- The Roundhill Memory ETF (DRAM) has surpassed $5 billion in assets under management just five weeks after its April 2 launch.
- A record-breaking $1.1 billion single-day inflow signals a shift in investor focus from logic chips to the AI memory bottleneck.
- Key holdings including Micron, SK Hynix, and Samsung represent over 70% of the portfolio as high-bandwidth memory demand spikes.
The Roundhill Memory ETF (DRAM) has emerged as the most explosive thematic launch of 2026, officially crossing the $5 billion assets under management (AUM) threshold. The fund’s rapid ascent was punctuated by a staggering $1.1 billion inflow in a single trading session, reflecting a massive rotation of capital into the semiconductor storage layer as Wall Street identifies memory as the primary bottleneck for Artificial Intelligence (AI) infrastructure.
Launched on April 2, 2026, the fund has benefited from a “Memory Supercycle” where demand for High-Bandwidth Memory (HBM) and DRAM has outpaced manufacturing capacity. While traditional consumer electronics demand remains soft, the data center requirements for training Large Language Models (LLMs) have pushed contract prices for AI-grade memory up by more than 50% year-over-year. This divergence has led institutional investors to seek pure-play exposure outside of broader semiconductor indexes.
The ETF’s strategy focuses on a concentrated basket of 13 global leaders. Industry titans Micron Technology (MU), SK Hynix, and Samsung Electronics anchor the portfolio, benefiting from what analysts describe as a “zero-sum constraint” in manufacturing; every wafer diverted to high-margin AI memory starves the rest of the market, maintaining a firm supply floor. According to Roundhill Investments, the fund is the first U.S.-listed vehicle to offer targeted exposure to this specific segment of the hardware value chain.
“The memory sector sits at the critical intersection of AI demand and constrained supply,” said Dave Mazza, CEO of Roundhill Investments. “Reaching these milestones so quickly reflects how decisively investors have embraced memory as the solution to the AI infrastructure gap.”
Despite the fund’s momentum, some market analysts warn of the risks associated with thematic ETF volatility. The fund carries a 0.65% expense ratio and is heavily concentrated, meaning it is susceptible to sharp corrections if the projected $240 billion global memory market growth faces technical or geopolitical hurdles. Currently, however, the “prepayment mania” from second-tier cloud providers suggests that the upward trajectory for memory pricing may persist through late 2026.
Disclaimer: This article is for informational purposes only and does not constitute advice of any kind. Readers should conduct their own research before making any decisions.
