5 "Magnificent Seven" Stocks I'm Buying and 2 That I'm Selling

The real signal here is not “AI is good,” but that capex leadership is becoming a return-on-capital contest. NVDA is the cleanest toll collector in the near term because hyperscaler spend still outruns end-demand normalization, while AMZN/MSFT/GOOG/META face a more mixed setup: revenue can improve, but depreciation, power, networking, and financing costs will hit margins before monetization fully catches up. That creates a likely 1-3 month divergence where semis outperform the platforms on guide revisions, even if both themes remain structurally positive.

AAPL and TSLA are the obvious valuation air pockets if market breadth narrows. For AAPL, the non-obvious risk is not demand collapse but gross margin erosion from AI-related component inflation and a slower mix shift, which can compress multiple before any earnings miss shows up. For TSLA, the stock still trades on a far more ambitious terminal story than current auto economics justify; absent a visible autonomy/robotics monetization milestone, the 6-18 month risk is multiple compression rather than a fundamental reset.

Contrarian takeaway: consensus is probably underestimating how much of the current AI build-out is already priced into the hyperscaler spend narrative and how sensitive NVDA could be to even a modest capex cadence slowdown. If next year’s spending growth disappoints, the first-order hit is NVDA multiple compression, followed by a second-order air pocket in suppliers and the broader AI complex. The thesis is falsified if hyperscaler capex guides keep reaccelerating into the next two earnings seasons and AAPL/TSLA show durable margin or delivery inflection despite higher input costs.