The AI-driven surge in AI infrastructure investment, memory demand, and data-center spending is underpinning a strong, though imperfect, growth outlook for semiconductors and related ETFs, with NVIDIA and memory chips playing pivotal roles.

Global semiconductor sales jumped to about $403.3 billion in the second quarter of 2026, a 35.1% rise from the first quarter, with June sales up 123.6% year over year to $134.5 billion driven by demand across the Americas, Asia Pacific, and China.

Memory chips, especially DRAM and NAND, are in high demand as AI servers require high-bandwidth memory, with the Memory segment expected to surge roughly 250% year over year in 2026 and potentially account for more than half of total semiconductor revenues per industry analysis.

WSTS projects global semiconductor sales around $1.5 trillion in 2026 and about $1.9 trillion in 2027, signaling rapid growth driven by AI and advanced computing.

Data centers are the main growth engine as GPUs, networking gear, processors, and memory undergird AI workloads; major analysts expect hyperscalers to spend trillions on AI infrastructure over the 2026–2028 window, with estimates ranging from about $1 trillion in 2026 to $4.1 trillion in AI infrastructure spending by hyperscalers through 2028.

NVIDIA stands out as a key growth driver, projecting roughly 70% revenue growth in fiscal 2028, supported by ongoing demand for AI infrastructure, though tighter memory pricing could pressure margins.

Semiconductor ETFs offer growth potential but come with risks, including valuation concerns, volatility in AI spending, trade restrictions, supply shortages, and macro factors; notable funds include SOXX, SMH, SHOC, XSD, CHPX, and SOXQ, with varying sizes, fees, and liquidity.