Michael Gustafsson, now Solutions Engineer at Cloudflare and previously the steady hand behind Swedish government core rebuilds, has seen companies hit this wall at every stage. At October’s CTO Breakfast, we went deep on the inflection point where infrastructure stops being ‘just plumbing’ and becomes a real competitive moat.

Why Listen to Michael?

• Led national-scale rebuilds under public scrutiny, and lived to tell.
• Partner to Europe’s fastest-growing CTOs at Cloudflare.
• Advocates for discipline, transparency, and the right kind of speed.

We caught up to talk about the point where infrastructure stops being plumbing and starts being strategy.

You’ve worked in both public and private sectors. Which one taught you more about moving fast without breaking things?

The public sector, easily. People assume it’s slow, but the discipline is the point. Strong guardrails, clear governance, predictable patterns. It forces you to build correctly from the start.

Startups romanticize speed, but most of them mistake motion for progress. They hack it together, tell themselves they’ll fix it later, and never do. Then you end up with an architecture no one fully understands.

The public sector showed me that speed built on shaky ground isn’t speed, it’s deferred pain. If you want to move fast, start by building something that can survive it.

Infrastructure used to be something investors ignored. Now it’s part of the pitch. When did that flip for you?

When I joined Cloudflare. That’s when I saw infrastructure being used as strategy, not support.

Customers weren’t just buying reliability, they were buying trust, reach, and confidence that the platform would stay fast under pressure.

Your architecture is your velocity. It decides how often you can deploy, how fast you can respond, how safely you can experiment. The smartest companies I work with understand that infra isn’t a cost line, it’s leverage.

If you’re still treating it as plumbing, you’re already behind.

Every scale-up says they’ll handle tech debt later. You’ve seen it go wrong, what’s the pattern?

Tech debt doesn’t kill you in a straight line, it bleeds you out slowly. You lose observability, repeatability, and institutional memory. Suddenly every release is a risk assessment. The common thread is lack of discipline early on: no infra-as-code, no consistent patterns, no documentation. When you don’t codify how you build, you can’t scale how you build.

The teams that do it right treat automation and standards as force multipliers. That’s not bureaucracy, it’s how you buy future speed.

What’s the most expensive mistake you’ve made?

"Cloud cost blindness."

I’ve seen companies spend absurd amounts of money on traffic they didn’t even know they were generating. Not because they’re reckless, just because the systems became too complex to reason about. AWS taught me that lesson the hard way. You can architect your way into six-figure bills without realizing it.

That’s why I value Cloudflare’s “scale to zero” principle. If you don’t use it, you don’t pay for it. Simplicity scales. Cost visibility is a design requirement, not something finance should figure out after the fact.

One infrastructure trend that’s overrated, and one that’s underrated, for European builders?

Overrated: Multicloud from day one. It’s complexity cosplay. You’re adding cost, latency, and governance overhead before you even have product-market fit. Get one stack solid, learn how to run it well, then expand.

Underrated: Edge-native architectures. Running closer to users, reducing ops load, building once and deploying everywhere, that’s a real advantage. And Europe, with its data laws and latency constraints, benefits more than most regions from that model.

Edge-native isn’t just faster: it’s simpler, cheaper, and safer when done right.

What’s something you stopped doing as a technical leader that made your team better?

Writing clever code.

The older I get, the less I care about elegance and the more I care about repeatability. I used to want to be the one who solved the hardest problems personally. Now I’d rather design a framework that prevents those problems entirely.

Hero engineering doesn’t scale. Teams that rely on heroes burn out; teams that rely on systems compound velocity. If your team’s success depends on one person’s brilliance, you don’t have a system, you have a liability.

Three years from now, what infrastructure decision will separate Europe’s unicorns from its casualties?

The ones that learn to build with, and around, AI.

We’re heading toward “agentic” infrastructure: systems that generate, deploy, and retire themselves based on intent. That’s going to compress build cycles from months to hours. The real challenge won’t be capability, it’ll be governance. Who owns the guardrails? How do you control cost, security, and compliance when code writes itself?

The companies that solve that will lead. The rest will be buried in their own automation debt.

Final thought?

What makes Michael interesting isn’t his résumé, it’s the mindset. He treats infrastructure as product, culture as leverage, and discipline as a multiplier. In a world chasing “move fast,” he’s reminding European builders what keeps speed sustainable.
That’s exactly the kind of voice we built this network for: not loud, just signal.