The internet looks distributed.
Millions of websites.
Thousands of services.
Independent systems everywhere.
But underneath, it’s not that distributed.
A large part of the internet runs on a small number of cloud providers.
Not by design.
By accumulation.
A few providers dominate the infrastructure layer
Today, most internet services rely on cloud infrastructure.
And that infrastructure is highly concentrated.
Amazon, Microsoft, and Google together control roughly 60–70% of the global cloud market
Individually:
- AWS ~30%
- Azure ~20%
- Google Cloud ~13%
Everything else is fragmented.
Which means:
a large part of the internet depends on a small number of systems.
Cloud replaced ownership with access
Before cloud computing, companies ran their own infrastructure.
Servers.
Data centers.
Networking.
Now they rent it.
Cloud providers offer:
- instant scalability
- global distribution
- managed infrastructure
So instead of building systems, companies connect to them.
This shift is what made centralization practical.
Convenience scales faster than independence
Running your own infrastructure is possible.
But it’s harder.
Cloud platforms remove friction:
- faster deployment
- lower upfront cost
- built-in reliability
So the rational decision is simple:
use what already works.
And when millions of companies make that decision,
the result is concentration.
Optimization drives everything toward the same place
No one chooses a cloud provider randomly.
They optimize for:
- cost
- latency
- reliability
- ecosystem
And optimization tends to favor the largest providers.
Because they:
- have more regions
- offer more services
- integrate more tools
This follows the same pattern described in
Optimization Systems and Unintended Consequences.
Each decision is local.
The outcome is systemic.
Dependence turns providers into infrastructure
Once a system is built on a cloud platform,
moving away becomes difficult.
Not impossible.
But expensive and risky.
As described in
When Daily Life Depends on Software Infrastructure:
systems become essential.
And the platforms they run on become critical infrastructure.
Centralization increases efficiency — and risk
Large cloud providers make the internet faster and more reliable.
But they also create concentration points.
Which means:
when something goes wrong,
it affects a lot of systems at once.
This is exactly the fragility described in
The Fragile Infrastructure Behind “Always Online” Services.
Failures scale with the provider
When a small service fails, impact is limited.
When a cloud provider fails, impact is widespread.
A single issue can cascade into:
- outages
- service degradation
- global disruptions
As shown in
How Small Infrastructure Failures Become Global Outages
and
Why Modern Systems Fail All at Once.
Centralization wasn’t planned — it emerged
No one decided:
“a few companies should run the internet.”
It happened through:
- cost optimization
- infrastructure complexity
- developer convenience
- scaling requirements
Each decision made sense individually.
Together, they reshaped the system.
What this actually means
The internet is still technically decentralized.
But functionally, it depends on a few providers.
A few cloud platforms run most of the internet
not because they control it,
but because everything chose to run on them.