Until recently, AI was viewed primarily as a product domain. Investors focused on start-ups, teams and companies developing services based on machine learning and neural networks. Over time, however, it became clear that such stand-alone initiatives are not sufficiently resilient to support long-term growth. The success of an individual solution does not guarantee scalability; without infrastructure, talent and durable demand, these businesses can be highly vulnerable in the first downturn.

Leading market participants have quickly recognised a crucial point: investment in AI works only when capital is deployed not simply into technologies, but into entire ecosystems. AI cannot be developed in isolation. It requires a full value chain—from energy and computing capacity to human capital, educational institutions, and attractive places to live and work. This is why the new investment logic is built around a number of interconnected components.

The first of these is the creation of data centres. These are no longer merely technical server facilities; at scale, they resemble cities in their own right. Demand for data storage and processing is growing rapidly, and delivering the required computing capacity calls for vast quantities of microprocessors, graphics processing units, servers and non-ferrous metals. A single large data centre, for instance, may require up to 10,000 kilometres of copper cabling.

This entire infrastructure also depends on affordable, reliable energy. According to various estimates, powering global IT infrastructure at its current growth rate will require more than 7,000 TWh of electricity per year—roughly equivalent to the combined annual consumption of the European Union and the United States.

The energy architecture itself must be multi-layered, consisting of several lines of resilience. The first layer comprises large-scale power plants, providing the principal and most stable source of electricity. The second layer consists of gas- or diesel-powered generators. While less efficient, they can keep critical infrastructure running for extended periods if the primary electricity supply fails. The third layer is a passive reserve based on capacitor storage systems, designed to cover the load briefly in emergency situations.

Such an architecture becomes indispensable as AI is embedded more deeply into critical systems, because any interruption in power supply can bring the entire system to a halt. When essential data and processes are managed by artificial intelligence, there is no realistic ability to fall back on human resources—there is simply no “manual mode” built into the system. Without guaranteed continuity of power, a full transition to AI is impossible.

Once this level of energy resilience is in place, large-scale AI deployment becomes feasible. The recent example of London Heathrow Airport, which was paralysed for two days by an energy-related disruption, is a reminder of why such failures become unacceptable in an AI-dominated infrastructure environment.

Rare earth metals also play a central role in the development of AI infrastructure. They are used in processors and chips, data-storage equipment—particularly hard drives—and image and sound recognition systems. Rare earth elements are also critical for data centres, where they are needed in cooling systems and telecommunications equipment.

It is therefore unsurprising that US President Donald Trump has actively promoted policies aimed at expanding rare-earth extraction and processing in the United States, while also pursuing an interest in Greenland, whose territory is rich in these materials.

The next strategic component is human capital. Its full development requires an environment in which talent can be born, educated, live and work. This creates a need for so-called “cities of the future”: purpose-built environments combining universities, STEM schools, technology-company campuses and high-quality urban infrastructure. These cities do not merely attract talent; they retain it by creating dense ecosystems for shared growth and development.

Serbia offers one example of this strategy, having made AI development a matter of national policy. AI has been introduced as a compulsory subject in general school curricula, while local authorities have invested in a National AI Platform, including a supercomputer available free of charge to universities, research institutes and local start-ups.

The United Arab Emirates provides another notable example. Competition between emirates is being used to attract highly qualified professionals through Golden Visas and other immigration incentives. By creating an investment-friendly and talent-friendly environment, the UAE is laying essential foundations for the emergence of integrated AI ecosystems.

The defining feature of this new approach is its focus on systemic investment in self-contained digital environments. These ecosystems include everything: infrastructure, energy, talent and products. Investors are increasingly recognising that efficiency and returns cannot be achieved without this systemic approach.

If there is a shortage of qualified specialists, data centres will remain underutilised. If there is insufficient energy, data-storage and processing systems cannot operate. If there are no cities and liveable urban environments, there will be no people willing to live and work there. Each element is interdependent, and only their coordinated development can produce sustainable success.

This concept is replacing an older paradigm in which capital was allocated to narrow sectors such as banking or green energy. For example, in 2022, around 1,500 institutional investors, collectively managing US$40 trillion in assets, withdrew from investments in the hydrocarbon sector. Today, however, a portion of this capital is being redirected not so much towards green energy as towards resilient AI infrastructure—including rare-earth materials, advanced electronics and the energy systems that support them.

The financial sector is no longer at the forefront either: the share of institutional-investor allocations to financial services declined from 16% in 2018 to 14% in 2024. By contrast, the technology sector has gained significant weight in recent years, rising from approximately 13% to around 28%. The real investment opportunity is increasingly concentrated in ecosystems built around AI.

Europe is also mentioned less and less often among the leading investment destinations. A fading green agenda, fragmented regulation and stagnant growth are making the region progressively less attractive. China, along with other ambitious emerging markets in Asia and the Middle East, is moving to the forefront—particularly those capable of offering the complete cycle required to build an AI ecosystem.

We may be on the verge of seeing new Silicon Valleys emerge—not in California, but in less obvious parts of the world. The countries that can create the conditions for the integrated development of these ecosystems will become magnets for capital and technological progress.

These are the investments of the future: not merely investments in the AI sector, but investments in the world in which AI can live, develop and operate. Ultimately, wherever the infrastructure for AI emerges, the future emerges with it.