Why data, not code, is fuelling the AI revolution

In the 19th century, steel was the defining raw material of progress. It made the railroad possible, reshaped cities with skyscrapers and powered the Industrial Revolution. The steel industrialist Andrew Carnegie understood, from first principles, that whoever controlled this raw material would control the future and bet everything on it. That bet didn’t just build an empire. It helped the United States win the Industrial Revolution.

Today, as leaders gather in Davos for the World Economic Forum Annual Meeting, the same dynamic is unfolding with artificial intelligence (AI). We’ll hear about breakthrough models and trillion-dollar forecasts, but the modern raw material isn’t compute or algorithms. It’s data. And the nations and industries that understand this, who invest in the substrate AI runs on, will define the next century, just as steel once did.

But behind the optimism is a hard truth; 95% of AI pilots fail, according to MIT. Not because the algorithms are weak, but because the data is. We’ve built extraordinary digital intelligence, yet we’re asking it to operate with an incomplete view of the physical world. This is like trying to build a transcontinental railroad without sufficient steel to lay the track.

The AI revolution will only happen if we turn the world from analogue to digital and if we capture the physical truths in critical industries that underpin our economy – energy, manufacturing and shipbuilding. In every major sector, the nations and companies that master the transformation of the built world into ground-truth digital data will be the ones that win the AI revolution. Everyone else will simply be spectators.

As Carnegie won the Industrial Revolution with steel, the AI revolution will be won with data.

The metal, motors and concrete that make up the real economy, including our power plants, factories and shipbuilding industry, remain largely invisible to AI.

We still depend on people doing manual inspections, scribbling notes or taking photos from scaffolds and confined spaces. Most of our data is trapped in PDFs, spreadsheets and binders. In effect, AI is trying to see through a keyhole.

That’s why so many industrial AI projects struggle. We’re attempting to automate or optimize systems we haven’t truly digitized. It’s like trying to drive a car with the headlights off or operate a refinery with no gauges. Intelligence demands visibility.

Across industries, the AI conversation remains fixated on algorithms. Better models, faster chips, smarter code. But the frontier that matters most is measurement, the same way the frontier for Carnegie wasn’t the train, it was the steel required to build the track.

Modern AI can reason, plan and simulate with astonishing power if it’s grounded in real, reliable data. But the physical world still operates, in a data sense, like the 19th century. Infrastructure ages, corrodes, vibrates and fails, but very little of this is captured at the fidelity AI needs.

The result is predictable: AI models are deployed into blind spots. They speculate where they should be sensing. They extrapolate where they should be measuring. And we wonder why they fail.

So, how do we get AI the raw material it needs? That’s where robots come in and why they must work in tandem with AI. Robots in the field can capture immense volumes of real-world data: ultrasonic, vibration, lidar, thermal, visual. They turn physical infrastructure into high-fidelity digital models.

This is why my organization, Gecko, is making the same first-principles bet Carnegie made. He focused on steel because it was the substrate of the Industrial Revolution. We are focused on data because it is the substrate of the AI revolution. Without it, nothing moves forward.

A second, equally important corpus of data is human expertise. The knowledge living in the hands and heads of operators, inspectors and maintainers. This is part of the raw material AI must learn from too.

When AI is grounded in high-fidelity, structured data, everything changes. Instead of guessing at performance, it can calculate it. Instead of reacting to failures, it can prevent them. Instead of extrapolating, it can understand.

AI can detect losses no human could see, the tiny vibrations, temperature shifts or material stresses that signal inefficiency or danger long before they become problems. It is the difference between binders of inspection reports stuffed in filing cabinets and a complete digital map of critical infrastructure.

This is where value is created: turning atoms into bits. In infrastructure, that means longer asset life and higher reliability. In defence, more operational time at sea. In manufacturing, higher yield and lower waste.

This isn’t about replacing people, it’s about giving them x-ray vision and making them superheroes.

Every great leap in human progress has required new infrastructure. Steam needed steel. Electricity needed copper. The internet needed fibre.

AI needs data.

But not just any data, ground-truth, structured, high-quality data that connects the digital world to the physical one. That’s what turns AI from a curiosity into a tool of transformation. Just as steel rails enabled the movement of goods and ideas across continents, this new layer of data infrastructure will enable the movement of intelligence across every sector of the economy.

Without it, AI remains an engine without rails: powerful, but going nowhere.

At the World Economic Forum Annual Meeting, I hope every leader leaves with one simple question:

'Do we have the data on our assets to make our systems intelligent?'

Because the future won’t be written by algorithms alone. It will be built and rebuilt, on data.

The rails of the AI revolution aren’t made of code. They’re made of information. The raw, physical truth about how our world works.

Just as steel helped America lead the Industrial Revolution, data will determine who leads the AI revolution.