Investors rotate to 'steel' assets
Market commentary describes a $400 billion rebalancing away from crowded AI‑infrastructure names toward harder industrial assets like factories, power and logistics. (markets.financialcontent.com) That narrative frames a question about which firms capture value from deploying AI at scale versus those that simply house the narrative. (markets.financialcontent.com)
Investors are shifting money from the companies selling the artificial-intelligence story to the companies building the power, factories and freight behind it. (markets.financialcontent.com) A MarketMinute commentary published April 13 said that move erased about $400 billion in market value from crowded artificial-intelligence infrastructure names as the second quarter began. The piece described the trade as a rotation into energy, industrials and other “real economy” assets. (markets.financialcontent.com) The argument is not that chip demand disappeared. Nvidia reported $68.1 billion in quarterly revenue on February 25, including $62.3 billion from data centers, with both figures up more than 70% from a year earlier. (nvidianews.nvidia.com) The question is where the next dollar of profit lands. If artificial intelligence works like a factory line, investors are asking whether the richest returns go to the company making the machines or to the owners of the electricity, buildings and transport that keep the line running. (blackrock.com) That debate has sharpened because the physical bill is getting easier to measure. JLL said on January 6 that global data-center capacity could nearly double from 103 gigawatts to 200 gigawatts by 2030, and that about $3 trillion of investment would be needed to add 100 gigawatts of new supply. (jll.com) JLL also said artificial intelligence could account for half of all data-center capacity by 2030, while construction costs have been rising about 7% a year. That pushes attention toward utilities, equipment makers and landlords that can secure land, transformers and power faster than rivals. (jll.com) Power is the tightest choke point in many of these projects. Deloitte’s 2026 power and utilities outlook said rapid growth in artificial intelligence, electrification and data centers is straining grid capacity and forcing utilities to seek new financing and operating models. (deloitte.com) BlackRock Chief Executive Larry Fink made a similar case in his 2026 annual letter, telling investors that more economic value is being created in capital markets and arguing for long-term investment in the systems that support growth. His firm also used its Global Infrastructure Partners unit to convene an infrastructure summit in Washington on March 11 focused on bottlenecks. (blackrock.com; semafor.com) Not everyone reads the trade as a verdict against technology stocks. Nvidia told investors on February 25 that it expects sequential revenue growth through calendar 2026, and its earnings call pointed to a data-center business that has grown nearly 13-fold since fiscal 2023. (nvidianews.nvidia.com; q4cdn.com) What changed is the market’s tolerance for paying any price for that growth. As artificial-intelligence spending moves from buying chips to wiring substations, pouring concrete and signing long-term power contracts, investors are treating “steel” assets less like side characters and more like the toll roads of the boom. (markets.financialcontent.com; jll.com)
