wsj.com web signal

Retail Traders Rotate From Mag Seven Into SK Hynix, Marvell

nvidia apple funding ai-business

TL;DR

  • Retail traders accounted for just 6% of Magnificent Seven trading volume over the five days through Friday, a four-year low, per Citi.
  • The Bloomberg Magnificent Seven gauge is down 3.1% year to date through Monday's close, while the S&P 500 is up 8.7%.
  • Micron has surged more than 200% year to date, and Marvell Technology, Intel and Lam Research have all more than doubled.

Retail flows into the Magnificent Seven have collapsed to a four-year low, and the money is chasing the picks and shovels instead. According to a Citi team led by Stuart Kaiser, retail traders accounted for just 6% of total trading in Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla over the five days through Friday. In 2023 and 2024 that share was routinely above 20%; in 2025 it mostly held above 15%. Nvidia's retail share slipped to 8.1% from 9.6% the prior week. Tesla, the perennial retail favorite, still led the group at 10%, but that reading is near its lowest since 2022.

The price tape rhymes with the flows. The Bloomberg Magnificent Seven gauge is down 3.1% year to date through Monday's close while the S&P 500 is up 8.7%, per the same reporting. The rotation the Wall Street Journal and other outlets are tracking points to a very different scoreboard among AI-adjacent chip names: Micron is up more than 200% this year, and Lam Research, Marvell Technology and Intel have all more than doubled. The seven best-performing S&P 500 stocks year to date, as of July 9, are Sandisk, Dell Technologies, Micron, Western Digital, Seagate, Intel and Marvell, a memory-and-hardware roll call rather than a hyperscaler one.

SK Hynix's U.S. debut on the Nasdaq on July 10 sits at the center of this shift. The South Korean memory maker sold almost $27 billion of stock in its market-debut Friday, with orders exceeding available shares by more than sevenfold, and its U.S.-listed options started trading the following Tuesday. CNBC's framing, of everyday investors scouring for companies supplying the computational revolution with power and memory, captures why the flows are moving from platforms toward bottlenecks.

The honest caveat is that a share-of-volume snapshot is not a directional bet. Retail could be selling Mag Seven names and buying them back through ETFs, and Citi's window is only five trading days. What the reporting does not give you is a clean breakdown of net dollar flows into SK Hynix or Marvell versus what actually left Nvidia, nor whether institutions are following retail into the memory names or quietly fading them.

If this rotation sticks, the winners are the AI supply chain, memory, custom silicon, networking and power, and the exposed name is any hyperscaler whose story now depends on justifying capex against a market that has decided to price the shovels directly.