Morgan Stanley: Institutions Overweight AI Hardware Chain in Q3, Software Stocks Systematically Underweighted
nashnova research
Morgan Stanley's analysis of Q3 13F filings shows active managers accelerating their bets on upstream AI hardware — semiconductor equipment, storage, memory — while systematically underweighting traditional software leaders. The money is chasing pick-and-shovel sellers, not the coders.
The Mega-Seven — how much did institutions actually own?
At quarter-end the seven mega-cap tech names — Nvidia, Apple, Microsoft, Amazon, Alphabet, Meta, Tesla — sat at an aggregate -124 basis points underweight, narrowing 5 bps from Q2's -129 bps. This means → institutions are still underweight, but the selling pressure is easing at the margin — a shift from "hard no" to "a little less no."
In plain terms = if the S&P 500 index weight is the "standard position," active managers held 1.24 percentage points less of these seven names than the benchmark demanded. The money did not follow market cap.
The remaining 21 large-cap tech stocks carried a +28 bps overweight, down from +39 bps in Q2. This reflects a clear intra-sector rotation, not a retreat from tech overall.
Nvidia and Tesla — who sits at each extreme?
Nvidia remains the single most underweighted stock among institutions, reaching -2.75% at Q3-end — widening 22 bps from Q2 and hitting a record underweight. This means → at the peak of the AI narrative, active funds were actually cutting their largest AI-chip position.
Morgan Stanley flags an important caveat: 13F data captures a static snapshot as of September 30. In the 45 trading days after that date, Nvidia's stock fell 32% while Apple rose 6% — Apple has likely overtaken Nvidia as the most deeply underweighted mega-cap tech name by now.
Tesla completed a landmark position reversal: flipping from underweight to an overweight of +0.71%, a 101 bps swing quarter-over-quarter — the only mega-cap tech name actively overweighted by institutional managers. In plain terms = the entire active-management community pivoted from bearish to bullish on Tesla in a single quarter.
The AI hardware chain — where exactly did the money go?
Top three overweights: semiconductor-equipment maker KLA at +1.29%, storage maker Seagate at +1.01%, and flash-memory firm Western Digital at +0.86%. Lam Research and Dell also carried positive position deviations.
This reflects a very clear stock-picking logic: follow the AI supply chain upstream to the bottlenecks — whoever sells the picks and shovels, whoever controls the chokepoint, gets the capital.
In plain terms = AI training devours chips, chips need equipment to manufacture and storage to feed data — institutions are betting on the "equipment → storage → compute" chain where revenue certainty is highest.
Why are software stocks being left behind?
IBM, Palo Alto Networks, Adobe, Oracle, and ServiceNow all carried active positions well below their S&P 500 index weights — a sector-wide systematic underweight.
This means → the market does not doubt AI itself, but it lacks confidence in when AI will show up in software companies' actual revenue. Hardware revenue has already landed; software-side AI monetization remains at the expectations stage.
Morgan Stanley's view: whether software stocks see a turnaround in Q4 positioning depends on tangible progress in AI application-layer commercialization — not the story, but the real revenue numbers.
市场有风险,内容仅供研究参考,不构成投资建议。
