Memory Shortage Just Entered Year Two: 3 ETFs to Watch Now

Market Backdrop: memory shortage just entered Year Two

The AI-driven memory crunch has stretched into its second calendar year, a reality that continues to reshape pricing, allocations, and capex across the chip supply chain. In late June, Micron Technology offered a cautious but constructive tilt for fiscal Q4 2026, guiding revenue to roughly $50 billion, a signal that the memory segment remains a key revenue driver for the sector. Lead customers report that HBM4 qualification samples are still rationed, underscoring ongoing tightness at the high end of memory technology.

Analysts say the pattern is less a quick snap back and more a gradual normalization—a trend that could keep certain memory categories tight through 2027. For investors, that means selecting vehicles that reflect the different layers of the memory and semiconductor stack rather than chasing a single heavyweight name.

Three ETFs as distinct bets on the same macro theme

Three exchange-traded funds stand out for reflecting various slices of the same AI-driven memory cycle: the VanEck Semiconductor ETF (SMH), the DRAM-focused ETF, and the Invesco PHLX Semiconductor ETF (SOXQ). Each offers a different balance of exposure, cost, and concentration, letting investors tailor risk and horizon to their view on memory and chips.

• SMH — VanEck Semiconductor ETF: The broadest blue-chip option among peers, SMH carries meaningful weight in Micron and other mega-cap memory and equipment names. In this cycle, the ETF’s exposure to core memory players helps capture upside from tight supply and equipment-maker gains tied to AI-inflected demand. Market observers point to a roughly 9% weighting to Micron in the fund’s latest composition as a lever for memory-sensitive alpha.
• DRAM ETF — a pure-play memory sleeve: This fund concentrates upstream memory exposure into a single ticker path, with about 73% of its portfolio focused on the three major memory fabricators: Samsung, SK Hynix, and Micron. The concentration means the ETF can swing with pricing cycles in DRAM and related memory technologies, offering a direct channel to the memory cycle rather than broader chip bets.
• SOXQ — Invesco PHLX Semiconductor ETF: A broader semis bet at a lower price of admission, SOXQ mirrors much of SMH’s risk-on exposure but comes with a leaner fee profile. Its expense ratio sits at about 0.19%, roughly half of SMH’s cost, making it a cost-efficient choice for long-horizon holders who want broad semiconductor exposure aligned with the memory cycle.

“Investors are juggling two realities: the memory shortage just entered a second year, and AI-driven demand shows no signs of a rapid normalizing spike,” said Lisa Romero, Senior Analyst at MarketBridge Research. “The ETFs above let you pick a lane—drill into memory specifically, take a broader chip-market stance, or opt for cost efficiency with a wide semis sleeve.”

How to think about each pick

The choice among SMH, DRAM, and SOXQ hinges on your time horizon and risk tolerance. Here’s a quick framework:

• Want memory-specific exposure? The DRAM ETF offers a direct bet on the downstream pricing cycle of the main memory players. If DRAM pricing or supply tightens further, this vehicle can amplify the move.
• Want a blue-chip, diversified chip basket? SMH provides a familiar, weighty exposure to flagship semiconductor names, with a meaningful tilt toward memory and infrastructure that supports AI workloads.
• Want broad semis exposure with lower cost? SOXQ balances risk by tracking a wide set of chipmakers, while saving on fees relative to SMH. It can work as a core sleeve in a diversified tech portfolio.

In practice, many investors are layering exposures: a core position in SOXQ for broad semis tilt, complemented by a DRAM sleeve to lean into the memory cycle, with SMH used tactically on supply-tight weeks when Micron and peers rally on earnings calls.

Data snapshot and current conditions

• Micron guidance: Fiscal Q4 2026 revenue expected near $50 billion, signaling durability in memory demand despite ongoing price competition.
• HBM4 timing: Qualification samples remain rationed among lead customers, underscoring continued supply discipline at the high-bandwidth memory layer used in AI accelerators and data centers.
• Memory share of SMH: Micron accounts for a material portion of SMH’s weighting, with other core positions supporting memory-enabled equipment and design cycles.
• DRAM ETF concentration: Roughly 73% exposure to Samsung, SK Hynix, and Micron keeps the fund tightly aligned with the DRAM pricing cycle and supplier dynamics.
• SOXQ cost advantage: Expense ratio around 0.19%, roughly half the cost of many broad semi ETFs, which can matter for long-horizon compounding.

“The memory shortage just entered a new chapter in 2025, and the capital allocation response is still playing out through 2026,” noted James Liu, Portfolio Strategist at Northline Capital. “For patient investors, cost-efficient exposure that captures the macro AI demand cycle can pay off as normalizing supply chains gradually unfold.”

Strategic takeaways for the memory cycle

The AI memory shortage just entered forward-looking commentary from executives and analysts, but the path forward remains nuanced. Several factors will shape outcomes in the second half of 2026 and beyond:

• Continued AI deployment in cloud data centers will keep memory and high-bandwidth memory demand elevated, supporting pricing power in select segments.
• Capex discipline among memory makers could temper the pace of new capacity, prolonging cycles and keeping tightness in certain layers.
• Technology transitions from DRAM to newer memory formats and stacked configurations may create pockets of volatility but also opportunities for ETF-based access to the macro trend.

For investors, the question is not whether the AI memory shortage just entered but how to position portfolios to navigate a multi-year cycle. The ETFs discussed here offer a practical blueprint: carve out a DRAM-led sleeve for direct memory exposure, deploy SMH for broader but still memory-relevant upside, and lean on SOXQ for a low-cost, diversified semiconductors core that still captures AI-driven demand dynamics.

Bottom line

The memory bottleneck remains a defining feature of the AI era’s supply chain. With Micron signaling durable revenue potential in late 2026 and HBM4 allocations still tight, the focus for many investors has shifted from rapid shifts to longer, more deliberate positioning. The memory shortage just entered a new phase, and the trio of ETFs—SMH, DRAM, and SOXQ—offers complementary routes to participate in the evolving landscape.

As market conditions evolve, practitioners will watch quarterly updates from hyperscalers and memory suppliers, hoping for clarity on timing and scope of any easing in the supply crunch. Until then, these funds provide a framework to align exposure with both risk tolerance and the pace of AI-driven demand.