DRAM vs HBMX: Which Is the Better Pick in 2026?

Overview

Both DRAM and HBMX are thematic ETFs built around memory semiconductors and the AI infrastructure ecosystem, but they differ sharply in scope and management style. DRAM passively tracks companies in memory production with a 0.65% expense ratio and $25.9B in assets, while HBMX is an actively managed, concentrated fund that extends beyond DRAM and NAND producers to include packaging, testing, and equipment suppliers, charging 0.95% with $25.4M in assets.

How they differ

The most immediate difference is management approach: DRAM is a rules-based, passive tracker of memory companies, whereas HBMX uses active stock selection and concentration to pursue what its prospectus describes as a "memory stack" — the full supply chain from chipmakers through equipment and assembly services. DRAM has a broader asset base at $25.9B compared to HBMX's $25.4M, reflecting vastly different adoption curves since both funds launched in mid-2026. Neither fund reports a distribution rate or yields income; both are designed for capital appreciation. DRAM's 0.65% expense ratio undercuts HBMX's 0.95%, a gap typical between passive and actively managed strategies. The portfolio construction also differs: DRAM targets "at least 80% of net assets in the equity securities of memory companies," a narrower mandate than HBMX's supply-chain inclusive approach, meaning HBMX may hold equipment and semiconductor-equipment makers where DRAM would not.

Who each is best for

• DRAM: Fits investors seeking pure-play memory semiconductor exposure through a passive index structure with lower fees, comfortable with a simpler rules-based methodology that avoids active manager discretion.

• HBMX: Designed for investors who believe the memory opportunity extends beyond chipmakers themselves to the tools, testing platforms, and advanced packaging layers that enable AI hardware, and who are willing to accept active management and higher fees for concentrated conviction in that broader ecosystem.

Key risks to know

• Concentration in a single semiconductor subsector. Both funds limit themselves to memory-related exposure. A downturn in DRAM pricing, NAND demand, or AI infrastructure capex would affect both simultaneously; holdings likely overlap significantly, so owning both does not reduce this risk.

• Semiconductor cyclicality and capex sensitivity. Memory chip demand is historically volatile and tied to PC and data-center refresh cycles. An inventory correction or delayed cloud spending could compress valuations across both portfolios.

• Active management and portfolio drift in HBMX. Active selection introduces manager skill risk; HBMX's extended mandate (packaging, testing, equipment) means it may hold companies more loosely tied to memory fundamentals, creating potential underperformance if those satellite holdings lag core semiconductor strength.

• Valuation dependency on AI narrative persistence. Both funds are positioned to benefit from sustained AI infrastructure demand. A pivot in chip-architecture preferences away from memory-intensive models, or a slowdown in AI datacenter buildout, could erode investor appetite for these thematic buckets. Both are thematic bets on AI durability and semiconductor memory demand — verify that your holdings in memory, AI infrastructure, or chipmakers don't overlap with either before committing capital. Past performance does not predict future results.