The iShares Semiconductor ETF (SOXX) gained 79% year-to-date and 152% over one year as of June 5, 2026, as AI infrastructure spending reshapes chip demand across three continents.1 Leveraged 2x semiconductor funds posted approximately 196% over the same period.2
Nvidia's Vera Rubin architecture is now shipping to hyperscalers worldwide. But the cycle is moving past U.S. and European cloud facilities.
Taiwan's Phison Electronics is pushing AI computation into consumer hardware globally. Its aiDAPTIV technology, developed with Intel, expands available memory on AI PC platforms—enabling larger models and agentic applications locally without cloud infrastructure.3,4 "AI PCs are evolving into platforms for more sophisticated local AI workloads," said KS Pua, Phison's CEO.3
At the design layer, capital requirements are narrowing the competitive field globally. ASIC development costs have risen nearly an order of magnitude since FinFET adoption. An IEEE Spectrum engineer described the academic baseline: teams receive 40 chips from a TSMC prototyping service and declare success after five to ten functional units.5 Industry targets failures measured in parts per million.5 That gap defines the moat incumbents now hold—concentrated overwhelmingly in the U.S., Taiwan, and South Korea.
Geopolitics is restructuring supply chains in parallel. U.S. rare earth export restrictions are accelerating decoupling from Chinese inputs—a pressure felt across Asian manufacturing networks. Chinese challengers—including the Zhenwu V900 and J900—are targeting the high-end AI chip market, but sourcing constraints limit their development pace.
The supercycle's reach continues to expand beyond semiconductors proper. GlobalFoundries launched a Quantum Technology Solutions initiative, citing photonics work with PsiQuantum as proof U.S. manufacturing can serve emerging compute demand. Satellite compute is emerging as an additional demand vector alongside edge AI—relevant to markets where terrestrial infrastructure remains thin.
For global investors, the central question is duration, not direction. Fewer companies worldwide can compete for the largest contracts, amplifying both upside and drawdown exposure. The 79% YTD gain in SOXX reflects repriced expectations as much as realized demand.1 The next leg depends on hyperscaler capex holding and edge AI volumes following the trajectory Phison and Intel are building toward.3,4
