The Micron-led Roundhill Memory ETF (DRAM) just reached $6.5 billion in assets in 36 days, making it the fastest ETF to hit that mark and beating the early 2024 bitcoin ETF boom, according to Bloomberg Intelligence ETF analyst Eric Balchunas.
BlackRock’s iShares Bitcoin Trust ETF (IBIT) took 43 days to reach $6.5 billion, while Fidelity’s Wise Origin Bitcoin Fund (FBTC) took 51 days.
The ETF has also nearly doubled out of the gate. Balchunas said Monday after the bell that DRAM was up 98% since launching five weeks ago, calling it “easily the best perf ETF out of the gate,” while also noting it was the seventh-most-traded ETF Monday with $4.5 billion in volume.
That makes the AI memory trade one of the hottest ETF stories on Wall Street in the post-pandemic era — and a major acceleration from last week’s surge, when DRAM was barely a month old.
The fund’s latest jump in assets came as DRAM soared 13% Friday and pulled in another $1 billion in inflows, according to Balchunas. In early trading Tuesday, both DRAM slipped as the memory trade cooled after Monday’s surge.
Micron makes up 27% of DRAM — barely edging out SK Hynix (000660.KS) at 26% and Samsung Electronics (005930.KS) at 20%, according to the fund’s latest holdings. Sandisk (SNDK), Kioxia (285A.T), Seagate Technology (STX), and Western Digital (WDC) round out the next tier.
Micron closed at a fresh record Monday, its 26th record close of the year, before pulling back Tuesday. The stock was down 3.5% in early trading, tracking its worst day in two weeks.
That’s also where the bull case gets interesting.
The concentration of the ETF is the point. DRAM is a focused bet on the memory supply chain, where investors are treating DRAM, storage, and high-bandwidth memory as critical pieces of the AI build-out.
In a Monday note, D.A. Davidson analysts argued that investors are still underestimating “the new math of memory in the AI age,” writing that “the bigger the models, the more memory they require.”
The note explains that longer AI context lengths — essentially how much information a model can process at once — create a cycle of more memory demand, better models, and still more memory needs.
But that same setup cuts both ways.
Memory has a long history of boom-bust cycles, and even the bullish D.A. Davidson note flagged the risk directly. The firm warned that the industry is “historically prone to booms and busts” and that weaker demand as Micron expands capacity could lead to oversupply and lower memory prices.
Nevertheless, the firm reiterated a Buy rating on Micron and a $1,000 price target — about 34% above Friday’s close.
