Railway’s $100M Bet: Can a 30-Person Team Really Take On AWS?

Railway just announced a $100 million Series B, and honestly, I wasn’t surprised. I’ve been watching this San Francisco startup for a while now, and they’ve been quietly doing something most cloud companies only talk about: building infrastructure that actually keeps up with AI-generated code.

Two million developers have signed up without a single dollar spent on marketing. That’s not a typo. They raised only $24 million before this round—a $20 million Series A back in 2022—and now they’re processing over 10 million deployments per month and handling more than a trillion requests through their edge network. Those numbers rival companies with ten times the funding.

The Core Problem: Three-Minute Deploys Are Killing Developer Velocity

The pitch is simple and brutal: the tools we use to deploy software were designed for a slower era. A standard build-and-deploy cycle with Terraform takes two to three minutes. That was fine when humans wrote code at human speed. But now AI coding assistants like Claude, ChatGPT, and Cursor can generate working code in seconds. That two-minute delay? It’s a bottleneck.

Jake Cooper, Railway’s 28-year-old founder, put it better than I could: “When godly intelligence is on tap and can solve any problem in three seconds, those amalgamations of systems become bottlenecks.”

Railway claims deployments in under one second. That’s fast enough to keep pace with AI-generated code. Customers report tenfold increases in developer velocity and up to 65% cost savings. These aren’t internal benchmarks—they’re from actual clients.

Daniel Lobaton, CTO at G2X, saw his deployment speed improve seven times and his infrastructure bill drop from $15,000 per month to about $1,000. “The work that used to take me a week on our previous infrastructure, I can do in Railway in like a day,” he said.

The Controversial Bet: Abandoning Google Cloud to Build Their Own Data Centers

Here’s where it gets interesting. In 2024, Railway made the unusual decision to ditch Google Cloud entirely and build their own data centers. That’s a massive bet for a startup. It echoes Alan Kay’s famous line: “People who are really serious about software should make their own hardware.”

Cooper explained it as designing hardware for a differentiated experience. Full control over network, compute, and storage lets them optimize for fast build and deploy loops. It also paid off during recent cloud outages—Railway stayed online while major providers went down.

This vertical integration lets them undercut hyperscalers by about 50% and newer cloud startups by three to four times. They charge by the second for actual compute usage: $0.00000386 per gigabyte-second of memory, $0.00000772 per vCPU-second, and $0.00000006 per gigabyte-second of storage. No charges for idle VMs. That’s a stark contrast to the traditional model where you pay for provisioned capacity whether you use it or not.

“The conventional wisdom is that the big guys have economies of scale to offer better pricing,” Cooper noted. “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.”

Can 30 People Really Take On AWS?

Here’s the part that gives me pause. Railway operates with just 30 employees. Thirty people running infrastructure that processes over a trillion requests. That’s impressive, but it also raises questions about resilience, support, and long-term scalability.

AWS has hundreds of thousands of employees, decades of enterprise relationships, and a compliance portfolio that covers every regulation you can name. Railway is building for developers who value speed and simplicity over vendor lock-in. But enterprises need more than fast deploys. They need SLAs, security certifications, and someone to call at 3 AM.

Cooper acknowledged this in the interview: “We’re not trying to be everything to everyone. We’re building for the developer who wants to ship fast and not think about infrastructure.”

Fair enough. But as they scale, they’ll need to address enterprise concerns without losing the simplicity that made them popular in the first place.

The Pricing Model That Actually Makes Sense

Railway’s per-second billing is refreshing. No more paying for idle VMs. No more guessing how much capacity you need. You pay for what you use, and the rates are transparent.

Compare that to AWS where you’re still paying for EC2 instances even when they’re doing nothing. Or Google Cloud where sustained use discounts require you to run workloads for a full month. Railway’s model aligns cost with actual value delivered.

Is it enough to disrupt the cloud market? Probably not overnight. But for startups and AI-native applications where every millisecond and dollar counts, Railway offers a compelling alternative.

The Bottom Line

Railway is building infrastructure for the AI era. Sub-second deploys, per-second billing, and full vertical integration. The $100 million Series B gives them runway to expand their data center footprint and hire more engineers.

But the real test will be whether they can maintain that developer experience as they scale. Thirty people can only do so much. And the hyperscalers aren’t sitting still—AWS just launched faster deployment options and Google Cloud is pushing its own AI-native tools.

For now, Railway has a window. Developers are frustrated with legacy cloud complexity. AI is accelerating the pace of code generation. If Railway can keep delivering sub-second deploys and cost savings, they might just carve out a meaningful slice of the market.

I’ll be watching. This is one of the more interesting infrastructure bets I’ve seen in years.