For ETF investors, the key signal here is not a simple โAI trade overโ message, but a possible rotation within the theme. When hyperscalers become crowded and expensive, capital can move toward the suppliers that sit higher up the AI spending chain: memory, storage, and other infrastructure names. That matters because many technology ETFs are built around market-cap weighting, which can leave them heavily exposed to the same few winners and the same valuation risk.
This is where index composition matters. A broad tech ETF may still own both the big cloud platforms and the semiconductor beneficiaries, but the portfolio outcome can differ sharply depending on whether the index is concentrated in mega-cap software and platforms or more evenly spread across chipmakers, equipment, and hardware. Investors using sector ETFs as a tactical tool should check whether the fund is actually capturing the โpicks and shovelsโ layer, or simply repackaging the most familiar AI beneficiaries.
The second issue is financing discipline. The article highlights that AI capex is increasingly being funded not only from operating cash flow but, in some cases, from debt issuance. For long-term holders, that introduces a useful distinction between businesses that are spending to build the AI ecosystem and businesses that are selling the essential components. ETF investors should think about how much of their exposure depends on the marketโs willingness to keep rewarding spending rather than monetisation.
Finally, this kind of rotation can test behaviour. Sector ETFs often look diversified, but performance can still be driven by a narrow set of names. Rebalancing rules, concentration limits, and the fundโs definition of the sector all matter when sentiment shifts from AI enthusiasm to AI fatigue.
The Earnings Test: Will Meta, Amazon, Alphabet & Microsoft Pass?ย
Earnings, which begin on July 22, will put hyperscalers to the valuation test. These companies have already scaled their projected capex to a massive 92% of their projected operating cash flow during the last earnings season. Contrary to expectations, the projection proved to be a bullish catalyst for almost each of them. This is because investors perceiveย Alphabet, Microsoft and Amazon asย direct beneficiaries of AI capex spending via their cloud computing divisions. A lens which, unfortunately, bypassed Mark Zuckerberg’s Meta Platforms, according to Jefferies. Nonetheless, investors are growing wary of whether any of these companies will ever successfully monetise their AI capex. Jefferies flags a real possibility of none of them being able to do it.Big Borrowing Bane?ย
In addition to spending their cashflow, all these four companies, (with the exception of Microsoft Corp) are also increasingly borrowing money. The four major US hyperscalers have issued $169 billion worth of bonds so far this year, including $25 billion of Amazon bonds issued recently, compared with $83 billion in 2025, according to Bloomberg. “Alphabet, Amazon and Meta issued US$52bn, US$92bn and US$25bn year-to-date, while Microsoft has not yet issued any bonds. In addition, Oracle has also issued US$25bn of bonds year-to-date,” Jefferies highlighted.What Does The Future Hold?ย
Jefferies estimates AI capex by hyperscalers will total about $700 billion this year, exceed $800 billion next year, and surpass $1 trillion in 2027 when Oracle, OpenAI, Anthropic and neo-cloud providers are included. “The hyperscalers have underperformed the S&P 500 by 11% since early May and are down 8.7% from the peak reached in late May on a market capweighted basis,” Jefferies underlined. Going ahead the hyperscalers’ share price performance, both in absolute and relative terms, will remain crucial to monitor as massive selloff risks by traders grow on concerns surrounding lack of AI capex monetisation. Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories โ On NDTV Profit.Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.