VanEck Semiconductor ETF Gains from $500B Chip Deal and 49% 2025 Rally
SMH, which holds 49.8% of assets in Nvidia, TSMC, Broadcom, Micron and AMD, stands to benefit from the $500B Taiwan–US chip manufacturing deal that will drive U.S. fab investments. The ETF delivered a 49% return in 2025 and trades at 33× trailing earnings.
1. Taiwan-US $500B Chip Deal Spurs U.S. Fab Investments
The recently announced $500 billion agreement between Taiwan and the U.S. to bolster semiconductor manufacturing will channel massive capital into domestic fabrication plants over the next decade. SMH, which tracks leading semiconductor equipment makers, foundries and chip designers, stands to benefit directly from expanded factory build-outs. Industry forecasts project U.S. fab capital expenditures to grow at a compounded annual rate of 12% through 2030, and SMH’s broad exposure positions it to capture growth across equipment orders, silicon processing and advanced node capacity.
2. Concentrated Exposure to AI Powerhouses
SMH’s top five holdings—Nvidia, Taiwan Semiconductor Manufacturing Company, Broadcom, Micron Technology and Advanced Micro Devices—represent nearly 50% of its assets, while the top 10 account for over 73%. This concentration provides investors with direct access to firms driving AI compute advances, from GPU designers and logic foundries to memory suppliers and lithography equipment manufacturers. By focusing on the core players in the AI value chain rather than a single equity, SMH delivers diversified yet targeted semiconductor leverage.
3. Track Record of Outperformance
Over the past decade, SMH has generated annualized returns of approximately 30.9%, more than double the S&P 500’s 12.9% over the same period. In 2025 alone, the ETF posted a gain near 49%, significantly outpacing broad-market benchmarks. Trading at a price-to-earnings multiple that aligns with large-cap tech peers, SMH has demonstrated resilience through multiple industry cycles, underpinning its appeal as a core holding for long-term growth investors.
4. Long-Term AI Infrastructure and Inference Growth
As enterprise AI spending shifts from episodic training to sustained inference workloads, demand for GPUs, power-efficient memory and networking components is expected to accelerate. Deloitte forecasts inference compute will constitute two-thirds of total AI processing by 2026, up from one-third in 2023. This durable, usage-driven demand underscores a multi-year runway for SMH’s constituents, suggesting the ETF is well positioned to capture the next phase of semiconductor adoption.