Billionaire David Tepper of Appaloosa Slashed His Micron Stake and Piled Into a "Fabulous" AI Stock for a 6th Consecutive Quarter

David Tepper’s Appaloosa cut its Micron stake by 41% in the June quarter after Micron shares more than tripled to over $1,200, but warned that an AI demand/bubble risk could still pressure memory names. At the same time, Appaloosa increased its Taiwan Semiconductor (TSMC) position by 24% via a sixth consecutive-quarter buy, adding 322,500 shares as TSMC holds ~3/4 of global foundry share and is described as benefiting from AI-driven GPU infrastructure build-outs with potential for 50%+ revenue CAGR. Net impact is incremental, signaling mixed positioning rather than a fresh fundamental shock.

Analysis

The actionable read is a preference for the less reflexive part of the AI stack: foundry and advanced packaging exposure over memory. That matters because foundry monetization is driven by utilization and node mix, while memory is still the cleaner cyclical beta trade; if AI capex merely cools rather than collapses, TSM should hold up better than MU, but MU will likely suffer the larger multiple reset if growth decelerates.

Near term, the filing itself should not move fundamentals; it is mostly a positioning data point. The 1-3 month catalyst path is earnings and guidance: if MU can keep HBM pricing tight and maintain delivery commitments, the de-risking thesis is wrong; if not, the market will likely rotate away from the most levered beneficiaries of AI optimism and into higher-quality tollbooths.

The contrarian issue is that the market may already own the “TSM is the AI toll road” narrative, so incremental upside may be capped unless utilization or pricing surprises materially to the upside. Conversely, MU may be the more interesting squeeze candidate if supply discipline persists and AI server demand stays stronger than skeptics expect. A Fed-driven rates backup or a broader AI multiple unwind would hurt both, but MU would likely underperform on a 6-18 month view because its earnings path is more cyclical and less defensible.

Key Decisions for Investors

• Pair trade: long TSM / short MU into any post-13F strength in semis; use a 1-3 month horizon and target relative outperformance of TSM if the market starts rewarding cash-flow visibility over cyclical beta. Falsifier: MU next earnings confirms sustained HBM pricing and raises the market’s confidence in demand durability.
• If already long MU, reduce exposure or add a short-dated collar ahead of the next print; the risk is not just a pause in AI spend but a sharper de-rating if inventory or pricing inflects lower. Falsifier: management commentary shows no sign of order digestion.
• Use pullbacks in TSM to add rather than chase; the better entry is on any macro/rate-driven weakness, not after a momentum extension. Risk/reward is strongest if the market is offering TSM at a discount because rates rise, while AI buildout expectations remain intact.
• Watch NVDA and the broader AI capex complex as confirmation rather than trade directionally from the filing; if TSM strength is paired with weaker MU and stable NVDA guidance, it reinforces a supply-chain mix shift away from memory toward foundry/equipment.
• If the semi tape turns risk-off, prefer hedging MU first; it should have the higher downside convexity versus TSM. A break in MU guidance or HBM pricing is the key trigger to press the relative-value short.