Greg Abel Just Dumped Amazon. David Tepper Nearly Doubled His Stake. One of Them Is About to Look Very Smart.

Greg Abel reportedly exited Berkshire Hathaway's Amazon position in Q1 2026, likely reflecting the unwinding of stakes formerly managed by Todd Combs. At the same time, David Tepper nearly doubled Appaloosa's Amazon stake, apparently buying the stock after its February pullback as a bet on AWS and AI infrastructure spending. The piece is mostly interpretive commentary on portfolio moves rather than new fundamental disclosures, so direct market impact is limited.

Analysis

The real signal here is not the headline trade direction in Amazon; it is the split between a forced portfolio cleanup and a conviction add. That distinction matters because it means Berkshire’s sale is likely low-information for AMZN fundamentals, while Tepper’s buy is a higher-value read on where capital is being mispriced in the AI buildout. In other words, the market may be conflating de-risking at Berkshire with a negative view on Amazon’s capex cycle, when the more important takeaway is that large pools of capital still want exposure to AI infrastructure at a discount.

For AMZN, the near-term pain is tied to margin optics, not demand: aggressive AI and logistics investment suppresses reported profitability before it expands optionality in cloud, ads, and satellite connectivity. That creates a classic 2-4 quarter setup where the stock can rerate on operating leverage if AWS re-acceleration becomes visible, but can also derate again if management disappoints on near-term free cash flow. The second-order winner is anyone selling infrastructure into the AI arms race; the loser is any company that needs AI exposure but cannot fund the capex without compressing margins.

The contrarian point is that this is less a Buffett-vs-Tepper debate than a timing debate. If AI demand slows or hyperscaler spending pauses, Tepper’s added exposure becomes crowded beta; if spending remains durable, the pullback entry will look prescient. The market is still underestimating how much of AMZN’s valuation can be defended by AWS durability alone, even before newer monetization vectors become material.

For BRK.B, the portfolio housekeeping angle suggests little direct read-through to conglomerate fundamentals, but it does reinforce that Berkshire’s post-Buffett capital allocation may be less patient with smaller legacy positions. That can create modest pressure on similarly sized, non-core holdings across the market as investors anticipate future simplification and liquidation of inherited sleeves.

Key Decisions for Investors

• Long AMZN on weakness over the next 1-3 months; target a 12-18 month hold if AWS growth stabilizes and AI capex converts into higher revenue per unit of spend. Risk/reward is favorable if you can tolerate another 10-15% drawdown from macro risk.
• Buy AMZN call spreads 6-12 months out rather than outright calls to express upside from re-rating while limiting decay if margin compression lasts another quarter.
• Pair trade: long AMZN / short a basket of less-capitalized AI infrastructure beneficiaries that rely on external financing. The trade benefits if market starts rewarding balance-sheet-backed AI spend over speculative spend.
• Tactically avoid chasing BRK.B on this headline; treat the sale as non-fundamental cleanup unless subsequent filings show broader disposition patterns. Any weakness in BRK.B should be viewed as a portfolio-positioning opportunity, not a thesis break.
• For investors already long AI beta, trim some crowded names and rotate toward AMZN as a higher-quality catch-up trade if the market continues to reward AI exposure without penalizing capex.