Railway’s $100M bet: Can an AI-native cloud really take on AWS?

Railway, the San Francisco-based cloud platform that somehow got two million developers without spending a dime on marketing, just announced a $100 million Series B. TQ Ventures led the round, with FPV Ventures, Redpoint, and Unusual Ventures joining in. The valuation makes it one of the more interesting infrastructure plays to come out of the AI boom.

And honestly, the pitch makes sense. The old cloud stack — AWS, Google Cloud, Azure — was built for a world where developers typed code slowly and deployments took minutes. That world is dead. AI coding assistants like Claude, ChatGPT, and Cursor can spit out working code in seconds. Waiting two to three minutes for a Terraform build-and-deploy cycle feels like watching paint dry.

Jake Cooper, Railway’s 28-year-old CEO, put it well: “When godly intelligence is on tap and can solve any problem in three seconds, those amalgamations of systems become bottlenecks.” He’s not wrong.

Railway claims its platform does deployments in under a second. Customers report 10x faster developer velocity and up to 65% cost savings. Those aren’t just internal benchmarks — Daniel Lobaton, CTO at G2X, saw his monthly infrastructure bill drop from $15,000 to about $1,000 after moving to Railway. Deployment speed improved seven times. He said, “The work that used to take me a week on our previous infrastructure, I can do in Railway in like a day.”

What really caught my attention was Railway’s decision in 2024 to abandon Google Cloud entirely and build its own data centers. That’s a gutsy move for a startup. Most companies would just rent more capacity. But Cooper cited the Alan Kay maxim: “People who are really serious about software should make their own hardware.” It paid off during recent cloud outages — Railway stayed up while others went down.

This vertical integration lets them undercut hyperscalers by roughly 50% and newer cloud startups by three to four times. Pricing is per-second: $0.00000386 per gigabyte-second of memory, $0.00000772 per vCPU-second. No charges for idle VMs. That’s a direct jab at the traditional model where you pay for provisioned capacity whether you use it or not.

Cooper pointed out, “The conventional wisdom is that the big guys have economies of scale to offer better pricing. But when they’re charging for VMs that usually sit idle in the cloud, and we’ve purpose-built everything to fit much more density on these machines, you have a big opportunity.”

Here’s the part that makes me skeptical: Railway has only 30 employees. Thirty. They’re processing over 10 million deployments monthly and handling more than one trillion requests through their edge network. That’s impressive, but also terrifying. One key person leaves or gets sick, and the whole thing could wobble. AWS has armies of engineers. Railway has a small team that’s probably running on caffeine and pure will.

Still, the traction is undeniable. Two million developers without any marketing spend? That’s organic growth most companies would kill for. The question is whether Railway can scale its infrastructure and team fast enough to handle the demand without breaking.

For now, it’s a compelling alternative for developers who are sick of the complexity and cost of the old guard. If you’re tired of waiting minutes for deployments and paying for idle VMs, Railway might be worth a look. Just don’t expect it to replace AWS for every use case — not yet, anyway.