Sandisk has climbed 695% in 2026, extending a run that began when it finished 2025 as the best-performing stock in the S&P 500, and the memory maker now sits at the center of the question every momentum investor is asking about next year.
The precedent is unusually specific. Over the past decade, the top-performing S&P 500 stock in each calendar year returned a median 91% during the following year, according to the analysis that flagged Sandisk's position. That figure has become the reference point for traders deciding whether to hold the AI-infrastructure trade into 2027 or take profits on a position that has already tripled several times over.
"The follow-year stat is a description of momentum persistence, not a forecast," said Priya Mehta, an equity market structure analyst who tracks index flows and positioning. "When the leader is a memory name tied to a capex cycle, the second year depends almost entirely on whether supply stays tight and hyperscaler spending keeps rising."
The mechanism behind Sandisk's advance is the same one lifting the broader memory complex. AI data centers require more DRAM, more high-bandwidth memory and more storage, and advanced memory cannot be produced quickly because HBM requires complex manufacturing, packaging and customer qualification. That gap between demand and supply has given memory makers pricing power that historically belonged to logic and GPU suppliers.
Micron Technology, the largest U.S. memory maker, has become the clearest read on the same thesis, with DRAM, NAND and HBM exposure across data centers, AI servers and enterprise storage. The bull case for both names rests on the argument that memory is no longer a boom-and-bust commodity business but a structural AI infrastructure supplier. The bear case is simpler: when prices rise, producers eventually add capacity, and margins can fall quickly once customers slow orders.
Competitive pressure sits with Samsung Electronics and SK Hynix, which remain the two largest memory producers and can shift the supply picture faster than any single demand signal. SK Hynix shares fell on Sept. 23 as investors took profits after a rally to record highs, with the decline coinciding with the approaching conclusion of a 40 trillion won buyback program and caution ahead of a U.S.-China meeting on semiconductor supply chains.
The wider AI hardware complex is already showing the strain of a crowded trade. MaxLinear dropped on Sept. 23 as part of a semiconductor sector retreat following a large year-to-date rally, compounded by a CEO share sale exceeding $30 million. FormFactor traded lower as investors questioned whether AI infrastructure spending may be plateauing. Both moves came after the Federal Reserve raised rates by 25 basis points and issued hawkish forward guidance, which lifted Treasury yields and pressured high-multiple technology names.
Not every corner of the complex is retreating. Astera Labs extended a multi-day winning streak on Sept. 23 on optimism about its role in cloud and AI connectivity. Arm Holdings surged after its chief executive said demand for the company's chip architecture is off the charts. STMicroelectronics rose after projecting that its AI data-center revenue will surpass $2 billion by 2027. The divergence matters: capital is rotating within AI hardware rather than leaving it, which is the pattern that would need to hold for a second-year advance in the group's leaders.
For Sandisk specifically, the 91% median is a small sample drawn from a decade of leaders that included very different businesses, from energy producers to software platforms. A memory maker's second year is governed by a supply cycle that a software company never faces. The relevant question is not whether the historical average repeats, but whether HBM demand, DRAM pricing and hyperscaler capex guidance keep confirming the thesis through the next several earnings cycles.
The next confirmation points are the memory makers' quarterly reports and any update to hyperscaler capital expenditure plans, both of which will test whether the supply-demand gap that produced a 695% gain is narrowing or widening. Until those numbers arrive, the 91% figure functions less as a target than as a measure of how much expectation is already priced into the stock.
This article is for informational purposes only and does not constitute investment advice.