Goldman Sachs Group Inc. strategists report that companies with substantial tangible production assets are significantly outperforming the global equity market. This trend is driven by global investors, including hedge funds and retail participants, actively seeking safe havens to shield against the "AI disruption" sell-off. These investors are increasingly turning to heavy-asset, low-obsolescence (HALO) stocks, which are perceived as less vulnerable to rapid technological displacement by artificial intelligence.
According to Goldman's research team, a basket of "heavy-asset-intensive" stocks—whose economic value derives from physical assets or production means—has outperformed a control group of "light-capital" stocks, which rely more on human or digital resources, by approximately 35% since the beginning of 2025. The "HALO effect" referenced by Goldman is not the psychological halo effect but denotes companies whose value stems from physical assets, core production capacity, manufacturing networks, or infrastructure that are costly to replicate and have long lifespans. Consequently, these firms are seen as resistant to quick AI substitution or technological obsolescence, often earning a "safe-haven premium" during periods of AI-related anxiety.
Key characteristics of HALO stocks include high barriers from asset-heavy production means, such as power grids, mines, oil and gas assets, large utility networks, and key AI infrastructure manufacturers. Replicating these assets with AI is extremely expensive, and their human technical value is difficult for AI to overcome. The second feature is low obsolescence risk, referring to core production capacities that are unlikely to be fully replaced by software or robots in the short term, even as AI advances. Examples include supply chains for semiconductor equipment, chip foundries, and advanced packaging and testing processes.
While AI is disrupting profit structures in asset-light industries, it is also fueling a real-world "capital expenditure supercycle," such as the boom in AI and memory chips. Goldman estimates that the five hyperscale cloud providers will cumulatively invest about $1.5 trillion in AI infrastructure between 2023 and 2026, transforming them from traditionally capital-light winners into capital-intensive players. This surge in orders will directly benefit heavy-asset manufacturing companies in sectors like power, energy, materials, equipment, data centers, and cooling/power distribution chains.
The Goldman team, including strategist Guillaume Jaisson, noted in a client report that investors are increasingly shifting toward stocks with the "HALO effect"—those with substantial assets and low AI displacement risk. These are predominantly found in traditional manufacturing sectors such as utilities, basic resources, semiconductor equipment manufacturing, and energy. "The market is rewarding production capacity, dense manufacturing networks, infrastructure, and highly complex engineering projects—assets that are prohibitively expensive to replicate and require AI systems to incur significant costs for trial-and-error production testing, making them less susceptible to AI-driven obsolescence," the strategists wrote.
As illustrated, heavy capital-intensive stocks are markedly outperforming the market, particularly when asset-light sectors like software—deemed vulnerable to AI disruption—experience significant declines. Data is standardized to percentage gains as of December 31, 2024.
Anxiety that AI applications will颠覆 traditional business models has swept through major industries, from SaaS software to wealth advisory and real estate consulting, triggering sharp declines in stocks once considered inevitable AI winners. This extreme fear of "AI disruption" has led to indiscriminate, irrational selling, even spreading to sectors not obviously at high AI risk, such as labor-intensive logistics and transportation.
However, not all heavy-asset-intensive companies are performing well. Goldman's strategists pointed out that the race for AI leadership has turned previous long-term outperformers in the light-asset camp—the five hyperscale cloud providers—into heavy-asset-intensive investment targets. Yet, massive AI capital expenditures linked to the data center construction wave have pressured their stock prices, as investors question whether continued large-scale investments in AI computing infrastructure (projected to exceed $700 billion this year for the four largest U.S. tech giants, a potential 60% surge) can generate returns strong enough to justify high valuations.
The strategists estimate that these major tech firms—Amazon, Microsoft, Alphabet Inc. (Google's parent), Meta Platforms Inc. (Facebook's parent), and Oracle—are expected to invest a cumulative $1.5 trillion in building extensive AI computing infrastructure from 2023 to 2026, compared to roughly $600 billion in total historical investments prior to 2022.
Goldman's team also noted that higher real yields, along with geopolitical factors driving increased fiscal spending and manufacturing support, are bolstering the shift of capital into capital-intensive market segments. They highlighted that earnings momentum is turning in favor of these long-standing heavy-asset stocks: consensus estimates for increasingly strong earnings per share (EPS) growth and return on equity (ROE) are now significantly higher for capital-intensive companies than for their light-capital peers.
Another financial giant, Morgan Stanley, confirmed that the market is witnessing a major exodus from light-asset sectors like software. In a Monday report, Morgan Stanley's strategists wrote that some long-only funds in European markets had already begun reducing their exposure to SaaS software stocks facing AI disruption risks by the end of 2025.