Semiconductor ETFs Hit New Highs on AI Memory Rotation

Benzinga reported Tuesday that semiconductor ETFs are setting fresh records as investors look well past graphics processors toward the memory infrastructure underpinning modern AI systems.

SMH Logs Its Strongest Quarter on Record

The VanEck Semiconductor ETF climbed to an all-time high on Tuesday. The fund has risen roughly 57% in the second quarter alone and is up more than 65% year-to-date. That makes the current period the strongest quarter in the fund’s history.

The advance has drawn significant capital. Approximately $5.6 billion in net inflows have entered the fund so far this year, according to ETFDb data cited by Benzinga.

Institutional conviction has followed. Billionaire hedge-fund manager Dan Loeb recently initiated a meaningful new stake in SMH through his firm Third Point, according to the latest 13F filings. The move signals broad-based confidence in the AI chip trade rather than any single company.

Also Read: Nvidia’s Blackwell Ramp Lifts Data-Center Revenue to Record

A Structural Shift in How Memory Chips Are Valued

The background matters here. Micron Technology has long been treated as a commodity-driven cyclical stock, its fortunes tied to unpredictable swings in DRAM and NAND pricing.

That framing is now being challenged. UBS recently argued that high-bandwidth memory, or HBM, used in AI accelerators and data-center servers is structurally changing Micron’s business model. The bank pointed to newer multi-year supply agreements featuring fixed-volume commitments and partially locked-in pricing. That framework is a meaningful departure from the older spot-driven arrangements that historically amplified the memory industry’s notorious boom-bust cycle.

UBS projected Micron earnings per share of $155 in 2027 and $167 in 2028, with EPS holding above $100 through 2029. The bank suggested the market could begin awarding Micron a valuation multiple closer to those seen at other structural AI infrastructure players.

Also Read: What Rising AI Capex Means for Chip Suppliers in 2026

ETF Investors Broaden Their AI Exposure

The wider implication for ETF holders is that the AI hardware trade is diversifying. During the early phase of the generative AI boom that followed ChatGPT’s late 2022 launch, most semiconductor fund gains were concentrated in Nvidia.

Investors are now looking at memory suppliers, hyperscaler infrastructure plays and data-center supply chains as separate sources of return. HBM chips have become essential for training and running advanced AI models at scale. That creates a durable demand floor that did not exist in prior chip cycles.

Sector ETFs tied to AI infrastructure are attracting fresh attention as a result.

Read Next: AI Infrastructure Spending Shows No Sign of Slowing in 2026