Engineering

Why we built our own hardware instead of renting hyperscaler capacity

Renting capacity is the fastest way to launch. Owning the iron is the only way to control jurisdiction, margin and operations. We chose the harder path.

The easiest way to start a cloud today is to own nothing. Rent capacity from a hyperscaler, put an API on top, take a margin. We chose the opposite: our own servers in our own racks in Falkenberg. Here’s why.

Rented capacity inherits someone else’s constraints

When you rent capacity on another provider’s stack, you inherit their jurisdiction, their pricing model and their operational decisions. If their parent is subject to non-EU law, your infrastructure effectively is too. Sovereignty can be subcontracted away into a layer you don’t control.

Owning the iron gives you three things

  • Jurisdictional cleanliness — the hardware is owned by a Swedish EU company, all the way down to the disk.
  • Predictable economics — we don’t pay a hyperscaler’s margin, so pricing reflects actual cost, not markup on markup.
  • Control over operations — we decide hardware generations, redundancy and maintenance windows ourselves.

You can’t promise sovereignty on hardware you don’t own. Everything else is a promise of someone else’s promise.

— from Kepler’s engineering principles

The price we pay

Owning hardware is more expensive to start. It takes capital, rack space, cooling and people who know the iron. We took that cost deliberately, because it’s one-time — and what it buys, control over the whole chain, can’t be bought after the fact. That’s the difference between running a cloud and reselling one.

Want to see what it means in practice? Read about the platform or get started today.

Bygg på ett moln du faktiskt äger.

Ingen bindningstid. En körande VM innan du betalar en krona.