For first time, more central banks are set to shrink dollar holdings, survey finds

An OMFIF survey says 79% of central banks plan to cut dollar allocations in the coming decade, citing rising political risks around the U.S. currency, even as the dollar has rallied ~3% year-to-date on higher U.S. rates and a flight to safety. The same survey finds gold is gaining central-stage reserve use, with record prices and 82% of central banks holding gold and a net 30% intending to increase allocations over the next 1–2 years. It also reports strong momentum toward AI in reserves-management, with 66% of central banks planning near-term AI integration, alongside a growing preference for emerging markets (38% of public funds up from 27% last year).

Analysis

The important mechanism is not an imminent dollar break, but a slow repricing of the marginal reserve buyer. If central banks and sovereign funds keep trimming USD reserve growth, the first-order winners are gold and non-U.S. real assets; the second-order winner is any asset whose cost of capital is set globally rather than domestically. That argues for a bid in GLD/IAU and, more selectively, EM local-currency debt and infrastructure/real-asset vehicles, while the incremental headwind for U.S. duration is softer than for the dollar itself.

The AI angle is more actionable than it looks: public institutions are not trying to “alpha” AI, they are trying to cut operating friction in back-office and risk work. That favors incumbents with sticky enterprise distribution and secure cloud/data plumbing more than pure-play model names, so MSFT, ORCL, and cybersecurity vendors can capture budget share without needing a dramatic spend cycle. The flow is still small relative to hyperscaler capex, so this is a multi-quarter adoption story, not a one-week catalyst.

The contrarian point is that reserve diversification surveys usually overstate actual transaction speed. Higher U.S. real yields and a functioning Treasury market still anchor the dollar, so any FX move is likely to be gradual unless policy uncertainty or geopolitical stress worsens again. For the names provided, there is no direct single-name edge; DLTR is the only one with a mild imported-goods FX sensitivity, but the pass-through would be slow and likely drowned out by consumer demand and margin execution.

Key Decisions for Investors

• Long GLD / short UUP as a 3-6 month macro pair: best expression of gradual reserve diversification; stop if DXY reclaims prior highs or real yields resume moving materially higher.
• Add a small overweight to EEM versus EFA over 1-3 months: if reserve managers shift marginally away from USD, EM asset flows should improve faster than developed ex-U.S. because positioning is lighter and valuation support is better.
• Buy MSFT or ORCL on pullbacks into the next quarter as a low-beta AI adoption beneficiary: central-bank AI spend is not explosive, but it supports recurring workflow/software demand; risk is valuation compression if cloud growth slows.
• Watch GLD breakout levels and central-bank buying data as the falsifier: if reported reserve diversification does not translate into higher gold official-sector purchases over the next 1-2 quarters, fade the move.
• Avoid making a directional bet on CBSU, FISI, or WWRL from this note; the article is too macro and flow-driven, and any equity impact is likely second-order and weak.