AI ETFs Lead Sectors; Only 5 of 393 Thematics Beat S&P
BAI's 38% 3-month return headlines AI ETF dominance in 2026, but only 5 of 393 thematic ETFs are beating the S&P 500. Here's the data traders need.
Key Points
- BlackRock's BAI AI ETF posted a 38% three-month return to become the largest thematic ETF by AUM at $13.73 billion, but only 5 of 393 total thematic ETFs are outpacing the S&P 500 in 2026.
- The AI and semiconductor trade is structurally concentrated: the MSCI ACWI IMI Semiconductors index is up 109.30% through June 30, making SMH and SOXX the most-compared ETF pair of the year.
- Morningstar's warning about the 1,100-plus ETF launches in 2026 — nearly a third are leveraged or inverse trading tools — makes the next sharp market drawdown the critical stress test for potential mass fund closures.
BlackRock's actively managed AI ETF, BAI, put up a 38% return over the past three months, vaulted to $13.73 billion in AUM to claim the title of largest thematic ETF by assets, and made a compelling case that the AI infrastructure trade is the defining sector story of 2026. The catch: it is the exception, not the rule. Only 5 of the 393 thematic ETFs currently listed in U.S. markets are beating the S&P 500 this year, and the remaining 388 collectively manage $256.69 billion at an average expense ratio of 0.63% — a quiet tax on conviction trades that aren't working.
The AI Trade in Hard Numbers
BAI's mandate covers AI infrastructure providers, semiconductor companies, and software names, and its 38% three-month return reflects the violent repricing of that ecosystem over Q2. The S&P 500 has gained more than 10% through the first half of the year and the Nasdaq-100 is up over 20%, so BAI's outperformance margin is real but requires context: it is an actively managed product at a 0.55% expense ratio, meaning it charges 12 basis points more than a typical passive semiconductor index fund. In a 38% return environment that spread is trivial. In a mean-reversion quarter, it matters considerably.
The semiconductor benchmark validates the AI thesis in absolute terms. The MSCI ACWI IMI Semiconductors & Semiconductor Equipment ESG Screened Select Capped index is up 109.30% through June 30 — a number that would have seemed like forecast error at the start of the year. That performance has made SMH (VanEck Semiconductor ETF) versus SOXX (iShares Semiconductor ETF) the single most-compared ETF pair of 2026 among retail traders. The two funds track different indices, carry slightly different expense ratios, and have meaningfully different top-10 holdings weightings, which is why the comparison is commercially useful: investors who already decided to own semiconductors are now arbitrating between the two wrappers rather than questioning the theme itself. That level of second-order decision-making within a sector is a reliable sign of a trade that has moved from speculative to consensus.
The newest entrant with genuine structural differentiation is the Roundhill Memory ETF, DRAM, which launched in April and has accumulated $12.73 billion in inflows — a figure that rivals the AUM of established sector funds built over years. Tema ETFs' announced exclusive partnership with SemiAnalysis, an independent semiconductor and AI infrastructure research firm, points toward the next evolution: research-house-backed ETFs designed to give institutional-grade proprietary analysis a retail-accessible wrapper. If that model produces differentiated stock selection within the semiconductor space, it represents a credible competitive threat to both passive semi-index ETFs and generalist active AI funds like BAI.
Where the Sector Rotation Is Breaking Down
Rate-sensitive sectors tell the opposite story. Utilities, REITs, and financials entered 2026 with broad market expectations for multiple Fed rate cuts. Those expectations have eroded substantially by midyear. The Fed Funds Rate sits at 3.63% as of July 1, SOFR is at 3.66%, and CPI remains at 4.2% year-over-year — a combination that gives the Federal Reserve no credible justification for near-term easing. Prologis filed an 8-K on July 1, and with the 10-Year yield at 4.48%, REIT valuations remain compressed by the capitalization rate math that links directly to Treasury yields. Every 25-basis-point move in the 10-Year is a material headwind to REIT net asset values, and the current yield level is not supportive of sector recovery.
PAVE — the Global X U.S. Infrastructure Development ETF — occupies the middle ground, posting an 8.6% three-month return on $13.26 billion in AUM at a 0.47% expense ratio. Infrastructure is neither a pure growth play nor a pure rate-sensitive defensive — it draws its revenue from federal spending mandates that are largely insulated from monetary policy. That positioning explains why PAVE has attracted billions in inflows even as rate-sensitive sectors struggle: the spending pipeline is contractually committed and the revenue streams are regulated. Health care ETFs have also shown consistent inflow momentum, particularly in biotechnology, where structural demand from aging demographics and accelerating drug development timelines offers a thesis that does not depend on rate-cut timing.
The 5-out-of-393 thematic outperformance statistic is the most important number in the sector landscape right now. It is not an argument against thematic investing — BAI, DRAM, and the semiconductor complex are real stories with real earnings behind them. It is an argument against the reflexive assumption that a compelling narrative translates to a market-beating return, especially when the S&P 500 itself is running at a 10%-plus pace. The 393 funds collectively charge 0.63% on $256.69 billion, generating roughly $1.6 billion annually in management fees regardless of whether they outperform. The incentive to launch thematic products is structurally divorced from the incentive to outperform a benchmark.
The Structural Risk Building in New Launches
More than 1,100 ETFs launched in 2026 through the first half of the year, with active strategies representing over 80% of new launches and capturing 36% of total inflows — $313 billion — up from 31% in 2025. Dimensional Fund Advisors has filed to list five ETF share classes of its systematic fixed income strategies, and the broader ETF share class structure now gives investors access to leading discretionary active mutual fund strategies in a tax-efficient wrapper. That structural shift is durable and represents a legitimate long-term gain for retail investors who previously paid higher minimums for mutual fund access.
The risk is concentrated in the nearly one-third of 2026 launches classified as leveraged or inverse trading tools. Morningstar has flagged directly that a sharp market drawdown could trigger widespread closures among these products — a non-trivial risk given that the Nasdaq-100 is already up 20% and any mean-reversion event would hit leveraged long products with amplified force. More than three space-themed ETFs launched in Q1 2026 with six more in the pipeline ahead of a potential SpaceX IPO, and J.P. Morgan's standing observation applies directly: the ETF wrapper does not validate a theme, it simply makes it easier to express one. The funds that will survive a volatility event are the ones with assets above the $50 million minimum that most institutional platforms require for inclusion — a filter that eliminates the vast majority of 2026's new launches before they face their first real stress test. The date to mark is the September FOMC meeting: if the Fed signals no cuts through year-end, the second-order pressure on leveraged rate-sensitive ETFs and speculative thematic names will be immediate and measurable, and the 5-out-of-393 outperformance ratio could deteriorate further before any recovery.
The Weekly Investor
Daily market analysis for active traders. Free.
