Meet the Unstoppable Memory ETF With 74.7% of Its Portfolio Parked in Micron, Samsung, and SK Hynix

Roundhill launched the Roundhill Memory ETF (DRAM) on April 2, and it has already surged 121%—driven by a 74.7% concentration in Samsung (25.2%), SK Hynix (24.8%), and Micron (24.8%). The backdrop is an AI-driven HBM4 shortage, with Micron’s HBM4 touting 60% more capacity and 20% better energy efficiency. However, UBS survey data suggests ~60% of businesses are beginning to cut AI spending by moving to cheaper, less compute-intensive models, and accelerating capacity buildouts could eventually pressure memory prices and trigger a stock correction.

Analysis

The tradeable read is not that memory is ‘good,’ but that the market is still paying peak-cycle multiples for a supply shock that can unwind once capacity, packaging, and customer procurement behavior normalize. The cleanest relative winner is Micron: the U.S.-listed name has the most obvious scarcity premium and the best chance to re-rate on AI credibility, while a broad memory basket dilutes that upside with Korean FX, governance, and cycle beta.

The bigger second-order loser is not another chip vendor but any OEM that cannot fully pass through higher BOM costs. Apple is the most visible proxy: memory inflation can compress gross margin or force mix shifts before it shows up in headline guidance, and that effect is usually most visible over the next 1-2 quarters. If device makers start shipping lower-memory configurations or if enterprise buyers migrate to cheaper AI models, that is a leading indicator that pricing power is already leaking.

Contrarian risk to the bearish ETF view: HBM is not commodity DRAM, and the real bottleneck is yield plus advanced packaging, so supply may stay tighter for longer than consensus expects. That makes any outright short of the memory complex dangerous unless HBM pricing rolls over or Micron’s next guide stops beating. Falsifiers: a sustained pause in spot/contract price increases, or evidence that enterprise AI capex is re-accelerating rather than downshifting.

Key Decisions for Investors

• Favor MU over DRAM: initiate or add to MU on any 5-10% pullback over the next 1-3 weeks; target a 2:1 upside/downside setup into the next earnings cycle. Stop if Micron’s guide or HBM pricing momentum deteriorates.
• Short DRAM or buy DRAM puts on strength if the ETF extends its current momentum; this is a crowded, concentrated wrapper around a cyclical supply squeeze. Cover if HBM4 lead times keep extending or if the top-three names continue to raise guidance.
• Relative-value pair: long MU / short AAPL for the next 1-3 months to express memory-cost inflation as a margin headwind at the device layer. Falsify the trade if Apple successfully offsets costs with pricing and mix without unit deterioration.
• Watch-list, not trade yet: if UBS-style evidence of lower-intensity AI adoption broadens, reduce exposure to the entire HBM complex and rotate into higher-quality semis with less cycle beta; the signal to act is two consecutive quarters of weaker capex commentary from hyperscalers.