UBS SELL

Global equity strategy

Aug 22, 20264 pages

From the report报告摘录Correlation Erosion: AI exposure across compute, power, real estate, and infrastructure creates sector-wide correlations (e.g., semiconductors, copper, REITs), eroding diversification benefits during AI cycle…

Inside the report报告内文 Verbatim from the original PDF — first pages原版 PDF 开篇原文 · 逐字摘录

21 August 2026, 16:56 UTC Chief Investment Office GWM Investment Research

Signal over noise: The illusion of diversification: The hidden AI factor Global equity strategy Authors: Ulrike Hoffmann-Burchardi, Chief Investment Officer Americas and Global Head of Equities, UBS Financial Services Inc. (UBS FS); Nikolaos Fostieris, Global Equity Strategist, UBS Switzerland AG

• AI exposure now reaches well beyond the technology sector, spanning compute, power, real estate, infrastructure, metals, and emerging markets. But this apparent breadth can mask a shared dependence on the same AI investment cycle, reducing diversification when that cycle is reassessed.

• The sensitivity to AI spending is uneven across the value chain. Semiconductors and large-cap technology respond most directly, while copper, power producers, and data-center real estate are influenced by demand for electricity and deployable capacity, alongside their own sector-specific drivers.

• True diversification requires looking beyond static sector and geographic labels. A resilient AI strategy should balance infrastructure exposure with companies monetizing adoption and assess where AI could disrupt existing business models, rather than relying on a single capex-driven source of returns.

AI is broadening, but it is not one trade intensity. Conversely, when semiconductors sell off sharply, The AI investment universe has long expanded from just large-cap technology, IPPs, and copper have generally GPU manufacturers and mega-cap hyperscalers. Directly or weakened alongside them. indirectly, it affects a large number of sectors, including real estate, power utilities, electrical infrastructure, and industrial This does not mean that semiconductors mechanically cause metals. This broader opportunity set creates more ways to moves elsewhere, but it does show that markets can participate in the theme, but at the same time reduces reassess the broader build-out simultaneously, on both the the diversification benefits of previously uncorrelated asset upside and the downside. categories. The common factor is AI capex, but the route differs Asset prices across these sectors increasingly reflect a A useful way to further assess this linkage is to regress daily common risk factor: expectations for the pace, scale, and returns across sectors against semiconductor-stock returns. returns of AI infrastructure investment. This relationship The results support the view that AI investment has become operates in both directions (see Figs. 1 and 2). Stronger a common risk factor, although the relationship varies by semiconductor sentiment can lift the parts of the market asset (see Table 1). most exposed to the physical AI build-out as investors raise expectations for compute demand, hyperscaler capex, power load, data-center construction, and materials

This report has been prepared by UBS Financial Services Inc. (UBS FS), UBS Switzerland AG. Please see important disclaimers and disclosures at the end of the document.

Figure 1 - AI compute demand becomes a materializes, but also greater sensitivity if deployment slows common risk factor or expected returns on new generation decline. Median sector performance on year-to-date trading days when semiconductors declined by more than 2% Data-center REITs are more nuanced. The key constraint for AI is increasingly the ability to deliver power-ready capacity, rather than demand alone. Although they remain correlated with the AI cycle, they have a lower near- term beta that reflects the importance of lease structures, contracted cash flows, tenant concentration, development pipelines, financing conditions, and interest rates.

Emerging markets have historically shown the lowest correlation with semiconductor returns, at 24% over the past two years. But that linkage has strengthened in 2026, Source: Refinitiv, Datastream, UBS, as of 15 Aug 2026 rising to 34%, as the semiconductor rally has become Note: The analysis uses the following indexes: Nasdaq-100, S&P 500 increasingly concentrated in three stocks that together Data Center REITs, S&P 500 Independent Power Producers & Energy account for more than 60% of emerging-market equity Traders, S&P 500…

Read the full report + PDF阅读全文与 PDF

The full summary (4 key points) and the original UBS PDF are for MastermindX Pro members. 完整摘要(4 个要点)与 UBS 原始 PDF 为 MastermindX Pro 会员专享。

Read on MastermindX前往 MastermindX 阅读

Related institutional research相关机构研报

Not investment advice. MastermindX hosts third-party institutional research for reference and education; ratings and views are the authors', not ours. Browse the full Research Vault → 非投资建议。MastermindX 仅收录第三方机构研究,供参考与学习;其中评级与观点均属作者本人。浏览完整研报库 →