Unpopular opinion: Not every startup needs Kubernetes, not even managed Kubernetes.

In my experience as a tech lead, I’ve seen too many early-stage startups overengineer their infrastructure just to “look legit”: setting up Kubernetes clusters when all they needed was to ship a working MVP fast.

Here’s what I’ve learned after trying all the paths:

🚀 Fastest to get started: Use a PaaS like Vercel, Railway, Render, Fly.io, or Heroku. These platforms are amazing for speed and DX, with Git-based deploys and autoscaling. But when your usage grows, so does the cost… fast.

⚖️ Best middle ground: Container-based PaaS like Google Cloud Run or AWS App Runner. A bit more setup, but you get more control and better scalability at a reasonable price. For many startups, this is the sweet spot.

💰 Cheapest with control: Self-host on a VPS like Hetzner, Contabo, OVH, or Scaleway. Install Coolify for a Heroku-like experience with simple CI/CD via GitHub/GitLab. It’s predictable, cost-effective, and works well… until you need to scale fast or handle complex infra needs.

🛡️ Pro tip: No matter what stack you choose, Cloudflare Free Plan is a no-brainer. Basic WAF, DDoS protection, and caching (for $0!) can go a long way for early-stage products.

✨ The lesson: optimize for focus, speed, and sustainability. Don’t waste time (or money) maintaining a complex Kubernetes setup before you even have product-market fit.

Agree? Disagree? I’d love to hear your take in the comments.

Or DM me if you’re wrestling with infra decisions for your startup, happy to chat.