Eight hours. That's how long Railway's platform sat dark on May 19, 2026, after Google Cloud—ironically, the very hyperscaler Railway was trying to escape—suspended the startup's production account. Customers couldn't deploy. They couldn't manage workloads. For a company that had just raised $100 million three months earlier to build its own data centers and take on AWS, the outage was more than embarrassing. It was existential.
This is the paradox Railway finds itself navigating: a 30-person team in San Francisco betting it can outmaneuver the giants by owning metal, promising radical cost savings and speed, while discovering that infrastructure independence comes with its own fragilities. When you control the stack, you also control—and own—the failures.
The $100 million Series B, announced on January 22, 2026, was led by TQ Ventures with participation from FPV Ventures, Redpoint, and Unusual Ventures. Railway has now pulled in roughly $124 million since its 2020 founding, including a $20 million Series A back in May 2022. The company declined to disclose the latest valuation, though CEO Jake Cooper told VentureBeat at the time that Railway is generating "tens of millions" in annual revenue and operating "default alive"—startup parlance for profitability without needing another check.
The pitch is seductive if you've ever wrestled with AWS's labyrinthine console: Railway bundles compute, storage, and networking into what it calls an "intelligent cloud." Developers deploy from GitHub or Docker, use CLI or API tools, browse a template marketplace. Infrastructure, in theory, becomes invisible. Ship faster, configure less.
As of early 2026, Railway claimed north of 2 million users, adding about 200,000 developers monthly. The company says 31% of the Fortune 500 touch the platform in some capacity—Bilt, Intuit's GoCo, TripAdvisor's Cruise Critic, MGM Resorts among the named accounts. Enterprise features include bring-your-own-cloud deployments inside customer VPCs, SOC 2 Type II and SOC 3 compliance, HIPAA Business Associate Agreements. Standard enterprise checklist stuff, but the real story is what Railway started doing in 2024.
That's when the company launched "Railway Metal," its own data center initiative. Instead of reselling hyperscaler capacity—the safer, more boring route taken by most infrastructure startups—Railway began moving off Google Cloud Platform into co-located facilities. A September 2024 changelog entry introduced the Metal beta; by March 2026, Railway was discussing a Stripe integration spanning four regions and more than 40 points of presence.
It's an audacious move for a team this small. Cooper has touted cost savings up to 65% versus traditional clouds and "10x developer velocity," whatever that means in practice. One customer, G2X, reportedly measured an 87% infrastructure cost reduction and 7x faster deployments after switching. Railway bills per-second rather than AWS's hourly or monthly increments—granularity that adds up for workloads that scale unpredictably.
But owning the hardware means owning the risk.
The May 19 outage wasn't Railway's first operational stumble. A March 30 CDN configuration error cached authenticated GET responses for about 0.05% of domains over 52 minutes—small in scope, potentially serious in implication. Mid-February brought a DDoS attack tangled with Cloudflare-related disruptions that stretched across three days. A December 2024 network incident, while Railway was still leaning on GCP, had already prompted the company to accelerate its multi-provider strategy.
Then came May. Google Cloud suspended Railway's production account, which the company maintains was done incorrectly. The incident report Railway published doesn't mince words about the impact: eight hours down, customers locked out of deployments, workloads inaccessible. For a platform selling simplicity and reliability, it was the nightmare scenario—entirely dependent on a partner who could, and did, flip the switch.
Hyperscalers have redundancy baked in, compliance teams, decades of operational scar tissue. Railway has 30 people and a promise to move faster. Until, of course, they can't.
The $100 million is earmarked for expanding data center capacity and hiring, though exactly how fast Railway can scale remains an open question. The company has integrated Stripe's CLI for one-command deployments and built a template marketplace where creators earn up to 25% revenue share. Documentation updated through May 2026 shows active development on cost controls, storage options, and enterprise features—all the blocking and tackling required to compete.
And the competition is well-capitalized. Render raised $50 million in June 2023, Fly.io closed a $70 million Series C that same month. Vercel, the Next.js juggernaut, secured $250 million in May 2024 at roughly $3.25 billion. Supabase, another developer darling, raised $200 million in April 2025 at around $2 billion. Railway's angle—own the hardware, slash costs, market to builders exhausted by AWS—resonates. Whether 30 people can execute without more multi-hour disasters is the bet investors just wrote a nine-figure check on.
Cooper and his team are probably right that developers are hungry for simpler infrastructure. They might even be right that owning metal unlocks economics the hyperscalers can't match at Railway's scale. But May 19 sits there in the incident log, a reminder that independence and resilience don't always arrive together. The clock is ticking. Eight hours can feel like an eternity when your platform is dark.
