Powering the Intelligence Age: AI Data Centres, Energy Security and the Future of Tanker Trade

The next great infrastructure race is not only about computing power. It is also about electricity, fuel, geography and the ships that connect energy producers with the places where intelligence is being built.

We speak about artificial intelligence as if it exists in a world of its own: algorithms, processors, software and extraordinary computing capacity. Yet behind every prompt, every model and every new application there is a very physical reality. AI requires buildings, land, cooling systems, transmission lines, power stations and a dependable supply of energy, every hour of the day.

In my view, this is one of the most important infrastructure questions of the coming decade. The race to develop AI will increasingly be shaped not simply by who can design the most advanced chip, but by who can secure the electricity to operate it, where that electricity can be delivered, and how reliably it can be sustained. This is also where the energy and maritime industries meet.

The scale of the electricity challenge

The International Energy Agency’s 2025 analysis projected that global data-centre electricity consumption could rise to around 945 terawatt-hours (TWh) by 2030, more than double the level at the time. Its subsequent 2026 update reported that electricity use by data centres rose by 17% in 2025, while AI-focused facilities grew even faster; it expects total data-centre electricity demand to double by 2030 and demand from AI-focused centres to triple.

These are projections, not certainties. Efficiency improvements could moderate the growth, while new AI applications could increase it. The important point is the direction of travel: data centres are becoming large, concentrated electricity customers, and their demand can arrive much faster than conventional power infrastructure can be planned and built.

We should also remember that electricity is only part of the energy equation. A data centre needs cooling, backup power, transformers, batteries, water systems and a resilient connection to the grid. A site may have capital, land and fibre connectivity, but without a credible power solution it is not a viable AI hub.

Where are the data centres being built?

The United States and China remain the two largest data-centre markets. In the United States, major clusters have developed around established technology and connectivity centres, including Northern Virginia, Texas, the Pacific Northwest and parts of the Midwest and Southeast. Access to fibre, customers, skilled labour and existing grid infrastructure has been important. Increasingly, however, developers are also looking for available land and power capacity beyond the traditional hubs.

China’s facilities are concentrated around major economic and technology regions, while policy is also encouraging development in western areas where renewable resources may be more abundant. That creates a different infrastructure challenge: connecting remote generation and computing demand through substantial transmission investment.

Europe has important clusters in the Frankfurt–Amsterdam–London–Paris–Dublin corridor and in the Nordic countries. The Nordics can offer cooler climates and access to low-carbon electricity, while the major European hubs benefit from connectivity and proximity to business customers. But grid congestion, planning constraints, land and water availability increasingly influence where projects can proceed.

Japan and South Korea are investing in digital infrastructure close to major industrial and consumer markets. Singapore remains a significant connectivity hub, although land and energy constraints have encouraged a more selective approach. India and Southeast Asia are also important growth markets, with the availability and reliability of electricity likely to determine how quickly announced projects become operating facilities.

The Gulf deserves particular attention. The region combines capital, established energy industries, strategic location between Europe and Asia, and ambitions to become a centre for digital infrastructure. Yet an energy-exporting country does not automatically have unlimited electricity available at the right location. Data-centre development still requires generation, grids, cooling, water planning and dependable fuel logistics.

Power is becoming a location strategy

For years, the location of a data centre was often discussed in terms of connectivity, customers, tax policy and real estate. Those factors remain relevant, but power is moving to the centre of the decision. A developer may prefer a particular city, but if the grid cannot provide a large and reliable connection for several years, the project may move elsewhere or require dedicated generation.

This is why the next phase of AI infrastructure will be closely linked to energy security. The IEA has highlighted grid connection delays and supply-chain constraints for transformers and other electrical equipment. A power plant alone is not enough; electricity must be transmitted, transformed and delivered to the site. Permitting, skilled labour and equipment lead times are now strategic considerations.

In practical terms, I expect successful AI locations to be those able to assemble a complete energy system: diversified generation, robust grid connections, storage, backup arrangements, and long-term contracts that make the cost of power predictable. No single technology will provide the answer everywhere.

Why a diversified energy mix matters

Renewables will be central to the expansion because wind and solar can be competitive and can be deployed at scale. But their output varies with weather and time of day. Batteries and other forms of storage can help, while stronger grids allow electricity to move between regions. For facilities that need continuous power, the system must also have dependable sources available when renewable generation is low.

Natural gas is likely to remain part of that solution in many markets, particularly where it can provide flexible generation while grids and storage expand. Nuclear power, including potential small modular reactors over the longer term, may offer firm low-carbon electricity where projects can be financed, licensed and delivered. Hydropower and geothermal can also be important in suitable locations.

The right answer is not to choose one source and assume it will solve everything. It is to build a resilient portfolio that balances reliability, affordability, emissions, local resources and the needs of the grid. For AI operators, energy procurement is becoming a core part of corporate strategy, not a back-office utility contract.

The connection to LNG and the tanker business

This brings me to shipping. When we discuss the energy requirements of AI, the maritime dimension is sometimes overlooked. Electricity is produced locally, but the fuels that support generation are often part of international trade. In countries that rely on imported natural gas, LNG carriers connect gas-producing regions with receiving terminals, regasification facilities and power stations.

If gas-fired generation expands to support data-centre demand, it can reinforce the strategic importance of LNG supply chains. That does not mean every new data centre creates a direct or equal increase in LNG imports. The effect depends on the local generation mix, renewable output, nuclear capacity, pipeline gas availability, storage and government policy. A data centre powered by hydroelectricity or nuclear energy has a very different fuel footprint from one relying on gas-fired generation.

Nevertheless, in markets where gas is the flexible source available to meet new demand, the implications for LNG procurement, shipping and terminal capacity can be significant. Buyers may seek more diversified suppliers, longer-term contracts and greater flexibility in destination and delivery schedules. Importing countries may consider floating storage and regasification units (FSRUs) or new terminal capacity where onshore infrastructure is delayed or unavailable.

For the tanker business, it is important to be precise. LNG carriers are a specialised fleet and should not be confused with crude-oil or product tankers. Their demand is driven by the volume and distance of LNG trade, liquefaction capacity, long-term contracts, fleet availability and the development of import infrastructure. The growth of AI-related power demand may become one additional source of structural gas demand in certain markets, but it is not, by itself, a sufficient basis for ordering new ships.

Crude and product tankers may be affected more indirectly. The broader energy transition, changes in power generation, refinery demand and the movement of feedstocks all influence tanker trades. Data-centre growth could also increase demand for backup fuels in some locations, although this is likely to remain a small and highly site-specific part of the overall picture. The stronger and more immediate maritime link is likely to be through LNG and the infrastructure that supports gas-fired power.

Routes, chokepoints and energy security

Energy security is not only a question of how much fuel a country can buy. It is also about where that fuel comes from, which routes it follows, how many alternatives are available and what happens when a disruption occurs. LNG trade depends on liquefaction plants, specialised vessels, receiving terminals and maritime passages. The Strait of Hormuz, the Suez Canal and the Panama Canal can all matter to particular flows, depending on origin and destination.

For governments and operators, resilience means understanding these routes and avoiding excessive dependence on a single supplier, terminal or passage. Diversification has a cost, but so does interruption. In a world where digital infrastructure is increasingly critical to finance, communications, public services and industry, the consequences of an extended power disruption can reach far beyond the data centre itself.

That is why I see energy security and maritime security as connected subjects. A reliable AI ecosystem requires not only secure data and cyber systems, but also secure energy infrastructure, dependable logistics and the ability to respond to geopolitical or operational shocks.

What should investors and shipowners watch?

I would watch the conversion of announced data-centre projects into facilities with secured power, rather than relying on headline investment figures alone. The key indicators are signed electricity supply agreements, grid connection dates, generation projects reaching financial close, LNG import commitments, terminal utilisation and the actual commissioning of new computing capacity.

For shipowners, the opportunity should be assessed with the same discipline as any other trade. LNG fleet demand will depend on confirmed liquefaction projects, contracting patterns, vessel orderbooks, charter duration, financing costs and the timing of new terminals. A forecast of AI electricity demand is not a charter contract. The strongest opportunities will be those supported by real infrastructure and durable cargo flows.

There may also be opportunities for integrated energy and infrastructure partnerships: power producers, utilities, LNG suppliers, terminal operators, shipping companies and technology firms working together to secure supply. Such arrangements could make sense where demand is large, predictable and strategically important. But they require careful allocation of construction, price, volume, regulatory and geopolitical risks.

The next infrastructure map

We are entering a period in which the geography of intelligence and the geography of energy will increasingly overlap. Some data centres will locate near customers and fibre networks; others will follow abundant power, cooler climates or available land. Some countries will use renewable electricity and storage; others will rely on gas, nuclear power or a combination. The global map will not be uniform.

What seems clear to me is that AI is not weightless. It depends on physical infrastructure and international supply chains. The race to build it will create new demands on electricity systems and, in some markets, on the ships and terminals that move energy across the world.

For the maritime industry, this is a development worth following closely, but with a clear distinction between opportunity and speculation. The future of tanker business will not be determined by AI alone. It will be shaped by the wider evolution of energy demand, the economics of gas and renewables, geopolitical risk, shipping routes and the investment decisions that turn plans into actual cargoes.

The question I would put to policymakers, energy companies and shipowners is a straightforward one: as we build the infrastructure for an increasingly intelligent world, are we also building the resilient energy and transport systems capable of keeping it running?