Stop watching the looms: why the AI era belongs to infrastructure

I live in Manchester, England now. I moved here from Texas last summer (which is its own story), but the thing I wasn't prepared for is how the Industrial Revolution isn't history here. It's the city itself. And if you're American like me, you might need to hear this: the Industrial Revolution didn't start in the US. It started here. Manchester is where the modern world was born.

You see it everywhere. The old cotton mills converted into apartments. Warehouses that once stored raw materials now house coffee shops and coworking spaces. Canal paths where barges carried thirty tons of goods are running trails. The bones of the revolution are still holding the city up. You can't walk three blocks without stepping on a piece of it.

And the longer I live here, the more I realize we're making the same mistake with AI that people made with cotton.

Everyone's watching the wrong thing

Here's what most people know about the Industrial Revolution: it started somewhere around 1760, someone invented a machine, it changed everything, boom, modern world. The cotton gin. The spinning jenny. The power loom. Pick your favorite invention, credit it with transforming civilization, move on.

But that's not what happened.

The cotton gin was patented in 1794 by Eli Whitney in the American South. The power loom had been around since the 1780s, invented in England. America grew the raw cotton. But it was Manchester that figured out how to turn it into an industry that changed the world. These were extraordinary inventions. They mechanized work that had taken human hands countless hours. But Manchester didn't become the heart of the Industrial Revolution because someone built a clever machine.

Manchester became Manchester because of the infrastructure that powered the machines.

The Bridgewater Canal opened in 1761, before the cotton gin even existed. It connected Manchester to the coal mines at Worsley, and suddenly a single horse towing a canal barge could move ten times the cargo of a horse pulling a cart. Then came the Rochdale and Ashton Canals, linking the textile towns across Lancashire and Yorkshire. Then the Manchester Ship Canal in 1894, the largest river navigation canal in the world, which turned a landlocked city into an inland port.

The Liverpool-Manchester Railway opened in 1830. Not the first railway in England. The first inter-city railway in the world to run exclusively on steam power, the first to be double-tracked throughout, the first with a signaling system, the first with a full timetable. It connected Liverpool's port to Manchester's factories, and within decades, a rail network stitched together every major city in the country. The United States wouldn't complete its transcontinental railroad for another 39 years.

By 1830, Manchester had 99 cotton-spinning mills. It was producing so much that Friedrich Engels came here to study the conditions, and some would say his observations became the foundation for modern labor economics. By 1910, Trafford Park had become the world's first industrial park, and the influence had reversed: now American companies like Ford and Westinghouse were coming to Manchester to set up shop.

The machines were the catalyst. But the canals, the railways, the warehouses, the factories, the logistics networks, the signaling systems... that infrastructure was the revolution. The machine told you what was possible. The infrastructure made it real.

Sound familiar?

We're calling it the wrong name

Right now, the world is fixated on LLMs and AI agents the way the history books are fixated on the cotton gin. Everyone's watching the machine.

And look, the machines are extraordinary. I'm not dismissing that. Large language models can reason, write, code, analyze, and synthesize at a scale that was science fiction five years ago. Agents are starting to act autonomously, chaining together tasks, making decisions, operating in the real world. This is genuinely remarkable technology.

But if you think the AI revolution is about the models, you're making the same mistake as someone in 1790 thinking the Industrial Revolution was about the loom.

The models are the cotton gin. They're the catalyst. They show us what's possible, that machines can now do cognitive work the way the loom showed us machines could do physical work. But the real revolution? The thing that will actually reshape industries, economies, and how we live? That's the infrastructure being built around them.

People keep calling this "the AI revolution." I think that's like calling the Industrial Revolution "the cotton gin revolution." It names the spark and ignores the fire.

What we're actually living through is an industrial revolution. Not a metaphorical one. A literal one. The pattern is identical: a breakthrough invention mechanizes a category of human labor, and then an explosion of infrastructure, systems, and platforms is required to turn that invention into widespread, sustainable transformation. The first time, it was physical labor. This time, it's cognitive labor. But the shape of the revolution is the same.

I've started calling it the Cognitive Industrial Revolution. Not because the phrase is catchy, but because it's accurate. It forces you to see the whole picture, not just the machine. It reminds you that revolutions aren't about the invention. They're about the world you have to build around it.

The part nobody's talking about

Here's what I keep seeing in the AI conversation right now: everyone's debating which model is best, which agent framework will win, whether GPT or Claude or open-source will dominate. Those are interesting questions. They're also the wrong ones.

The question that matters, the one that mattered in 1780 and matters right now, is: who's building the canals?

Because AI can write code. It's getting frighteningly good at it. But code isn't an application. An application is code plus hosting, plus routing, plus databases, plus environment management, plus deployment pipelines, plus scaling, plus security, plus observability, plus a dozen other things that have to work together seamlessly for any of it to matter.

AI is about to dramatically increase the volume of software being created. More people building. More ideas shipped. More code written in a weekend than a team used to produce in a quarter. That's the cotton gin moment: the mechanization of code creation.

But every one of those applications needs somewhere to run. Every one of them needs infrastructure. And that infrastructure can't be the same sprawling, duct-taped, manually-configured mess that most companies are running today. The volume alone would break it.

This is the part of the revolution that's barely in the conversation. Everyone's racing to build better models and smarter agents. Almost nobody is asking: when a million new AI-generated applications need to go live next year, what runs them?

Manchester knew this before anyone

Here's where the story gets almost too perfect.

Manchester didn't just birth the Industrial Revolution. It birthed artificial intelligence too. And I didn't know that when I moved here.

In June 1948, a machine called the Manchester Baby ran the world's first stored program. It was seventeen feet long, weighed a ton, and took 52 minutes to complete a calculation. It was built at the University of Manchester by Frederic Williams, Tom Kilburn, and Geoff Tootill. Within a year, it evolved into the Manchester Mark 1, which became the prototype for the Ferranti Mark 1, the first commercially available general-purpose computer.

Then Alan Turing moved to Manchester. In 1950, working at the university, he published "Computing Machinery and Intelligence", the paper that asked "Can machines think?" and introduced what we now call the Turing test. The foundational question of artificial intelligence was asked here. In Manchester. In the same city where the cotton mills had reshaped the world a century before.

I walk past the buildings where this happened. The same city that built the canals and railways that made the Industrial Revolution possible also built the machines and asked the questions that made AI possible. The DNA of this place keeps producing revolutions.

And both times, the pattern is the same. The breakthrough gets all the attention. The infrastructure does the actual work.

This is why I'm excited about where I am

I work at Upsun. And what we do maps so precisely onto this pattern that it almost feels like the metaphor was built for us.

Upsun is an application platform. In plain terms: we're the infrastructure layer that takes your code, whether a human wrote it, an AI wrote it, or both, and turns it into a running, scalable, secure application. Hosting, routing, databases, environment management, deployment, scaling, all of it, handled intelligently by design, without the overhead of cobbling together a dozen different services and hoping they play nice together.

Apps are more than just code. AI is extraordinary at generating code. But code sitting in a repository isn't an application any more than raw cotton sitting in a warehouse is a shirt. You need the systems, the canals, the railways, the mills, to turn the raw material into something useful.

That's what Upsun is. We're building the infrastructure that makes the Cognitive Industrial Revolution actually work. Not the cotton gin. The canals. Not the model. The platform.

And in a world where AI is about to increase the volume of code and applications by orders of magnitude, the infrastructure question becomes the only question that matters. Raw cotton didn't change the world sitting in a warehouse. It changed the world when the canals, railways, and mills connected it to the people who needed it. The application is the product. Infrastructure is the canal that gets it to the world. That was true in 1830. It's true now.

The same mistake, the same opportunity

I walk past the old mills on my way to get coffee. They're beautiful. Red brick, massive, built to last centuries. And they did last. But here's the thing I keep coming back to: the looms are gone. Every single one of them. The machines that started the revolution, the ones the history books obsess over, were replaced and scrapped and forgotten generations ago.

The warehouses are still here. The canals still run. The railways still carry passengers through Piccadilly and Victoria every morning. The infrastructure outlived the machines by two hundred years.

I think about that when I think about what we're building now. The models everyone's racing to build today will be replaced (some in a few weeks at this rate). The AI that feels miraculous in 2026 will feel quaint in 2036. Code changes. Tools evolve. The thing that looked like magic becomes a commodity.

But infrastructure lives on. The platforms, the systems, the invisible architecture that connects code to the world and keeps applications running... that endures. Because it was never about the machine. It was about what the machine needed to be useful.

The Industrial Revolution didn't start with the cotton gin. It started with the canal that connected the mine to the city.

The Cognitive Industrial Revolution isn't starting with the LLM. It's starting with the platforms that connect the code to the world.

I happen to live in the city that figured this out first. And I happen to work at the company that's figuring it out again.