David Tepper has never been a cautious investor. The founder of Appaloosa Management gained notoriety by making big, concentrated bets when most professional money was moving in the opposite direction. For example, he bought distressed bank debt during the 2008 financial crisis when everyone thought the system might not survive, producing returns that made his fund one of the best-performing in hedge fund history.
His name is associated with a particular kind of conviction that, depending on where you stand when the trade resolves, looks either brilliant or reckless. His portfolio’s present memory stock positioning follows the same pattern as a technological thesis: it is constructed around a belief that the market has not yet fully priced, heavy, and deliberate.
When David Tepper tripled Appaloosa’s ownership of Micron Technology in late 2025, the memory stock investment took on its current form. This move would have been noteworthy on its own, but it became even more significant when he followed it with an additional 11% increase in Q1 2026. One of the fund’s bigger assets was already Micron. Following these changes, it is now at the top of the portfolio with a focus that shows sincere belief rather than a varied exposure to the AI concept.
The thesis is explicit and important to comprehend: Micron is the main domestic manufacturer of high-bandwidth memory, which is the kind of specialized DRAM that AI data centers need in massive quantities in order for GPU clusters to operate at scale. The demand for HBM is coming from enterprise AI deployments and hyperscalers, where the computation requirements are both substantial and fundamentally expanding, rather than from consumer devices or casual computing.
Because SanDisk is new and Tepper made it his only new stock acquisition of the quarter, it is, in some ways, the more illuminating decision. It is not a tentative exploratory wager to open a position that instantly demands 3% of the portfolio. It’s a declaration. After being split off from Western Digital and reestablished as a stand-alone NAND flash storage company, SanDisk operates in a sector of the memory market that provides AI infrastructure with storage rather than compute.
This includes SSDs, enterprise storage arrays, and data centers that store the training datasets and model weights needed for AI systems to operate. The acquisition of SanDisk indicates that Tepper is considering the entire infrastructure stack rather than just the GPU-adjacent component that the majority of AI investors have already discovered.
He sold off European financial assets, aviation equities, and legacy tech brands to finance these positions. This kind of intentional rotation indicates a true shift in thesis rather than a margin reduction. One noteworthy aspect of the exit list is that European aviation and finance are both cyclically exposed and interest-rate sensitive in ways that might have been appealing in a different macroenvironment. Right now, rotating out of them and into memory chips reveals something about Tepper’s interpretation of the cycle’s subsequent phase.
Any competent watcher of memory stocks should approach this examination with the historical caution. The DRAM and NAND businesses have always seen booms and busts; they are fantastic during upcycles and terrible when supply meets demand and ASPs plummet. That type of decline followed the previous cycle high in 2021 and 2022. The current demand dynamic is structurally different, according to analysts, and Tepper’s stance appears to reflect this.
Hyperscalers are entering into multi-year supply deals that create an earnings floor in ways that prior consumer-driven cycles never could. Whether that argument remains true over the following two years or whether the bust still arrives on time is still up in the air. However, the argument makes enough sense that it would be incorrect to write it off as simple cyclical chasing. Tepper has made mistakes in the past. Additionally, he has been extremely accurate in ways that once appeared irrational.
