Model portfolios are driving the RIA tech-ETF rotation

The average RIA portfolio ended the second quarter holding 92.9 unique ETFs, up from 88.4 in the first quarter, and the natural read is that advisors are broadening their books. The AdvizorPro data behind that number tells a narrower story of a concentrated rotation into semiconductor and AI funds, with a separate ISS Market Intelligence analysis suggesting a large share of those allocation changes originated upstream.

AdvizorPro's Q2 2026 RIA ETF Trends Report tracked 5,398 RIAs that appeared in both the first and second quarters, and technology-focused funds drew 230 net RIA allocators, more than any other segment. The iShares Semiconductor ETF and the VanEck Semiconductor ETF each gained more than 200 net RIA relationships, while AI tickers from Global X and Defiance posted double-digit percentage gains. The money came out of the real-asset trade that defined the start of the year, as commodities-focused funds shed 103 RIAs and digital assets lost 86.

The rotation was broad but not evenly distributed: 63.4% of tracked firms added ETFs while 18.2% trimmed their lineups, and turnover held at 11.6% of holdings as adds outpaced drops by 1.6 to 1, producing a net gain of nearly 25,000 ETF positions across the tracked universe. Every one of the ten largest ETF issuers by RIA penetration gained advisors, with Invesco leading on 95 net new relationships, followed by Schwab ETFs, First Trust, and VanEck.

The sharpest percentage gains came from small active shops: EntrepreneurShares grew its RIA base 115.6%, largely on the strength of its XOVR crossover fund, while Baron Capital and ProcureAM each posted gains above 70%. Those boutique numbers point to a pocket of advisors still making specific product bets.

Only 41 tickers appeared in RIA portfolios for the first time in the second quarter, down from 140 in the first quarter, and advisors grew more selective, or, as the ISS data suggests, the selection increasingly happened before the product ever reached them.

ISS Market Intelligence estimates, based on 13F holdings data, that at least $1.8 trillion, or 35%, of the $4.3 trillion in ETF assets held in discretionary brokerage and advisory accounts as of the first quarter exhibited what it calls "model-like characteristics," holdings that resemble a structured model portfolio even when the account is not formally labeled one. The estimate does not say a model made every trade; it says 35% of the asset base sits in portfolios that look centrally built, which is the pattern a rotation into two semiconductor ETFs and a handful of AI funds would leave.

That should give an RIA principal pause, because the 92.9-ETF average looks like diversification until the components suggest standardization: book a model, take the semis and AI sleeve, watch the lineup drift upward. The collapse in new-launch adoption supports the same reading; if advisors were making the calls themselves, the products being added would be more heterogeneous, but the additions concentrate in the few tickers that model portfolios already favor.

Invesco led the issuer table with 95 net new RIA relationships, and it arrives with a platform overhaul behind it: the firm unified custom strategies across public and private markets in August, as this publication reported. The issuer that controls the model, or the custom strategy inside it, controls the flow, and Invesco is positioning for that with the data suggesting the positioning is working.

The real-asset reversal is the tell: commodities and digital assets led the first-quarter surge, and the second quarter erased it. That is a fast pivot for a channel that prides itself on long-term discipline, and it is more consistent with model rebalancing than with 5,398 firms independently changing their minds in the same quarter. RIAs still control discretionary books, but the 35% model-like share suggests a growing portion of the flow has already been delegated.

The next quarter's data will show whether the tech trade holds or reverts like the real-asset trade did, but the share of assets that look like models matters more than the 92.9 average. That number is likely to keep climbing, because the economics push that way: models are cheaper to run, easier to risk-manage, and they move the advisor's job toward client relationships and away from product selection. Advisors who want to keep the allocation call will need to be explicit about it. The market has already started moving it upstream.