooks)
  • 3 environments (production, staging, preview)

This is a completely normal workload for a growing SaaS company.

12-Month Total Cost of Ownership

PlatformComputeDatabaseBandwidthOtherTotalHidden %
Vercel Pro ,940
,800
,120
$890$6,75034%
Railway$3,480
,440
$680$420$6,02019%
AWS (optimized) ,180 ,640
,940
,710
$8,47051%

"AWS optimized" means we used a team that actually knows AWS well (not a junior engineer clicking buttons in the console).

Two notes on reading the table. *Bandwidth* is all-in data transfer (egress plus requests, CDN, and build/registry traffic), so the effective per-GB cost sits well above each platform's headline egress rate. Railway still comes out cheapest on bandwidth in both absolute and per-GB terms. *Hidden %* is the share of the total that teams typically fail to forecast up front, and it is spread across every column, not just "Other."

What the Pricing Pages Don't Tell You

Vercel

  • Function duration and memory are more expensive than advertised once you have non-trivial API work
  • Image optimization and edge config add up faster than expected
  • Database (Vercel Postgres/Neon + external) + connection pooling fees are significant
  • Preview deployments for every PR become expensive at scale

Railway

  • Much more transparent, but volume discounts kick in later than competitors
  • Egress is genuinely cheaper
  • Support is slower (you pay for it in engineering time)

AWS

  • The 51% "hidden" number is not unusual. It includes CloudWatch, WAF, data transfer between AZs, NAT gateway, support plan, and the 40+ small services most teams eventually turn on.
  • Reserved instances and Savings Plans only help if you can predict usage 12 months out.
  • The real cost of "just use ECS + RDS" is dramatically higher than the marketing suggests unless you have dedicated platform engineers.

When Each Platform Actually Wins

Choose Vercel if:

  • Your team values velocity over cost control
  • You have heavy frontend + serverless API patterns
  • You are willing to pay a premium for DX and will stay under ~$800-900/mo

Choose Railway if:

  • You want most of Vercel's experience at lower cost
  • Your workload is relatively predictable
  • You don't need the absolute cheapest price

Choose AWS (or GCP/Azure) if:

  • You already have deep platform expertise in-house
  • You're spending $4k+/month and can amortize specialized knowledge
  • You need specific compliance or regional requirements

The Real Decision Framework

Most teams should not be optimizing for the absolute lowest infrastructure cost at Series A/B. They should be optimizing for:

1. Engineering time saved 2. Reliability under load 3. Ability to iterate quickly on product

At $6k-8.5k/year, the difference between these platforms is often smaller than the cost of one engineer's time for two months.

The teams that get into real trouble are the ones who scale to $40k-80k/year on one of these platforms without ever doing a proper cost audit.

Our Actual Recommendation (2026)

For most startups between $0 and ~$3M ARR:

  • Start on Railway or Vercel Hobby/Pro
  • Move to dedicated AWS/GCP only when you have a platform engineer and consistent $6k+/month spend
  • Run the real numbers every 9-12 months instead of trusting pricing calculators

The biggest savings almost never come from switching cloud providers. They come from actually looking at what you're using and turning the expensive parts off.

---

*Full methodology and detailed monthly breakdowns available to paid subscribers. Raw billing exports available upon request for verification.*

This article is part of ongoing research into real technology costs. Figures are based on public pricing at publication time and may change.

Try the tools

Plan AI and cloud spend before it lands.

Open the pricing index, then use the calculators to model your real workload.

For Engineering

Model costs by token, understand the economics of feature complexity.

For Finance

Budget forecasting and vendor negotiation with live pricing updates.

For Product

Compare models, simulate scenarios, monitor pricing changes in real time.

Browse tools
ByteCosts

Cost intelligence for AI, cloud, and SaaS. Public pricing, normalized into an index and calculators that engineering and finance can use in the same room.

Catalog: 137 providers · 4,993 models · updated Jun 1, 2026

Prices via models.dev and custom scrapers · model quality benchmarks via Artificial Analysis

Disclaimer: All information provided is for reference purposes only. Actual costs may vary based on usage patterns and provider terms. Always monitor your own token consumption and billing dashboard to track real expenses.

© 2026 ByteCosts. All rights reserved.
Built on public pricing data and browser-side calculators. Figures are directional.