By Seth Kirkham
About the author: Seth Kirkham is chief investment officer of Galvanize Global Equities.
Investors are continuing to flee from software and pile into so-called HALO stocks -- Heavy Assets, Low Obsolescence. ServiceNow and Salesforce stocks sank 6% and 12%, respectively, on Tuesday. Both software firms have already fallen by more than 30% this year. Bloom Energy and Vertiv stocks, by contrast, are up roughly 67%.
The move out of software reflects a growing belief from investors that in the era of AI disruption, physical assets offer refuge. Power producers, grid operators, and heavy industrial firms cannot be replaced by code.
That sounds like a logical trade, but it isn't necessarily a profitable one. Infrastructure cycles like the current AI buildout have a habit of overshooting. When the narrative is compelling, capital sometimes becomes too abundant and returns compress.
Data center infrastructure will definitely become more capital intensive as AI expands, and companies that can deliver power and engineering firms that can execute stand to benefit. But benefiting from capital intensity and earning attractive returns on invested capital aren't identical outcomes. Not every power producer will become a growth stock. Not every grid operator will earn excess returns.
Infrastructure sectors have been rerated on the back of technological- demand stories before. Think fiber in the late 1990s, commodities in the 2000s, or shale in the early 2010s. In each case, the underlying thesis contained nuggets of truth. Data traffic surged, commodity demand rose, and U.S. oil demand grew. But capital arrived faster than that demand, timelines were misjudged, and balance sheets had to absorb the consequences. The direction was right, but the returns were uneven.
The HALO trade isn't insulated from repeating this capital cycle. It may be that the HALO infrastructure companies are attracting capital faster than they can absorb, and today's valuations aren't fully reflecting execution risk.
Whether or not a HALO trade is actually an attractive investment often lies in specifics: Are utilities signing long-term load agreements? Are regulators allowing timely cost recovery? Are data center projects backed by firm commitments or projections? Is capital being deployed against measured demand -- or anticipated demand? These questions haven't been dominating headlines lately, but they do determine investment outcomes.
Investors should also keep in mind that the seemingly endless flow of new demand for AI doesn't eliminate supply constraints. AI may scale in abundance, but the systems required to power it scale up slowly and unevenly. High-voltage interconnection capacity, skilled electrical labor, grid permitting, power electronics integration, and cooling density are finite inputs. When demand accelerates and supply remains constrained, pricing power will concentrate in the bottlenecks.
Investors who identify those choke points and allocate capital where constraint determines return will benefit significantly. The electrification of intelligence may reshape the economy, but the balance sheet will still determine who earns the return.
Guest commentaries like this one are written by authors outside the Barron's newsroom. They reflect the perspective and opinions of the authors. Submit feedback and commentary pitches to ideas@barrons.com .
This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
March 24, 2026 16:06 ET (20:06 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments