Artificial intelligence has become one of the biggest conversations of our time. Every day, we hear about new AI models, smarter applications, powerful chips and machines capable of doing things that seemed impossible only a few years ago. Companies are investing billions of dollars in AI, governments are developing national strategies, and almost every industry is trying to understand how artificial intelligence will change its future.

When I look at where AI is heading, however, I believe we are focusing too much on the technology itself and not enough on what will eventually make one AI system different from another. In my view, that difference will increasingly come from unique data.

Most artificial intelligence today has learned from an enormous amount of information that already existed. The internet, books, articles, images, research, public databases and decades of digital content have provided AI with an extraordinary foundation of human knowledge. This has produced results that only a few years ago would have been difficult to imagine.

But much of this knowledge is also available to everyone. As AI models become more capable and more companies gain access to similar technology, public information alone will become less of a competitive advantage. The real advantage will increasingly come from information that is not freely available to everybody.

I believe this will create an important shift. The first phase of AI was largely about learning from information humanity had already created. The next phase will increasingly be about creating new information from the real world.

We can already see the beginning of this transition. Cars are collecting information about roads, traffic and driving conditions. Satellites continuously generate information about agriculture, cities, construction and changes on the Earth’s surface. Factories can monitor how machines perform and predict when components are likely to fail. Logistics businesses can understand how goods move between countries, ports, warehouses and consumers. Energy companies can study electricity consumption and changes in demand, while agricultural businesses can collect information about soil, rainfall, temperature, crop health and actual yields.

The physical world is constantly producing information. Until recently, much of that information was either never collected or was difficult to analyse at scale. AI changes this completely. Information that once appeared ordinary can now become extremely valuable when it is collected consistently, organised properly and analysed over time.

This is why I believe businesses need to start looking at themselves differently. Traditionally, we measure companies through their products, revenues, margins, customers and physical assets. All of these will remain important, but another asset is emerging alongside them: the unique knowledge a company creates through its everyday operations.

A logistics company, for example, does much more than move products. After operating for many years, it can develop deep knowledge about routes, delays, transportation costs, ports, seasonal demand and trade flows. A construction company can gradually build knowledge about material costs, labour productivity, project delays, suppliers and maintenance. A mining company can develop years of geological and operational information. An energy company can understand consumption patterns and equipment performance across thousands of locations.

These companies may not think of themselves as data businesses, but in reality they are continuously creating knowledge.

AI gives them the ability to turn that knowledge into intelligence.

This distinction between public knowledge and proprietary knowledge will become increasingly important. Consider agriculture. An AI system can study almost everything publicly available about farming, including research papers, weather information, books and government reports. Another system may have access to exactly the same public knowledge but also possess ten years of information collected directly from thousands of farms, including soil conditions, rainfall, fertilizer usage, temperatures, crop health and final yields.

The second system has an advantage that cannot easily be copied. Its advantage does not necessarily come from better software or a more attractive application. It comes from years of accumulated knowledge about what actually happened in the real world.

This is what makes proprietary data so interesting. Software can be copied. Features can be copied. Business models can eventually be copied. But a high-quality dataset created through years of genuine operations can be extremely difficult to reproduce. Time itself becomes part of the competitive advantage.

For this reason, some of the most interesting AI companies of the future may not initially look like AI companies at all. They may operate farms, warehouses, energy systems, vehicles, buildings, mines or factories. On the surface, they may appear to belong to traditional industries. Underneath those businesses, however, they could be building enormous pools of proprietary information about how the physical world actually operates.

Over time, the distinction between a technology company and a traditional company may become less clear. A logistics company with exceptional proprietary data and strong AI capabilities could effectively become a technology business. The same could happen in agriculture, construction, energy, mining, manufacturing and real estate.

The important asset will no longer be only what a company physically owns. It will increasingly include what that company knows that others do not.

My own thinking about this has developed through experience. Running Maser Group as it has expanded into diversified fields has allowed me to look at businesses across different industries rather than through the lens of a single sector. Industries such as technology, infrastructure, real estate, energy, mining, construction materials and logistics can initially appear completely unrelated. The more closely I look at them, however, the more connections I see.

Every industry generates information. A physical asset generates data. A transaction generates data. A supply chain generates data. A machine generates data. Even operational problems generate data.

A single piece of information may have very little value. When the same type of information is generated thousands or millions of times and organised correctly, patterns begin to emerge. Those patterns can improve decisions, predict problems and eventually create completely new business opportunities.

This has also changed the way I think about diversification. The opportunity is not simply to operate in different sectors. The greater opportunity is understanding how the knowledge created across those sectors can eventually connect.

At the same time, data alone will not define the future of AI. There is an enormous physical infrastructure behind the digital world, and this infrastructure is becoming increasingly important.

AI needs data centres, electricity, land, fibre connectivity, cooling systems, advanced computing equipment and large amounts of capital. As AI becomes more powerful and widely used, these requirements will continue to grow.

This means the AI economy will create opportunities far beyond software. Energy producers, renewable-energy developers, data-centre operators, infrastructure companies, fibre networks and owners of strategically located land will all become part of the ecosystem.

I see two important resources developing alongside each other. Infrastructure gives AI the capacity to compute, while unique data gives AI differentiated knowledge. The combination of the two will be extremely powerful.

Energy will therefore become a major part of the technology conversation. A country may have talented people, available land and ambitious technology policies, but without sufficient reliable electricity, developing AI infrastructure at scale becomes difficult.

Technology policy and energy policy will increasingly become connected. Solar, hydro, wind and other forms of generation are no longer simply environmental investments. They can become part of the digital economy. A power project can support a data centre, a transmission network can unlock an industrial zone, and land close to reliable electricity and fibre can become far more strategically valuable.

The AI revolution does not make physical infrastructure less important. In many ways, it makes physical assets more important.

This is particularly interesting when I look at Africa, where I have spent a significant part of my professional journey. Africa is often discussed as a future consumer market because of its young population, growing cities and increasing adoption of digital technology. I believe its opportunity should be much larger.

Africa has a vast and diverse physical economy that is still developing. Agriculture, mining, energy, logistics, infrastructure, financial services and rapidly growing cities are producing new information every day. High-quality agricultural information collected across different climates, mining information developed through exploration and production, logistics information showing how goods actually move across borders and ports, and energy information showing where demand is growing could all become increasingly valuable.

Much of this knowledge does not already exist in an organised form on the public internet. This gives Africa an opportunity not simply to consume artificial intelligence developed elsewhere, but to create new knowledge that the rest of the world may eventually need.

The same opportunity exists around infrastructure. Africa has land, natural resources and significant renewable-energy potential. If these advantages are combined with reliable electricity, digital connectivity, data centres and locally generated information, some African markets could participate much more deeply in the global AI economy.

I also see a natural connection between the Gulf and Africa. The Gulf has capital, infrastructure experience, international connectivity and increasingly ambitious technology strategies. Africa has scale, resources, land, energy potential and major infrastructure requirements. These strengths can complement one another.

The Gulf itself demonstrates the importance of building foundations before expecting industries to grow. Dubai and Abu Dhabi did not become globally connected economic centres simply because technology arrived. Ports, airports, roads, telecommunications and reliable power were developed first. These physical systems created the environment in which other industries could grow.

The same principle applies to AI. Digital economies ultimately depend on physical foundations.

As companies begin to recognise the value of data, however, responsibility will become equally important. The future cannot simply be about collecting as much information as possible. Privacy, consent, ownership and security must remain fundamental.

There is also an important distinction between personal information and operational information. A machine’s performance, a crop’s yield, a building’s electricity consumption, geological information or a shipping route can create enormous value without requiring companies to invade people’s private lives.

The objective should be better intelligence, not unlimited surveillance.

In fact, I believe trust itself will become an important competitive advantage. Companies that build valuable datasets responsibly and protect that information properly will be in a much stronger position over the long term.

The growing importance of proprietary information may eventually influence how businesses themselves are valued. Today, investors place value on physical assets, intellectual property, brands, customers and future cash flows. In the future, high-quality proprietary datasets could become a much more important part of that calculation.

Two companies may have similar revenues and similar physical assets, yet very different long-term potential because one has spent years building structured information that can improve decision-making and train specialised AI systems.

That information can help predict demand, reduce costs, identify risks earlier, improve pricing, make machines more efficient and potentially create entirely new products.

At that point, data stops being simply a record of what happened yesterday.

It becomes a tool for understanding what could happen tomorrow.

Perhaps the most interesting aspect of this opportunity is that the best datasets cannot always be created quickly. Some require years of consistent operations and collection.

This reminds me of infrastructure investing. The most valuable infrastructure is often built before demand fully arrives. Ports are expanded before trade reaches its peak. Power plants are built before factories require all their electricity. Roads are constructed before cities completely develop around them.

Unique data may work in a similar way.

The companies building these datasets today may be creating an asset whose full value will only become obvious years from now. By the time everybody recognises its importance, recreating the same history may be impossible.

Nobody knows exactly what artificial intelligence will look like ten years from now. Models will become more powerful, computing will become more efficient, new companies will emerge and some of today’s leaders may disappear.

But I believe one principle will remain.

AI is only as useful as the infrastructure supporting it and the information available to it.

The first stage of artificial intelligence benefited enormously from the knowledge humanity had already placed online. The next stage will increasingly move beyond that existing information and into the physical world.

Companies will create information through machines, farms, buildings, vehicles, supply chains, energy systems, mines and infrastructure. They will organise it, protect it and use artificial intelligence to turn it into knowledge that did not exist before.

Twenty years ago, companies competed for distribution. Then they competed for internet users. Today, they are competing for computing power and AI talent.

Tomorrow, I believe they will increasingly compete for unique data.

Unlike information that already exists across the internet, the best proprietary data cannot simply be downloaded. It has to be created, collected responsibly, organised and improved over time.

That is why I believe one of the most valuable resources of the next economic era may not be something buried underground or stored in a warehouse.

It may be the unique knowledge that businesses quietly create every single day.

The next valuable resource will be unique data.