The private-markets pivot is an AI concentration trade

Morningstar's fifth annual Asset Owner Perspectives Survey arrived this week with the number the private-markets industry will quote for a year: asset owners holding more than $20 trillion between them expect private-market exposure to rise from 19 percent of assets under management to 23 percent within five years. The same survey of more than 500 owners across 11 countries ranks AI valuations and capital expenditures as its top macro-market concern, named by 73 percent of respondents—two findings that tend to get filed under separate headings but form a single position.

The demand side drew a second confirmation from Dynamo Software's fifth annual limited partner survey, where 63 percent of LPs plan to raise allocations to alternatives over the coming year, the strongest reading in that survey's history. Intent is not committed capital and a plan is not a subscription, so the LP figure is a leading indicator rather than a flow, but the direction is consistent across both samples and the five-year institutional target is the number gateway builders will sell against for the next several quarters.

The interest runs to the risk ledger. Generative AI ranks second among material global investment issues at 71 percent of asset owners, behind only rising inflation at 76 percent, while energy and supply-chain disruption, third at 66 percent, is a figure Morningstar ties to AI's electricity demand. Nearly six in ten owners said AI-driven energy demand could push up both energy costs and broader inflation, and the share flagging AI-related environmental considerations as material more than doubled year over year from 12 percent to 25 percent.

On macro-market risk, AI-related concerns hold the entire leaderboard: valuations and capex at 73 percent, AI-driven market concentration at 65 percent, overdependence on a handful of mega-cap technology providers at 63 percent. Lindsey Stewart, Morningstar's director of institutional insights, frames the exercise as weighing AI's opportunity case against valuations, concentration by sector and by geography, and governance questions at the largest AI companies—the bind behind the allocation shift.

The two halves of the survey connect through benchmark arithmetic every investment committee knows by heart.

A public sleeve that tracks a cap-weighted index cannot cut its mega-cap technology weight without accepting tracking error, and the 63 percent who flag overdependence on a handful of providers have described a constraint rather than a preference. Private markets supply the release valve—quarterly marks, no index weight to defend, and a duration profile that suits a five-year target—and the surveys say owners are opening it.

Whether the trade does what the committee believes depends on a question neither survey asks, let alone answers: how much of the private sleeve is itself AI exposure. Some share of the build-out—the power, the data centers, the credit that funds the capital expenditure—likely sits in the same vehicles the 19-to-23 percent move flows into. That is inference rather than measurement, and it is the inference worth testing. If the overlap is material, the higher target does not hedge the concentration; it relocates it behind a lock-up, where no daily price forces anyone to mark it.

There is a governance test that does not depend on the answer: before approving a larger private allocation, can staff produce one AI-exposure figure that spans public and private holdings? Where they cannot, the target is a hedge the institution is unable to document, and the 73 percent naming AI valuations as their top macro worry are describing a risk their own reporting is not built to size.

The right sequence runs the other way and is cheap to adopt: the consolidated AI line belongs in the policy statement first, with the private-markets target set underneath it as a response to what that line shows. Approving 23 percent as a diversification goal and meeting the overlap later costs almost nothing to avoid at the agenda stage and a great deal to unwind once the commitments are called.

None of that argues against private markets, and the appetite behind these targets is strong enough that the firms assembling the private-markets gateway now have a demand case assembled at institutional scale: the gateway itself has become an acquisition target and the blank offering line is the tell, with capital pre-sold before the Form D. A five-year move from 19 to 23 percent of institutional assets, joined to 63 percent of LPs intending to add, is the runway those platforms, custodians, and fund managers are underwriting right now.

The caution about evergreen secondaries cuts the same way: the dilution this publication flagged in evergreen secondaries is what happens when appetite outruns the supply of discounted assets, and a record reading in LP intentions forecasts more of that pressure, not less.

AI has moved into the advisor-client meeting, and the platform contest over that conversation is real. The allocator survey describes a different AI problem, one of measurement dressed as portfolio construction: tooling pitched at allocator workflow answers a productivity question this sample did not rank near the top, while the position the committee is genuinely short is the reporting layer that puts public and private exposure on one line. Whoever builds that connector gets a seat in the room where the 23 percent is approved.

Watch next year's installments for a single figure: a consolidated AI exposure number, public and private on one line. Until a committee can produce one, the move from 19 percent to 23 percent is a position an institution has taken without being able to write in its own investment minutes what it is a bet on.