
Railway V3: Faster and Cheaper
People told us we were crazy for building our own data centers, but since the dawn of Railway, that’s always been the plan.
If you ask us, we actually think our competitors are the crazy ones. Because without building their own data centers, their businesses won’t work. The incentives won’t work, the experience won’t work, and most importantly, the margins won’t work.
The fact of the matter is, you simply cannot build a cloud on someone else’s cloud.
The reasoning is simple, and I’ll tell you exactly what happens if you do:
The big clouds have fixed costs, compute, networking, and storage.
You’ll be unable to price below the resell price of these fixed costs, which will force you to create margin by charging 3-5x premiums or paywalling every one of your features behind some “Talk to a Sales Person” Enterprise paywall. Maybe even both!
This “hat on a hat” business model will actually work for a while. But you’ll go to bed every night thinking “Is tomorrow the day someone comes for our margin?”
Yes. Welcome to tomorrow.
Today we’re launching Railway V3. We’re calling it “Railway Metal.”
Metal is almost 50% cheaper than the “Big3” Cloud Providers.

The following changes are effective today if you’re a Railway customer with 80% or more of your usage on Metal:
- No more seat costs
- 50% cheaper egress
- 40% cheaper storage
- Zero incremental cost for “add-ons” (HIPAA, SLOs, etc.)
Combining the above with Railway’s “Pay only for what you use, by the minute” compute costs, and Railway is ~50% cheaper than the “Big3” providers on all axes.
Throughout testing on Railway Metal we’ve seen:
- Up to 6x faster on multicore benchmarks
- Up to 3x faster on networking benchmarks
- Up to 10x faster on volume write benchmarks
So yes — we’ve simultaneously decreased cost and increased performance. In some places (e.g. database/stateful workloads) the price-performance ratio is now 50x better.
We do this not because it was easy (it wasn’t), but because we’re a technology company. And technology companies innovate so that they can push savings and performance gains back onto users. That’s what we believe.
We do it faster, you pay us less, and you spend even less time messing around with infrastructure. If all of this is true, both Railway and our customers win.
That’s performance. Performance = less.
We have a tough job at Railway. Everybody else wants to sell you more — we want to sell you less.
Our bet is you don’t want 400 tools from the CNCF, you don’t want to configure a “PodDisruptionBudget,” and you sure as hell don’t want yet another 3-letter acronym solving a made-up problem imagined by some marketing devfluencer.
You want to deploy your code as fast as possible, have it scale for as cheap as possible, and handle failure while you sleep or count your money.
You do less, you get more.
That’s infraless, and that’s Railway.
This week, and moving forward, we’re going to be selling you less. Today, less cost. Tomorrow, less bullshit.
This week, you can expect:
- less latency
- less waiting for builds/deploys
- less time waiting for tickets/features
- And one more thing…
So allow me to be the first to welcome you to Launch Week 02.
Here’s to a week of less 🥂