The AI boom, which began in late 2022 with the public release of ChatGPT, has resulted in a nearly $33 trillion increase in the S&P 500 Index's market value. This surge in AI-related investments has made it increasingly difficult for Wall Street to achieve true diversification, as many seemingly distinct investments are ultimately tied to the same underlying AI infrastructure spending. Concerns are mounting among AI lab heads about the rapid pace of technological development, prompting a national debate on the balance between AI's promise and its potential for harm.
Traditional investment diversification strategies are proving insufficient in the current AI-dominated market. Many growth ETFs, despite holding hundreds of stocks, show significant concentration in a few large technology companies. For instance, the Schwab U.S. Large-Cap Growth ETF (SCHG) has its top five holdings—Nvidia, Apple, Microsoft, Amazon, and Alphabet—accounting for approximately 35% of the fund. Nvidia alone constitutes 10.83% of the fund. This concentration is driven by the fact that the fortunes of many AI-adjacent companies are directly tied to the capital expenditures of a few hyperscalers.
Microsoft, Amazon, Alphabet, and Meta are projected to spend between $700 billion and $725 billion combined on capital expenditures in 2026, an increase of 60% to 77% over 2025. The overwhelming majority of this spending is dedicated to AI data centers, chips, and networking equipment. Companies like Nvidia, Micron, Vertiv, Equinix, and Arista Networks are all deeply dependent on this sustained spending. This creates a scenario where owning multiple companies across different funds does not provide true diversification but rather stacks exposure to whether these four companies continue their unprecedented spending pace. Jared Gross, head of institutional portfolio strategy at J.P. Morgan Asset Management, highlights that this vast capital investment has overwhelmed traditional portfolio construction models, embedding AI-related risks across almost every asset class, from public equities to private markets.
Despite the significant capital outlay, the hyperscalers remain financially robust, issuing debt and follow-on equity to fund AI development. While their financial flexibility may be somewhat reduced, public hyperscalers continue to be profitable. Investors are encouraged to assess their total portfolio's reliance on these four companies and understand where they have positive AI exposure, unintended exposure, and genuinely defensive, non-AI exposures to balance the inherent uncertainties of the current market landscape. The sheer scale of debt issuance, estimated at $132 billion this year alone for data center rollouts by Google, Amazon, Microsoft, Meta, and Oracle, coupled with rising costs for AI infrastructure, could also pose a risk in fragile bond markets. Bloomberg has also noted that while the cost of building AI infrastructure is not collapsing, the price of AI services is, as exemplified by OpenAI's fee cuts to retain customers.
