Over the past two years, semiconductor stocks tracked by the iShares Semiconductor ETF (SOXX) have dramatically outpaced software names in the iShares Expanded Tech-Software Sector ETF (IGV). At the same time, their rolling 2-year correlation has flipped to roughly -0.25, an unusually negative reading for two technology subsectors tied to the same demand cycle.
Historically, semiconductor and software benchmarks have shown a consistently positive, relatively high correlation, particularly around major macro and tech cycles such as the late-1990s dot-com period, the 2007-09 crisis, and the 2013-19 cloud and mobile expansion. In those regimes, both groups typically traded as part of a single high-growth technology complex, rising and falling together with shifts in liquidity, risk appetite and enterprise IT spending.
The current decoupling is rooted in AI enthusiasm that has heavily favored hardware, with SOXX dominated by AI-centric names like NVIDIA (NVDA), Advanced Micro Devices (AMD) and Broadcom (AVGO). Market pricing embeds aggressive expectations for AI data-center and accelerator demand in these chipmakers, while software-heavy IGV constituents such as Microsoft (MSFT) have not enjoyed the same relative re-rating despite also having substantial AI exposure.
With semiconductors and software still economically linked through the same AI infrastructure and application stack, a -0.25 two-year correlation looks more like distortion than a stable new equilibrium. If past co-movement patterns reassert, normalization can arrive through semiconductor underperformance as AI expectations cool, software outperformance as AI monetization shifts toward applications, or some combination of both, with highly AI-levered chip leaders most exposed to a reversal of the current gap.
Terminology
- 01Rolling 2-year correlation: Correlation between two assets measured over a shifting 24-month window.
- 02Mean reversion: Tendency for extreme relative performance or valuations to move back toward averages.