After Months of Watching AI Innovation ETFs Run These 3 Active Funds Stand Out and the Revolution Has Barely Started

Goldman Sachs sees AI capital spending rising to $765 billion annually in 2026 and $1.6 trillion by 2031, underpinning a bullish case for AI infrastructure exposure. Among the three funds covered, BAI has led with about 79% trailing-year gains and 35% YTD, ARTY is up about 32% YTD, and IVES trails at 13% YTD. The article favors active and thematic approaches to capture the next wave of AI beneficiaries beyond the mega-cap leaders.

Analysis

The market is still pricing AI as a single trade, but the better expression now is a barbell between compute incumbency and supply-chain disintermediation. The strongest second-order beneficiaries are not just the model owners; they are the names that monetize every incremental watt, wafer, and rack deployment without bearing end-demand risk. That favors the hardware and infrastructure layer first, then software/data plumbing only after capex proves durable enough to absorb another 12-18 months of investment.

BAI and ARTY are effectively different ways to own that same spillover: BAI is the higher-conviction, more index-independent version, while ARTY is the broader “follow the buildout” expression. The key nuance is that the winners of the next phase are likely to be the firms with pricing power into constrained inputs, not merely the largest AI logos. That supports continued relative strength in NVDA/AVGO/CLS/FN/MRVL/TSEM, but it also raises the odds of a rotation into power and data-center adjacency names like GEV if investors start questioning how much of the capex wave is economically self-funding.

The main risk is not that AI spending stops tomorrow; it is that the market compresses multiples before revenue catches up. Over the next 1-3 months, any deceleration in hyperscaler capex guides or a single large customer pausing orders could punish the higher-beta supply-chain names more than the mega-caps. That is why IVES is the most fragile vehicle here: its smaller, more idiosyncratic holdings should outperform only if the market broadens beyond the obvious leaders and re-rates the newer infrastructure proxies.

Consensus appears too comfortable with “AI breadth” being automatically bullish. In practice, breadth is a double-edged sword: it validates the theme, but it also dilutes scarcity and increases the odds that capital shifts from scarce chips to crowded enablers with weaker moats. The better contrarian setup is to own the few names with genuine bottlenecks and use laggards in the higher-beta thematic basket as funding sources.

Key Decisions for Investors

• Long NVDA/AVGO vs short a basket of higher-beta AI enablers (PLTR/INOD/NBIS) over the next 4-8 weeks: thesis is that the market rewards monetized bottlenecks over narrative names if capex remains the headline.
• Buy CLS and FN on pullbacks, hold 3-6 months: these are cleaner second-order beneficiaries of AI infrastructure spend with less multiple risk than pure software exposure; risk/reward improves if hyperscaler capex guides stay firm.
• Pair long GEV vs short a basket of AI software proxies (SNOW/PLTR) into the next earnings cycle: if the market starts pricing the power constraint more explicitly, grid and generation names should rerate before software does.
• For tactical exposure, prefer BAI over IVES at current levels: BAI gives more direct participation in the dominant leaders with lower analyst-keyman risk; IVES is only attractive if you specifically want higher beta and are prepared for larger drawdowns.
• If you want thematic upside with less concentration, use ARTY as the core and size a smaller direct NVDA/AVGO overlay: this captures the broadening of AI spend while keeping the portfolio anchored to the names most likely to defend margins.