Over the past few months, I’ve been spending a lot of time studying cloud infrastructure companies. Not just AWS, Azure, and Google Cloud, but also players like DigitalOcean, OVHcloud, Hetzner, and newer regional providers.

The obvious conclusion would be: the cloud market is already dominated.

But that’s not what I’m seeing.

What I’m actually seeing is a market that’s splitting into very distinct layers, and that split is creating new opportunities for anyone thinking about building a cloud service provider today.

This newsletter is my attempt to simplify that into a practical playbook.

Most people still think cloud is about scale.

More servers
More regions
More services

That model worked in the past.

It does not work anymore for new entrants.

Today, cloud is about focus and positioning.

There are essentially two worlds now.

On one side, hyperscalers like AWS, Azure, and Google Cloud. They are building vertically integrated platforms with their own chips, AI stacks, and global infrastructure.

On the other side, a new wave of providers that are not trying to compete on breadth at all.

Instead, they are doing something very specific and doing it well.

And that’s where the opportunity is.

If I had to simplify everything I’ve seen, every successful cloud provider today fits into one of these four buckets.

This is all about trust.

Governments and regulated industries care less about features and more about:

If you are building in a market like Europe or India, this model is extremely relevant.

This is probably the most interesting segment right now.

There is massive demand for GPU compute, and not every company wants to deal with hyperscaler complexity or pricing.

The providers winning here are focusing on:

This is where simplicity wins.

If you’ve ever used AWS for a small project, you know how quickly things get complicated.

Providers like DigitalOcean built their entire business on removing that complexity.

The value here is not more features. It’s clarity.

This is more enterprise-driven.

A lot of companies are moving away from legacy setups but are not ready to go all-in on hyperscalers.

They want control, but they also want managed infrastructure.

That’s where this model fits.

One thing that becomes very clear when you study these companies is that the stack choice is not just technical. It’s strategic.

Here’s the simplified version of how a modern CSP is structured:

Every provider is essentially building some variation of this.

If you’re building today, there are two dominant directions.

Used mostly for sovereign and traditional IaaS models.

It gives you control, but it comes with operational complexity.

This is what I see more and more new providers choosing.

It aligns better with modern workloads, especially AI.

Faster to build
Easier to scale
More developer-friendly

If I had to start from scratch, I would not try to cover everything.

I would do something very specific.

Something like:

Start with:

Then layer on:

This part is important.

Basic compute is becoming a commodity.

The real revenue is moving toward:

So if you’re building only VM infrastructure, you’re competing in the lowest-margin segment.

This is not just a technical problem.

It’s a positioning problem.

The companies that are winning are not the ones with the best infrastructure.

They are the ones with the clearest answer to:

“Who is this cloud for?”

Cloud is not a closed market.

It’s evolving.

The rules have changed from:

“Who has the most infrastructure”

to:

“Who solves a very specific problem better than anyone else”

Hyperscalers dominate scale.

But there is still space for focused players who understand exactly where to compete.

If you’re thinking about building in this space, this is probably one of the most interesting moments to do it.

Not because it’s easy.

But because the market is no longer one-size-fits-all.