The iShares Semiconductor ETF (SOXX) is up 79% year-to-date and roughly 152% over one year, as of June 5, 2026.1 The rally tracks a surge in AI compute and memory demand across the chip supply chain.
AI PCs are pushing that demand further. Agentic applications and larger mixture-of-experts models require more memory capacity and faster responsiveness than earlier AI PC workloads.2 Phison's aiDAPTIV technology, built with Intel, expands the memory available to AI PC platforms so OEMs, developers and end users can run heavier AI applications locally while keeping data on-device.2 "Through our collaboration with Intel, aiDAPTIV helps expand the necessary memory available to AI workloads on Intel AI PC platforms," said KS Pua.2
Power semiconductor supplier Vicor Corporation illustrates the financial upside of the AI hardware buildout. The company reported trailing twelve-month revenue of $471.7 million and net income of $136.7 million, with a market capitalization of $11.3 billion as of the July 6, 2026 market close.3
The compute and memory race is also raising the cost of entry. Commercial chipmakers hold themselves to failure rates measured in parts per million, documenting even rare anomalies to trace root causes — a rigor far beyond academic prototyping runs, where five to ten working chips out of a 40-chip batch is enough to call a design proven.4 That gap in tolerance for failure reflects the capital and engineering investment now flowing into memory and compute hardware built for AI workloads, and it is a factor keeping new entrants out of leading-edge design.
Together, the memory expansion push and the equity rally point to the same bottleneck: AI systems are outgrowing the compute and memory headroom of existing hardware, and chipmakers are being rewarded for closing that gap.