David Tepper's Appaloosa Management now holds $3.1 billion across three positions—Amazon, Micron Technology, and Taiwan Semiconductor Manufacturing—representing 40 percent of the fund's disclosed $7.7 billion in long equity. The positioning, disclosed in Appaloosa's latest 13F filing, marks the highest three-name concentration in the fund since Tepper's pre-2019 China consumer thesis.
The allocation breaks roughly even across the trio. Amazon represents the largest single line at approximately $1.15 billion, followed by Taiwan Semi at $1.05 billion and Micron at $900 million. Appaloosa added $420 million to its Taiwan Semi stake in the quarter and built the Micron position from $310 million to $900 million. The fund did not materially alter its Amazon weighting. No other holding in the portfolio exceeds $650 million.
This matters because Tepper rarely concentrates without a macro catalyst he believes the market is mispricing. His 2009 bank concentration—BAC, Citi, Wells—returned 130 percent over eighteen months. His 2016 energy positioning in Chesapeake and Apache delivered 90 percent before he exited in early 2017. The common thread: Tepper sizes when he sees a supply-demand mismatch with a twelve-to-eighteen-month clearing mechanism. Here, the mismatch is AI capex against semiconductor fabrication capacity and hyperscale cloud infrastructure spend.
The Taiwan Semi and Micron pairing is a bet on memory and logic capacity constraints. TSMC controls 54 percent of global foundry revenue and 92 percent of sub-5nm production. Micron holds 24 percent of DRAM and 11 percent of NAND. Nvidia, AMD, and every hyperscaler building custom silicon depend on TSMC. Training clusters depend on high-bandwidth memory from Micron and SK Hynix. The Amazon position layers in the other end: the hyperscale buyer that converts chip capacity into inference revenue. Amazon Web Services now runs 31 percent of global cloud infrastructure and is deploying its Trainium2 chips at scale. The thesis is vertical: chip supply bottleneck, cloud margin expansion, and capital efficiency gains as inference costs decline.
Operators and allocators should watch TSMC's April revenue print, which will reflect March orders for Nvidia's Blackwell ramp and Amazon's Trainium production. Micron reports fiscal Q2 earnings on March 19; guidance will clarify HBM3E allocation between hyperscalers. Amazon's Q1 call in late April will disclose AWS revenue growth and capex run rate. If AWS capex stays above $75 billion annualized and operating margin holds above 38 percent, Tepper's thesis—that cloud operators can spend into AI infrastructure without margin compression—remains intact.
Appaloosa has filed only two 13F amendments in the past six years. Both preceded exits within ninety days.