AI adoption is collapsing fund lineup diversification

Of the 131 asset managers Mercer surveyed for its 2026 report, 55% have already wired AI into at least one investment process and 91% expect to lean on it harder over the next year; the tools they are reaching for are largely the same ones. 63% run off-the-shelf AI products, 58% buy vendor-supplied data, and a former top-quartile CIO, writing in WealthManagement.com on September 11, reads that as convergence rather than productivity — the survey numbers are the part of his case that does not depend on his opinion.

His argument starts upstream of the models: investment professionals arrive from a narrow group of undergraduate institutions, pass through the same MBA programs, and then through a single CFA curriculum that teaches thousands of them to value securities and frame risk in the same terms. Shared standards build competence, but a system designed to establish common foundations eventually produces common conclusions; layer AI on top — overlapping models asked overlapping questions about overlapping datasets — and a manager's distinct philosophy survives mostly in the marketing deck.

For advisors the fallout shows up as overlap, not style drift: a U.S. large-cap sleeve, a quality sleeve and a global sleeve assembled from the same inputs can leave a client holding one bet under five tickers. That is why a holdings-level overlap report and a look at factor concentration belong in due diligence rather than in a pitch about how seriously the firm takes manager selection.

If every manager rents the same off-the-shelf tools and feeds them the same data, the capacity gain nets out in differentiation; free valuation tools already reset the floor for advisors planning a move, and rented intelligence does the same to fund selection. As this publication has argued, tax planning is the last high-margin skill in commoditized portfolios; a lineup whose managers have quietly converged makes construction, asset location and harvesting the parts of the job a competitor cannot copy by licensing the same model.

The test is arithmetic. Before the next rebalance, run a name-level overlap check across the model lineup and a factor read on the top ten positions of the growth, quality and core sleeves; if those names repeat, the diversification the client is paying for is thinner than the sleeve labels suggest, and no amount of manager commentary will tell you so.