The wave of AI investment is entering a new phase. Since 2026, the allure of the "Magnificent Seven" tech giants, which long dominated US stock market trends, has been fading. Market capital is shifting from large technology platforms towards the AI infrastructure supply chain, particularly the memory and storage chip sector. As demand for AI computing power continues to expand, investors are reassessing who the most direct beneficiaries of this AI cycle are.
Year-to-date, the Nasdaq 100 index has gained nearly 18%, while the S&P 500 is up about 10%. However, an index tracking the "Magnificent Seven" has risen only 1.1%. In stark contrast, the Philadelphia Semiconductor Index has surged 82%, on track for its best annual performance since 1999. Within this rally, memory and storage chip companies like Micron Technology and SanDisk Corp. have become the market's new darlings, while the "Magnificent Seven" that previously led the AI rally are receding into the background.
Fund flows are confirming this shift. According to data compiled by Bloomberg, investors pulled $786 million from the Roundhill Magnificent Seven ETF in June, the largest monthly outflow since the ETF's inception. Simultaneously, the Roundhill Memory ETF attracted $930 million in inflows.
Decoupling of the "Magnificent Seven" and the Broader Market Signals a Shift in AI Trading Logic
For the past few years, the "Magnificent Seven" were virtually synonymous with the AI theme in US stocks. However, this tight correlation has been loosening this year. In April, the 40-day correlation coefficient between the "Magnificent Seven" and the Nasdaq 100 index exceeded 0.95, nearing perfect synchronization. Recently, this metric has fallen below 0.7, reaching its lowest level since 2017.
Jessica Rabe, co-founder of DataTrek Research, noted in a client report dated June 30th that the correlation between mega-cap tech stocks and the S&P 500 has reverted to 2015 levels. At that time, these companies' index weightings were only about 10% to 11%.
Nevertheless, the market influence of the "Magnificent Seven" remains significant. These companies still account for approximately 37% of the Nasdaq 100 index's weight and contribute nearly one-third of the S&P 500's market capitalization. Brian Barbetta, co-head of the technology team at Wellington Management, stated that over the past few years, the "Magnificent Seven" were among the few companies consistently delivering earnings growth that exceeded expectations. However, investors are now paying more attention to the risks and challenges they face.
Rising AI Capital Expenditure Pressures Challenge the Valuation Rationale for Giants
The core of this changing market sentiment lies in investors beginning to question whether the massive AI investments by large tech companies can generate commensurate returns. Companies like Microsoft, Amazon.com, Alphabet, and Meta Platforms, Inc. are all accelerating the construction of AI infrastructure. This massive capital expenditure is compressing free cash flow, while uncertainty remains about when these investments will translate into revenue and profit.
Microsoft shares have fallen about 20% year-to-date, with June recording one of its worst monthly performances since 2000. The market is concerned that the company needs to continue investing in AI infrastructure while simultaneously facing potential disruption to its traditional software business from AI technology.
Similarly, Meta Platforms, Inc. is under comparable pressure. Reports indicate that CEO Mark Zuckerberg has acknowledged that the company's development progress on AI agents has not met expectations. Meta is even reportedly considering building a cloud infrastructure business to absorb potentially excess computing resources.
Brian Barbetta pointed out that the market is currently revisiting a key question: whether cloud computing giants will face declining returns on capital and pressure on profit margins in the future.
AI Value Chain Repricing: Memory Chip Stocks Emerge as Clear Winners
In contrast, memory chip companies within the AI infrastructure chain are experiencing significant upward revisions to their profit expectations.
Bloomberg Intelligence data shows that the expected net profit growth rate for the "Magnificent Seven" next year is currently 18.9%, down from 21.4% three months ago. Conversely, future earnings growth expectations for chip manufacturers have surged from 34.3% to 48.5% over the same period. The market narrative is shifting from "AI application potential" to "AI infrastructure realization."
Mark Lehmann, Vice Chairman of Citizens Bank, noted that investors previously had high certainty about the growth prospects of the "Magnificent Seven," justifying higher valuations. Now, the market is re-evaluating the quality of their earnings, while profit expectations for companies like Micron Technology and SanDisk Corp. continue to improve rapidly.
Wall Street Divided on Whether the AI Leadership Has Shifted
However, not all investors believe chip stocks will continue to lead the market.
Nikolaos Panigirtzoglou, a global market strategist at J.P. Morgan, stated in a July 1st report that as cloud computing giants, AI model companies, and AI application firms improve their commercialization capabilities, the performance gap between large tech companies and chipmakers may narrow. He suggests AI value may ultimately reconcentrate in platform companies.
Mike Wilson, Chief US Equity Strategist at Morgan Stanley, believes the upward momentum for chip stocks is waning, and the market is beginning to look for previously lagging investment opportunities, including AI-focused cloud computing giants. "This divergence can't persist," Wilson said. He expects hyperscale cloud companies to stabilize, while semiconductor stocks could face pullback pressure.
In the short term, however, the market's focus remains on the ability to deliver profits. As long as the return on AI investments by large tech companies remains unproven and profit expectations for memory chip firms continue to rise, the dominant force in AI-related trading may well remain on the hardware infrastructure side.
Comments