With prices going down on certain Date Centre’s what is the future of data centres: a major growth industry, but with risks
The AI view is that data centres are likely to be one of the most strategically important infrastructure industries over the next 10–15 years, driven by artificial intelligence (AI), cloud computing, digital services and the enormous processing requirements of modern businesses.
However, there is an important distinction between the future of data centres as an industry and the investment prospects of individual companies building and operating them. Demand could rise substantially without every data-centre investment proving profitable.
What will drive growth?
1. Artificial intelligence
AI models require enormous computing power, particularly when training models and responding to millions of user requests. Advanced AI servers use much more electricity and generate much more heat than traditional servers.
Outlook: The strongest potential growth driver, although demand will depend on how profitably AI services can be commercialised.
2. Cloud computing
Businesses continue to move software, storage, databases and computing away from their own premises and into specialist facilities operated by cloud providers.
Outlook: A durable source of demand, even if AI investment fluctuates.
3. Electricity and power infrastructure
AI data centres require substantial, reliable electricity supplies, alongside backup power, cooling and grid connections. In some markets, access to power is becoming a greater constraint than access to land.
Outlook: Companies supplying power equipment, cooling systems and electrical infrastructure could benefit as much as data-centre owners.
4. Advanced cooling and networking
High-performance AI chips produce intense heat and need increasingly sophisticated cooling. Data centres also need high-speed networking to connect thousands of processors.
Outlook: Specialist engineering and equipment suppliers may have attractive opportunities because these systems are essential to operating the facility.
2. How large could the industry become?
The International Energy Agency’s 2026 analysis forecasts that global data-centre electricity consumption will rise from approximately 485 terawatt-hours (TWh) in 2025 to around 950 TWh in 2030 — almost a doubling in five years. AI-focused data-centre consumption is projected to grow even faster.
Projected global electricity consumption by data centres
2025
485 TWh
2030 forecast
950 TWh
Source: IEA, 2026. The figures are forecasts, not guaranteed outcomes.
That is an extraordinary increase in demand for a single category of infrastructure. It creates opportunities across property, power generation, electrical engineering, computing equipment and specialist services.
But electricity consumption is not the same thing as revenue or profit. The critical question for investors is whether customers will pay enough for the computing capacity to justify the cost of constructing and operating these facilities.
3. The UK opportunity
The UK has an established financial-services industry, strong internet connectivity, major cloud customers and demand for AI services. London is the dominant UK data-centre market.
According to CBRE’s 2026 outlook, London is expected to see substantial new capacity, with expansion increasingly moving into surrounding areas where power is more readily available.
The main obstacle is electricity infrastructure.
A particularly revealing recent development is Microsoft’s withdrawal from a proposed 50-megawatt AI data-centre project in Loughton, Essex, after its anticipated power connection was delayed until the 2030s. Microsoft said this was a site-specific decision, not a withdrawal from its wider UK partnership.
This illustrates an important investment principle: having land, planning permission and willing customers is not enough if the electricity cannot be delivered.
In the UK, I would pay particular attention to:
– Sites with secured, credible grid connections.
– Locations with sufficient power capacity and planning approval.
– Regions with good fibre connectivity and access to skilled labour.
– Projects with long-term customer contracts rather than speculative demand.
– Developers that can bring facilities into operation on time and within budget.
4. Where are the best investment opportunities?
I would divide the sector into five categories.
A. AI chips and computing equipment
Companies such as Nvidia supply the processors used in advanced AI systems.
Potential: high growth, but considerable valuation risk and competition as customers seek cheaper computing.
B. Data-centre property and operators
Examples include Equinix and Digital Realty.
Potential: recurring rental and service revenue. Risks include high capital costs, borrowing costs, competition and the need for constant reinvestment.
C. Electrical equipment and infrastructure
Examples include Schneider Electric and Eaton.
Potential: beneficiaries of power distribution, backup systems and capacity expansion, without having to own the data centre itself.
D. Cooling and thermal management
Advanced cooling is increasingly important as server power density rises.
Potential: specialist equipment and engineering suppliers could benefit from the shift to high-density AI facilities.
E. Power generation and energy storage
Electricity generators, grid equipment companies, battery storage and some nuclear-energy suppliers may benefit from additional demand.
Potential: long-term demand growth, but returns depend on regulation, power prices, project costs and how new capacity is financed.
My initial preference would be to investigate the suppliers of essential equipment and power infrastructure alongside established data-centre operators. That offers exposure to the expansion without relying exclusively on the success of individual AI companies.
It does not automatically make these investments cheap: some suppliers may already have very optimistic growth expectations reflected in their share prices.
5. What could go wrong?
There are five significant risks that I would take seriously.
Risk Why it matters
AI investment bubble Technology companies may build more capacity than customers ultimately need or can afford.
Excessive construction A rush to build facilities could eventually produce oversupply and lower rental prices.
Electricity constraints Projects may be delayed for years or require expensive alternative power supplies.
Technology advances More efficient chips and AI models could reduce the computing resources needed for some tasks, although lower costs may also stimulate more usage.
High interest rates Data centres require substantial upfront investment, making financing costs particularly important.
The IEA itself highlights uncertainty over AI adoption, profitability, electricity demand and the timing of infrastructure investment.
I would also distinguish between AI demand and AI profitability. Even if AI becomes indispensable, that does not mean every company providing AI infrastructure will earn an attractive return on capital.
6. My outlook over the next 10 years
2026–2028: Infrastructure expansion
Strong investment in AI computing, power connections, cooling and networking. Delivery delays and equipment shortages will separate successful projects from speculative ones.
2028–2031: Winners and losers emerge
Investors will increasingly focus on utilisation rates, contracted revenues, operating costs and actual returns on capital. Some developers may struggle if they have overpaid for sites or cannot secure customers.
2031–2036: A more mature digital infrastructure industry
AI, robotics, autonomous systems, scientific computing and enterprise software could sustain demand. Power efficiency, specialised chips and new facility designs will become increasingly important.
These are scenarios rather than firm predictions; the speed of AI adoption and the economics of computing could materially alter the timetable.
My conclusion
I believe data centres have a strong long-term growth outlook. However, I would not simply buy shares in any company associated with AI and assume the investment will succeed.
For an investor, I would investigate three things first:
1. Who has pricing power? Companies supplying essential equipment or scarce capacity may be better placed than businesses competing solely on price.
2. Who has the strongest balance sheet? Heavy borrowing can turn a promising infrastructure business into a poor investment.
3. What price are you paying? An excellent company can still be a bad investment if its shares already assume years of exceptional growth.
Given your interest in longer-term investments and fixed-income opportunities, there is another angle worth examining: the debt issued by data-centre operators and infrastructure companies. That offers a different balance of income, credit risk and exposure to the sector than buying their shares.
AI report driven.. KMI