Peter Thiel and David Tepper Are Buying Amazon Stock Hand Over Fist: Should You Follow the Smart Money?

Amazon is increasing AWS data center capex to $220B for 2026 (from $200B earlier guidance), targeting continued AI-driven cloud demand. In Q2, AWS revenue grew 37% YoY and operating income rose 64%, indicating operating leverage as investments convert to monetizable capacity. The stock is already up ~10% in Q3, and the article argues demand will outstrip planned capacity through 2027, with signs of 2028 demand emerging.

Analysis

The key market mechanism is not just growth, but scarcity pricing: when hyperscaler capacity is tight, the firms with deployed compute are effectively selling an asset class with rising marginal value. That should keep demand flowing first into the picks-and-shovels stack — NVDA, networking, power, and data-center infrastructure — before it fully converts into AMZN free cash flow. In other words, the headline beneficiary is Amazon, but the cleaner near-term monetization often sits one layer upstream.

The risk is that investors may be paying today for operating leverage that arrives much later. Heavy capex can support revenue acceleration while still suppressing per-share cash generation, and that is usually where multiples crack if utilization or pricing discipline disappoints. The main falsifier over the next 1-3 quarters is any AWS deceleration from the current high-growth cadence without a corresponding slowdown in spend; that would reframe the story from compounding to deferred returns.

Consensus is likely underestimating the second-order drag on the broader software layer. If AI budgets stay concentrated in infrastructure, application vendors can see delayed purchase cycles as customers prioritize compute, which supports a bifurcation between infrastructure winners and software laggards. The contrarian view is that AMZN may be less attractive than it looks on a one-year basis if the market starts demanding proof of cash conversion rather than rewarding optionality.

Key Decisions for Investors

• Accumulate AMZN only on a 3-5% pullback over the next 2-6 weeks; use a 6-12 month horizon and expect upside if AWS sustains ~30%+ growth and capex begins to monetize. Falsify the thesis if AWS growth slows materially or margins fail to expand.
• Express the theme more cleanly with a long NVDA / short AMZN relative-value trade over 6-9 months. NVDA monetizes the build-out immediately, while AMZN carries execution and free-cash-flow risk; cover if AMZN shows stronger-than-expected operating leverage.
• Add a watchlist alert on data-center infrastructure beneficiaries such as DLR and EQIX for 3-12 months; they benefit from power and rack scarcity without bearing Amazon’s capex burden. Treat as a secondary trade, not a primary recommendation, because valuation already discounts some AI demand.
• Do not chase the 10% post-rally move in AMZN on a one-week horizon; wait for either a digestion period or the next earnings update. The risk/reward is better after the market has had time to separate capex optics from actual monetization.