For Asia-Pacific stock markets, is REAL trading more suitable than HALO?

How do the four defensive moats of the AI·REAL framework resist AI disruption?

The impact of AI is reshaping valuation logic in global stock markets. Given the structural particularities of Asia-Pacific markets, Bank of America Securities believes that the crowded “HALO” (heavy assets, low淘汰率) trading strategy has clear limitations. Identifying companies using the REAL (Regulatory barriers, Enduring cycles, Asset-heavy, Local services) framework may be more appropriate—especially in markets with abundant capacity and lacking scarcity-driven moats in heavy industries, where the protective effect of the HALO strategy may be overestimated.

According to reports from Wind Trading Desk, Bank of America analysts Winnie Wu and her team stated in a report on the 16th that over the past five months, the market capitalization of the US software sector has evaporated by more than $2 trillion. The Indian IT sector has fallen over 40% from its December 2024 high, and media, e-commerce, and fintech stocks in Asia-Pacific have also experienced large-scale sell-offs. Meanwhile, capital has rapidly rotated into heavy asset sectors such as semiconductors, capital goods, energy, and utilities, boosting the popularity of “HALO” trades.

However, Bank of America warns that the HALO strategy has limitations in markets with ample industrial capacity. In heavy industries lacking regulatory or scarcity-driven moats—such as automotive, solar, steel, and cement—supply can easily outpace demand, leading to fierce price competition. Heavy assets in these sectors not only fail to provide protection but may become burdens—especially as AI shortens R&D cycles and opens up alternative technological paths, exacerbating overcapacity and competition.

Therefore, the REAL framework is introduced to reassess corporate survival risks. It evaluates companies across four dimensions: regulatory barriers, enduring cycles, asset-heavy nature, and local service intensity. Bank of America’s research shows that even in the most AI-impacted sectors like software and consumer internet, leading companies with REAL characteristics still demonstrate strong resilience and long-term investment value.

Limitations of the HALO strategy: Heavy assets are not a universal moat

The core logic behind market chasing HALO sectors is that tangible assets take a long time to build and are difficult for AI to quickly replace. This makes sense in fields where asset scarcity is assured, but in markets with abundant engineering talent and industrial capacity, the limitations are obvious.

Bank of America points out that heavy industries lacking regulatory constraints or scarcity moats—such as automotive, solar, steel, and cement—are prone to rapid supply surpassing demand, leading to intense price competition. In these sectors, heavy assets not only fail to offer protection but may become liabilities—especially as AI further shortens R&D cycles and opens up alternative technological routes. Conversely, some light-asset industries that rely heavily on manual services (like healthcare and dining) may have more durable resistance to AI substitution.

Bank of America also emphasizes that the REAL framework does not imply that companies in these sectors are immune to AI. AI can expand their addressable markets, reduce operational costs, and potentially compress innovation cycles, lowering industry barriers. The key takeaway is: leading companies in high-REAL moat industries should face significantly lower survival threats than their low-REAL counterparts. Especially in environments of rapid disruption driven by AI, valuations of low-REAL sectors may be compressed more persistently and deeply.

The four components of the REAL moat: redefining defensiveness

Bank of America defines the four dimensions of the REAL framework as follows:

Regulatory Barriers (Regulatory Critical): Systemically important banks, telecom operators, power and energy suppliers, and defense-related industries. These sectors involve social stability and national security, with strict regulations on licensing, foreign ownership limits, and critical infrastructure operations. The introduction of AI often entails higher monitoring requirements and increased oversight burdens, raising compliance costs rather than lowering barriers.

Enduring Cycles (Enduring Cycles): Semiconductors, capital goods, aerospace and shipbuilding, biopharmaceuticals, and gaming IP. The moats here stem from uncompressible real-world time accumulation—advanced chip processes depend on multiple generations of process knowledge and EUV equipment; new fabs take years to build; aircraft require complex certification; drugs undergo multi-stage clinical trials and regulatory review; gaming copyrights last 50-70 years, supporting ongoing IP monetization. AI can optimize some processes but cannot bypass the long certification and validation periods.

Asset Heavy (Asset Heavy): Natural resources and commodities, power grids and utilities, railways and ports, livestock farming. The scarcity in these sectors comes from physical resource constraints—limited mineral reserves, lengthy approval processes, high infrastructure costs—making it difficult for new entrants to economically justify replicating existing assets.

Local Services (Local Services): Hotels, restaurants, property management, childcare, eldercare, pet care, and on-site IT deployment and maintenance. These jobs involve non-standardized environments requiring high manual adaptability, with very low tolerance for errors. Currently, AI and robotics are not yet economically capable of effective substitution in these areas.

Distribution of REAL across markets: ASEAN’s short-term advantage and long-term concerns

Looking at the MSCI Asia-Pacific index composition, the AI risk exposure varies significantly across markets. Southeast Asian markets are heavily weighted toward high-REAL sectors: Singapore about 79%, Malaysia about 87%, Indonesia about 94%, with core weights in banking. This structure provides relative resilience in an AI-driven market environment. Since the start of the year, Thailand has gained 14.6%, Malaysia 5.1%, outperforming India (-10.6%).

However, Bank of America warns that the short-term resilience of Southeast Asian markets may turn into long-term vulnerability. If AI-driven automation reduces manufacturing costs in developed economies and nearshoring becomes more attractive, offshore demand for labor-intensive production could decline. Countries like Vietnam, Malaysia, and Thailand have high FDI-to-GDP ratios and export dependence, but without sufficient digital infrastructure and AI talent, they face structural challenges such as widening technological gaps and disrupted global supply chains.

AI disruption combined with aging populations: a dual-axis analysis of sectoral structure

Bank of America analyzes the impact of AI disruption alongside Asia’s low fertility rates and accelerating aging, creating a dual-axis matrix to assess the medium- and long-term structural positioning of sectors.

Healthcare, semiconductors, capital goods, and insurance are in the best position: They have relatively high short-term moats against AI disruption, benefit from aging-driven automation demand, and see increased spending from wealthier elderly consumers. These are “structural opportunities” combining defensive qualities with long-term growth potential.

Real estate, utilities, and banks, while possessing high REAL moats that can buffer AI valuation shocks in the short term, face long-term downward pressure from aging—affecting housing demand and discretionary spending. Bank of America judges these sectors “can provide short-term protection against AI volatility but have limited upside for long-term valuation re-rating.” Consumer durables, media & entertainment, retail, and automotive sectors are under the most pressure—due to low AI moats and the drag of aging on discretionary consumption—making them the most risky in the current landscape.