The AI memory-chip trade is getting a new, more speculative flavor. Just days after a filing for a leveraged DRAM ETF, GraniteShares has submitted paperwork for an "Autocallable DRAM ETF," according to Bloomberg Senior ETF Analyst Eric Balchunas. The move signals that Wall Street is racing to package the explosive momentum around memory chips into increasingly complex products.
The underlying Roundhill Memory ETF (DRAM) has been one of 2026's hottest launches. Since debuting in April, it's soared almost 80% and rapidly amassed roughly $6.25 billion in assets under management as investors pile into AI infrastructure plays. Unlike broad semiconductor ETFs, DRAM is narrowly focused on companies tied to high-bandwidth memory, DRAM chips, NAND flash storage, SSDs, and related technologies—increasingly viewed as the "picks and shovels" of the AI boom.
The filing comes on the heels of Themes ETF Trust's submission for the Leverage Shares 2X Long Memory Daily ETF, designed to deliver twice the daily performance of the DRAM ETF. Now, GraniteShares is taking it a step further with an autocallable structure.
AI's New Bottleneck Trade
Top holdings in the DRAM ETF include Samsung Electronics, SK hynix, Micron Technology (MU), SanDisk Corp (SNDK), Western Digital Corp (WDC), and Seagate Technology Holdings (STX). The AI infrastructure race has dramatically reshaped sentiment around memory stocks, once viewed as among the semiconductor industry's most cyclical segments. Training and inference workloads for large AI models require enormous bandwidth and storage capacity, turning memory chips into a critical supply bottleneck.
That demand shock has triggered eye-popping rallies across the sector. SanDisk shares have surged roughly 384% this year, while Micron has climbed more than 116%. Seagate has jumped nearly 158% as hyperscalers continue ramping AI data-center spending. Industry pricing trends have added further fuel: memory-chip prices reportedly climbed between 80% and 90% during the first quarter as demand overwhelmed production capacity. The pricing boom has translated into blockbuster earnings growth across the industry.
From Leveraged ETFs to Structured Products
The proposed GraniteShares autocallable ETF suggests issuers now believe investor appetite extends beyond traditional ETF wrappers into structured-income and derivatives-based strategies. Autocallable products typically use options-linked structures that can generate enhanced income or capped upside under specific market conditions, but they also introduce additional complexity and downside risks. Combined with an already volatile thematic ETF, the filing underscores how aggressively firms are trying to capitalize on the AI-memory frenzy.
At the same time, warnings about overheating are growing louder. Investor Michael Burry recently cautioned about bubble-like conditions in parts of the semiconductor market, while technical indicators show several memory-related names trading in deeply overbought territory. The DRAM ETF's Relative Strength Index has reportedly climbed above 80, a level often associated with stretched momentum.
Still, the rapid progression from niche thematic ETF to leveraged funds and now autocallable products highlights one thing clearly: everybody on Wall Street wants a piece of DRAM.