Billionaire David Tepper’s Appaloosa Management Has 31% of its portfolio Invested in These 2 Artificial Intelligence (AI) Stocks

Micron reported fiscal Q4 2026 revenue of $54.2 billion, up 379% year over year, and non-GAAP EPS of $33.42, up 1,003%; memory-chip tightness tied to AI demand could persist through 2028. Its shares have risen more than 200% this year, while Appaloosa cut its Micron stake about 41% in Q2 but it remained 15.06% of the portfolio. Amazon's Q2 revenue rose 20% to $200.6 billion, AWS sales increased 37% to $42.2 billion, and the company raised its 2026 capex projection to $220 billion from $200 billion; Appaloosa increased its stake about 15.7%.

Analysis

The key asymmetry is earnings quality, not headline growth. MU’s low forward multiple may use peak-cycle earnings as its denominator: if memory prices or mix normalize, earnings can fall faster than revenue, making the stock look cheaper than it is. A long supply lead time and competing capacity additions could eventually weaken pricing power; Samsung and SK hynix are relevant supply-side checks. Conversely, tight memory supply can raise costs for cloud operators, so Amazon’s larger capex budget is not automatically a positive for near-term AWS returns. For AMZN, the test is whether AI demand converts into durable AWS operating profit and free cash flow after depreciation—not bookings or capex alone. Over the next 1–3 months, track memory pricing/supply commentary and Amazon’s AWS margins and capex-to-cash-flow trajectory. Over 6–18 months, utilization and new memory capacity determine MU’s earnings durability, while cloud workload monetization determines whether AMZN’s investment earns an adequate return. Contrarian read: MU’s low multiple is not a reliable cushion at peak profitability, but an abrupt bearish turn on memory could also miss a genuine multi-year capacity constraint. Tepper’s portfolio changes are not an independent catalyst.

Key Decisions for Investors

• Avoid chasing MU after the sharp run-up; consider trimming or keeping exposure hedged until earnings are assessed against mid-cycle memory pricing rather than peak-period EPS. Revisit if supplier commentary shows capacity coming online faster than demand or pricing weakening.
• Prefer a relative-value position: long AMZN against a smaller, dollar-neutral short or underweight in MU, entered in tranches rather than on a single headline. The thesis is comparatively more diversified monetization versus greater exposure to memory-cycle earnings; cap the pair because AMZN’s AI capex may also pressure cash flow.
• For AMZN, add on weakness only if AWS growth remains accompanied by stable/improving operating economics and a credible path from capex to cash generation. Reduce or exit if AWS margins deteriorate as investment rises, or if capex expands without evidence of monetization.
• Falsifiers: sustained memory-price strength alongside disciplined capacity would undermine the MU underweight; material AWS margin improvement and cash-flow conversion would support AMZN, while weaker cloud demand or rising investment without returns would invalidate the long.