The Expanding Surface Area of Software

The Expanding Surface Area of Software

Raghib MurtRaghib Murt·March 19, 2026
Related post →

Many people think software engineering as a craft will become less important in the coming years.

AI writes code.
Agents build tools.
Anyone can ship software, right?

So the assumption is obvious: Software will matter less.
But the data and the direction of the world suggest something very different.
The surface area of software is expanding faster than most people realize.
Not because engineers suddenly became more productive.
Because the number of problems that want to become software is exploding.

Three forces are pushing this forward.

First is the sovereignty pressure.
For years, the world ran on a handful of cloud providers and platforms.
Now governments and regions are realizing how fragile that dependency is.
Countries want domestic cloud capacity.
Domestic AI models.
Domestic payment rails.
Domestic data infrastructure.
Every push toward technological independence multiplies new software systems.

Second, institutions are being refactored into software.
Industries were the first wave. Media. Retail. Finance.
Now the deeper layers are changing.
Government systems.
Healthcare infrastructure.
Tax platforms.
Courts.
Education.
These were slow, bureaucratic, paper-heavy systems for decades.
Now they’re being rewritten as software.
And every refactor spawns integrations, platforms, compliance layers, and entire ecosystems of tools.

Third, AI collapses the cost of founding.
This one is subtle. A single builder with AI can now do work that used to require a small team.
That doesn’t mean every builder creates a unicorn.
But it does mean something important.
The number of attempts explodes.
More experiments. More tools. More micro-products.
More weird little software businesses solving narrow problems.

The result is a strange paradox.
The importance of software systems actually increases.
Because when the cost of creation drops, the number of systems explodes.
Software will become the default way humans coordinate work in.
And in that world, the bottleneck won’t be writing code.
It will be distribution, trust, regulation, and legitimacy.