"AI disrupts everything" sweeps the stock market, where is a safe haven? Goldman Sachs gives the key word: HALO

Goldman Sachs Group Inc., a financial giant on Wall Street, has released a research report stating that companies with tangible production assets are significantly outperforming the global stock market. This is mainly because global investors, including hedge funds and retail investors, are actively seeking safe havens to dodge the "artificial intelligence disruption storm." As a result, they are shifting their investment focus towards companies with heavy asset-intensive stocks known as HALO (Heavy Assets, Low Obsolescence). The Goldman Sachs research team stated that their compiled "heavy asset-intensive" stock basket, which derives its economic value from physical assets or physical production materials, has outperformed the "light capital-intensive" stock comparison group, which depends more on labor or digital resources/capital, by roughly 35% since early 2025. The "HALO effect" cited in the report by Goldman Sachs refers to companies whose value primarily comes from high replication costs, long lifespans of physical assets/core capabilities/manufacturing networks/infrastructure. As a result, investors consider them less susceptible to rapid AI replacement or "technological obsolescence," making it easier for them to gain a "hedge premium" during the AI anxiety. The typical characteristics include heavy asset barriers to production materials, such as power grids/mines/oil and gas assets/large utility networks, key AI infrastructure manufacturers, where the replication costs of AI are extremely high, and human technological value is difficult to be overcome by AI. Another characteristic is the low technology obsolescence rate, such as core capabilities that are difficult to be completely replaced by software and robotics in the short term, making it challenging for AI to replace human technological value, as seen in the semiconductor equipment supply chain, chip outsourcing, chip advanced packaging and testing, and other technological processes. AI is not just "destroying" the profit structure of some light-asset industries, but also creating a real-world "capital expenditure supercycle" - such as the AI chip and storage chip supercycle. Goldman Sachs estimates that the five "super-sized cloud providers" will cumulatively invest around $1.5 trillion in AI infrastructure from 2023 to 2026, transforming themselves from "capital-light winners" in the traditional sense to "capital-intensive players." This substantial increase in orders will be more directly allocated to chip manufacturing, electricity, energy, materials, equipment, data centers, and cooling/electrical distribution chains. HALO effect Goldman Sachs research team, including strategists like Guillaume Jaisson, stated in this client report that investors are increasingly turning to heavy asset-intensive stocks with the "HALO effect" - those stocks that hold extremely heavy assets and face much lower risks of AI obsolescence. These stocks are mainly concentrated in industries such as utilities, basic resources, chip manufacturing, semiconductor equipment manufacturing, and energy in traditional manufacturing sectors. "The market is rewarding capacity, high-density manufacturing networks, infrastructure, and extremely complex manufacturing project engineering - these assets have extremely high replication costs and require significant investment by AI systems for 'trial and error' processing or production experiments, making them less susceptible to the impact of AI technological obsolescence," wrote the Goldman Sachs strategists. As shown in the chart above, heavy capital-intensive stocks are significantly outperforming the market. When sectors like software, considered to be at risk of "AI disruption," suffered significant declines, real-world assets performed well. The data above is normalized based on percentage increases as of December 31, 2024. The anxiety over AI applications disrupting all traditional business models has spread across various large industries, including SaaS software, wealth advice and management, and real estate consulting, leading to a significant drop in stocks previously seen as "inevitable AI winners." This extreme concern about "AI disrupting everything" has evolved into indiscriminate irrational selling and spread to industries seemingly unaffected by AI risks, such as labor-intensive logistics and transportation. However, some "heavy asset-intensive" companies are not performing well. Goldman Sachs strategists stated that the race for AI leadership has transformed the previously long-term outperformers in the light-asset sector, the so-called five "super-sized cloud computing service providers" (hyperscalers), into trading targets for heavy asset-intensive stocks. Yet, the significant AI capital expenditures closely linked to the AI data center construction wave have caused these companies' stock prices to continue to fall. This is mainly because investors have begun to question whether the massive ongoing investment in AI computing infrastructure (with the top four US tech giants' AI capital spending expected to exceed $700 billion this year, representing a potential increase of 60%) will generate robust enough returns to support their high valuations. The strategists estimated that the five large tech companies - Amazon, Microsoft, Google's parent company Alphabet Inc., Facebook's parent company Meta Platforms Inc., and Oracle - are expected to collectively invest around $1.5 trillion in building vast AI computing infrastructure from 2023 to 2026. In comparison, these companies had invested approximately $600 billion in the entire historical period leading up to 2022. The Goldman Sachs team stated that higher actual returns, as well as geopolitical factors driving increased fiscal spending and manufacturing support, are supporting the shift of funds to capital-intensive market segments. They also pointed out that profit momentum is shifting towards these long-term heavy asset stocks: the significantly higher expected earnings per share (EPS) growth rates and return on equity (ROE) rates, now significantly higher in these capital-intensive companies than in light capital companies. Another financial giant, Morgan Stanley, also stated that the market is moving away from sectors like software that involve light assets. Morgan Stanley strategists wrote in a report on Monday that some long-term European-only funds had already started reducing their exposure to SaaS-type software stocks facing AI disruption risks by the end of 2025.