Sparkline Analytics: A New Lens

Sparkline argues that investors should diversify away from crowded AI infrastructure by owning companies that adopt AI rather than companies that build the infrastructure, while modernizing value portfolios to capture intangible assets and innovation.

• The firm’s platform combines standard fund data with proprietary measures of intangible value, AI adoption, and trade exposure across more than 10,000 ETFs and mutual funds grouped into 127 peer categories.
• The iShares Semiconductor ETF rose 240% in five quarters, while hedge-fund leverage and crowding in AI infrastructure reached record levels before the trade unwound in July.
• AI thematic funds returned 53% over the past year and showed 39% exposure to semiconductors and semicap equipment, 17% to the Magnificent 7, and 69% to the broader AI infrastructure complex.
• AI infrastructure exposure reached 48% in the S&P 500, 68% in the Nasdaq 100, and 69% in AI thematic funds; the S&P 500’s exposure had nearly doubled from 25% in 2019.
• Traditional alternatives such as value, equal-weighted, developed international, and small-cap indexes reduce infrastructure exposure but hold between 41% and 83% in AI laggards, while emerging markets have 53% infrastructure exposure.
• Screening for funds with more than 33% in AI adopters and less than 33% in both infrastructure and laggards produced 111 funds with only 7% semiconductor exposure and 10% capex intensity.

Takeaway: Sparkline prefers the adopter-heavy side of the AI frontier because it offers participation in AI-driven earnings without the capital-cycle risk of infrastructure builders, while traditional value portfolios may need to add innovative, asset-light businesses.