The 3 "Magnificent Seven" Stocks That I'm Buying Now

The article identifies Nvidia, Alphabet, and Amazon as its preferred Magnificent Seven investments, led by Nvidia's projected 70% revenue growth in 2027 and 14.4x forward P/E valuation. Nvidia estimates the five largest AI hyperscalers will lift data-center capex to $1.3 trillion in 2027 from nearly $800 billion in 2026; if Nvidia meets estimates and retains its 29x trailing earnings multiple, the author argues its shares could double by January 2028. Alphabet and Amazon are expected to spend $200 billion and $220 billion, respectively, on data-center capex this year, supported by Google Cloud's 82% Q2 revenue growth and 36% operating margin and AWS's 37% Q2 growth.

Analysis

The useful signal is not the promotional valuation arithmetic but the implied capex concentration: sustained hyperscaler spend supports NVDA’s accelerator revenue, while AMZN and GOOG bear the near-term cash-flow and depreciation burden. The market will increasingly distinguish purchased capacity from monetized capacity; cloud growth alone is insufficient if incremental AI workloads carry lower utilization or require aggressive pricing. Over the next 1-3 months, NVDA remains the cleanest read-through on capacity commitments, whereas AMZN/GOOG need backlog, remaining-performance-obligation trends, and cloud operating-margin durability to validate returns on spend.

Second-order beneficiaries are AI networking and power/thermal infrastructure—AVGO, ANET, VRT and ETN—where deployment bottlenecks can preserve pricing even if accelerator unit growth moderates. Conversely, MSFT and META are the relevant competitive risks: both can absorb elevated capex with different monetization paths, and any evidence that internally designed silicon or alternative accelerator architectures reduce NVDA content per data center would pressure NVDA’s terminal multiple before it materially affects revenue.

Consensus appears too willing to annualize current cloud growth and too focused on NVDA’s forward P/E. A lower multiple can reflect unusually elevated earnings estimates rather than a free option on further upside. The key 6-18 month risk is a hyperscaler capex digestion phase: if AI revenue fails to scale proportionately with depreciation and power costs, procurement shifts from scarcity-driven ordering to ROI-driven optimization, compressing NVDA growth and exposing cloud-margin expectations simultaneously.

Key Decisions for Investors

• Maintain NVDA as an overweight only through the next hyperscaler earnings cycle; add on evidence of upward 12-month revenue estimates and unchanged gross-margin guidance. Use a 10-12% downside stop from entry or exit if a major customer signals 2027 capex moderation; upside case is estimate revisions plus sustained premium multiple over the next 3-6 months.
• Prefer a basket long ANET/VRT/ETN versus a smaller NVDA hedge rather than adding outright hyperscaler beta. Networking and electrical infrastructure benefit from cluster build-outs but are less exposed to a single-chip competitive transition; reassess if data-center order books or backlog conversion weaken for two consecutive quarters.
• Pair long GOOG / short AMZN over 1-3 months if cloud-margin data confirm the claimed divergence: GOOG offers a clearer path for incremental cloud profit to offset AI capex, while AMZN requires AWS reacceleration without deterioration in consolidated free cash flow. Cover if AWS growth reaccelerates by more than 5 percentage points or Google Cloud margin falls below its recent run rate.
• Do not chase a broad Mag-7 trade from this article. Set an alert around quarterly capex guidance, AI revenue disclosure, and depreciation growth for AMZN, GOOG, MSFT and META; a capex increase without matching revenue/backlog acceleration is the actionable bearish catalyst for AI infrastructure multiples.