Railway

A developer-focused Platform-as-a-Service (PaaS) for deploying and hosting applications, databases, and services — marketed as an "all-in-one intelligent cloud provider."

Deploy from GitHub, Docker images, or CLI; managed databases (PostgreSQL, MySQL, Redis, MongoDB); persistent volumes; object storage; custom domains with automated TLS; environment management; cron/scheduled tasks; monitoring, logging, alerting; usage-based per-second billing on actual CPU/memory utilization. Runs on self-owned hardware across US, EU, and Asia rather than reselling hyperscaler capacity.

Not an AI product per se; positions itself as infrastructure for AI startups and AI-driven development. Offers API and CLI; emphasizes making cloud infrastructure "invisible" for AI-accelerated development. No first-party LLM features.

Usage-based with plan minimums. Free Trial ($5 one-time credits); Hobby; Pro; Enterprise (SLA, compliance, dedicated infra). Bills per-second on actual CPU/RAM utilization; supports hard spending limits. Customers report significant savings vs. Heroku/AWS.

GitHub, Docker, CLI, REST API; integrates into standard dev workflows.

HIPAA BAA available; role-based access, SSO, extended audit-log retention on higher tiers; Enterprise adds dedicated infrastructure and SLAs/SLOs. SOC 2 status should be confirmed directly. A May 2026 platform-wide outage (~8 hours, caused by Google Cloud suspending Railway's production account) and a December 2025 EU-West build incident were reported.

Founded 2020; HQ San Francisco; founder/CEO Jake Cooper. Raised a $20M Series A led by Redpoint (May 2022); per Railway's January 22, 2026 announcement, raised a $100M Series B led by TQ Ventures with FPV, Redpoint, and Unusual Ventures. Total funding ~$124–130M+. Angel backers include Vercel's Guillermo Rauch and GitHub's Tom Preston-Werner.

Niche/indirect for a VC firm itself — most relevant for portfolio companies or for a firm's internal engineering team building tools/apps. Useful context for technical diligence on developer-infrastructure investments.

Credit/usage-based billing can be unpredictable at scale; background workers require manual workarounds; volume/storage features comparatively newer; reported reliability incidents in 2025–2026; not designed for regulated heavy-compliance workloads out of the box.