The Roundhill Memory ETF, listed under the ticker DRAM, reached $10 billion in assets under management (AUM) in just 50 days. The fund launched on April 2 and crossed the $10 billion mark on Friday — since inception it has risen 87 percent and attracted $8.6 billion of fresh inflows.
According to Morningstar, no prior ETF reached $10 billion so quickly: the previous thematic record-holder, the iShares AI Innovation and Tech Active ETF (BAI), took 556 days to hit the same milestone. DRAM achieved the result in a small fraction of that time.
Heavy concentration in three memory-chip makers
The DRAM portfolio is highly concentrated: 73 percent of its assets are invested in just three companies — South Korea’s SK Hynix and Samsung Electronics, and the U.S. firm Micron Technology. Micron’s shares rose as much as 19 percent on the Tuesday before publication, pushing the company’s market capitalization above $1 trillion.
Investor enthusiasm tied to data-center buildout
Investor interest is largely driven by demand for data-center capacity needed to train and run AI models. The Goldman Sachs Global Institute estimates roughly $7.6 trillion will be spent on data-center construction over the next five years, with a large share of that spending going to semiconductors.
Morningstar data show that assets in ETFs and funds focused on AI and big data have more than doubled since early 2025, reaching $70.8 billion. That expansion has been driven mainly by strong inflows rather than price gains.
Warnings from analysts and historical parallels
Todd Rosenbluth, head of research at TMX VettaFi, cautioned that the fund’s large exposure is concentrated in three stocks that all operate in the same industry, so the ETF does not cover different segments of the AI supply chain.
Before DRAM, the Fidelity Wise Origin Bitcoin Fund held the record for the quickest rise to $10 billion, reaching it in 81 days in March 2024 after U.S. regulators finally authorized crypto ETFs. Nate Geraci, president of NovaDius Wealth Management, said DRAM is a textbook example of a manager launching a product with perfect timing, a strong ticker symbol, and an investment thesis aligned with market demand, and he expects a wave of copycat funds.
Morningstar’s earlier research, however, urges caution: investors in thematic ETFs often underperform the targeted sector because they tend to buy at highs and sell at lows. Daniel Sotiroff, deputy director of ETF research at Morningstar, compared the current AI frenzy to the thematic investment mania seen during the COVID-19 period, when capital flowed into renewables and cannabis. He expressed concerns about the current valuations of chipmaker-focused ETFs.
Conclusion
Roundhill Memory (DRAM) reached $10 billion in AUM unusually quickly, reflecting investor optimism about AI-related data-center spending and demand for memory chips. But the ETF’s heavy concentration in three firms and broader concerns about thematic ETF performance suggest investors should weigh potential risks even as similar funds are likely to appear on the market.
This article does not constitute investment advice or a recommendation.
