Artificial intelligence is rapidly evolving from a software and semiconductor story into one of the largest infrastructure investment cycles of the decade.
For much of the current AI boom, investor attention has centered on advanced processors and companies positioned at the heart of accelerated computing. But as artificial intelligence systems scale, the physical infrastructure required to power, cool and connect those systems is becoming just as strategically important.
The next phase of AI investment is increasingly moving into electricity systems, transformers, liquid cooling, servers, networking equipment and hyperscale data-center capacity.
Reuters reported this week that companies supplying power and cooling equipment are emerging as significant beneficiaries of global data-center expansion. McKinsey estimates cited by Reuters suggest that cumulative worldwide investment in data-center infrastructure could approach US$7 trillion by 2030.
That scale reflects a fundamental reality of artificial intelligence: compute cannot expand without infrastructure.
Large AI models require enormous clusters of high-performance processors operating continuously. Those processors consume substantial amounts of electricity and generate significant heat, creating demand for increasingly sophisticated power-distribution and cooling systems.
This is beginning to reshape capital allocation across industries that previously sat outside the traditional technology sector.
South Korea's HD Hyundai Electric, China's Hainan Jinpan Smart Technology and Taiwan's Delta Electronics are among the companies benefiting from increased demand for power and cooling equipment associated with data-center construction, Reuters reported. New technologies including solid-state transformers and liquid-cooling systems are becoming increasingly important as operators seek higher efficiency and greater computing density.
Corporate transactions are reinforcing the trend.
Energy-services company SLB announced a US$4.1 billion acquisition of cooling-equipment manufacturer Kelvion, including debt, as it expands into infrastructure supporting AI-driven data-center growth. The transaction demonstrates how the artificial-intelligence investment theme is extending beyond conventional technology companies into industrial and energy infrastructure.
Demand at the computing layer also remains exceptionally strong.
Dell Technologies raised its annual outlook after reporting continued strength in AI-optimized server demand. Reuters reported on September 2 that Dell shares rose sharply after the company increased its revenue and profit forecasts as demand for servers designed for artificial-intelligence workloads continued to accelerate.
The infrastructure opportunity is also increasingly global.
Singapore's Keppel DC REIT and Keppel recently agreed to acquire a 90% effective interest in two Tokyo data centers for approximately 190 billion yen, or US$1.19 billion, highlighting continuing institutional demand for digital infrastructure assets in Asia.
Meanwhile, India's Yotta Data Services is targeting an IPO in early 2027 that could raise up to US$1.5 billion, with proceeds intended partly for GPU purchases and expansion of sovereign cloud infrastructure. The company's growth reflects increasing demand for AI computing capacity outside the United States and Europe.
NEXUS Analysis
The strategic implication is that the AI investment opportunity is becoming broader than the semiconductor cycle.
Artificial intelligence requires an interconnected infrastructure stack:
Compute. Advanced processors and AI-optimized servers remain the foundation of model training and inference.
Power. Data centers require increasingly large and reliable electricity supplies, creating opportunities across generation, transmission, transformers and grid equipment.
Cooling. Higher-density computing makes thermal management critical, accelerating demand for liquid cooling and other specialized systems.
Connectivity. AI clusters depend on high-speed networking and optical infrastructure capable of moving enormous volumes of data.
Physical Infrastructure. Land, buildings, substations and data-center campuses are becoming strategic assets within the digital economy.
From an institutional investment perspective, this changes how the AI theme should be analyzed.
The first phase of the cycle rewarded companies selling the most advanced computing hardware. The next phase may increasingly reward businesses that solve the physical bottlenecks created by AI itself.
Power availability, cooling capacity, grid access and data-center development timelines may become as important to artificial-intelligence expansion as processor performance.
There are also risks.
Rapid investment can create overcapacity, speculative development and infrastructure requests that exceed realistic demand. Reuters reported that electricity requests from U.S. data-center projects have become so large that regulators and utilities are increasingly scrutinizing whether all proposed developments are financially and technically viable.
That means investors will need to distinguish between genuine infrastructure demand and projects built primarily around expectations of future AI growth.
Strategic Outlook
The artificial-intelligence economy is entering a phase in which digital intelligence and physical infrastructure are becoming inseparable.
The companies that manufacture chips remain critical, but the investment universe is expanding toward the businesses that provide the electricity, cooling, networking, servers and physical capacity required to keep those chips operating.
For institutional capital, the opportunity may therefore be larger than AI software or semiconductor exposure alone.
The next infrastructure trade could be built around the systems that make artificial intelligence physically possible.
NEXUS PROJECT — Capital. Technology. Intelligence.