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

The economic transfer is less “AI wins” than “who converts capex into durable cash flow.” NVDA is the cleanest near-term beneficiary because every incremental dollar of hyperscaler spend still has a high attachment rate to its stack, but that advantage narrows if the buyers substitute toward custom silicon, networking, or delay rack deployment into 2025. The hyperscalers themselves are likely to see an earnings gap between reported revenue growth and true economic return as depreciation ramps; that creates a medium-term risk of multiple compression even if top-line growth looks fine.

AAPL and TSLA are vulnerable for different reasons: AAPL’s downside is slower-burn margin pressure from AI-driven component inflation, while TSLA’s downside is narrative fragility—its valuation still assumes future businesses that do not yet show up in current cash generation. The market is likely underestimating how much of the AI hardware boom is already forward-discounted in NVDA, while underestimating how much free-cash-flow resilience the best cloud platforms can retain if they maintain pricing discipline and utilization.

The key falsifier for the bullish NVDA / bearish AAPL-TSLA view is a visible slowdown in hyperscaler capex or a shift toward lower-NVDA-intensity architectures over the next 1-3 quarters. Conversely, if MSFT/AMZN/GOOG/META keep raising AI spend without a commensurate revenue reacceleration, the best relative trade becomes short the capex beneficiaries’ multiples, not the spenders’ stocks. Consensus is too linear on spend growth and too willing to pay for optionality in TSLA that remains 12-18 months from monetization.