Amazon and Alphabet Both Cost More on Next Year's Earnings Than on Last Year's

Amazon trades around 22x trailing earnings and 30x forward earnings, but the article attributes the “cheap” look to a Q2 $53.4B pre-tax non-operating income windfall from Anthropic-related gains. Alphabet similarly shows lower effective valuation due to Q2 $98.0B other income driven largely by $98B of unrealized equity-security gains; the core operating engines are solid (Amazon +43% op income to $27.5B; Alphabet +30% op income to $40.8B) but both firms face forward earnings drag from heavy capex (Amazon capex plan ~$220B; Alphabet raised to as much as ~$205B). The piece favors Alphabet on valuation (about 27x vs Amazon’s ~30x on a forward basis) and points to stronger cloud acceleration (Google Cloud +82% growth) despite rising spending.

Analysis

This is less a profitability story than a quality-of-earnings reset. When headline EPS is inflated by marks and equity revaluations, the market stops paying for trailing earnings and starts paying for durable free cash flow; that usually compresses multiples for the lower-quality headline number before it benefits the better operator. The near-term risk is that both names look optically expensive into the next print cycle because depreciation lags capex, so reported earnings can stay under pressure even if the underlying businesses keep improving.

On a 1-3 month horizon, the biggest second-order winner is NVDA: hyperscaler capex intensity is effectively demand visibility for accelerators, networking, and related infrastructure, while the cloud platforms absorb more depreciation and working-capital drag. Between the two, GOOG/GOOGL has the cleaner setup because its margin expansion gives more cushion against the accounting drag; AMZN has stronger top-line momentum but a more fragile FCF profile if memory and infrastructure costs stay elevated.

Contrarian take: the consensus may be underestimating how much of the 'expensive' debate is just arithmetic normalization, not a deterioration in the franchises. If AI-related spend converts into sustained cloud share gains, current capex becomes a moat investment and the market will eventually look through the P&L noise. The thesis is falsified if the next 2 quarters show capex still rising while cloud growth decelerates or operating margins fail to hold; then the multiple should compress on real fundamentals, not optics.

Key Decisions for Investors

• Long GOOGL / short AMZN as a 1-3 month relative-value pair: target a modest 5-8% spread move in favor of GOOGL; stop if AMZN shows a clear free-cash-flow inflection or GOOGL cloud growth stalls.
• Buy NVDA on 5-7% pullbacks tied to hyperscaler capex headlines; 6-12 month upside remains tied to sustained >$200B spend trajectories, with invalidation if AMZN/GOOG cut capex guidance by >10%.
• Avoid initiating fresh outright longs in AMZN on headline EPS alone; wait for either a better entry after depreciation noise fades or evidence that incremental capex is driving margin expansion.
• Watch EPS revision trends in GOOG/GOOGL over the next 30-45 days: if estimates hold while the stock de-rates, that is the cleaner buy-the-dip signal than chasing a nominally cheap multiple.