TO: Everyone who bought gold as an insurance policy
FROM: The market
DATE: March 25, 2026
RE: A clarification on what you actually owned
Please be advised that the commodity you purchased as a geopolitical hedge against war, inflation, and systemic breakdown has, during an active war that is causing inflation and systemic breakdown, declined approximately 18% from its January high.
We understand this is confusing. You were told gold goes up when things go wrong. Things have gone wrong. Gold has gone the other direction.
We appreciate your patience.
Let's reconstruct what actually happened, because the financial press is going to spend the next three weeks writing about "safe haven rotation" and "technical levels" and all of it will miss the point.
Gold peaked at $5,595 on January 29th. It was, at that moment, the consensus trade — central bank demand, de-dollarization, Fed pivot, AI-driven CapEx inflation, geopolitical fragmentation, every macro thesis that had been true for three years was still true and everyone from your barber to Deutsche Bank was long. Deutsche Bank had a $6,000 target. J.P. Morgan had $6,300. The metal had returned 65% in 2025, its best year since 1979.
Then Iran closed the Strait of Hormuz, and Brent crude surged past $112 a barrel, and gold — gold, the asset explicitly designed for this scenario — sold off 18% in six weeks.
As of Tuesday, XAU/USD is trading around $4,565. The all-time high is $1,030 per ounce lower, sitting up there like a reproach.
Here's the mechanism, because the mechanism is the whole story. When oil surges, real inflation expectations surge with it. When inflation expectations surge, the Federal Reserve — which had been expected to cut three times this year and is now being priced for a possible hike — loses whatever latitude it had to ease. When the Fed loses latitude to ease, real yields stay elevated. When real yields stay elevated, gold, a non-yielding asset, becomes expensive to hold relative to alternatives. So the oil shock that should have triggered a flight to gold instead transmitted, through the yield channel, into a reason to sell it.
This is not a new dynamic. It is a very old one that kept getting papered over by the 2020–2025 expansion in central bank balance sheets and a rate environment that made gold's zero yield irrelevant by comparison. That paper is no longer covering anything.
The flash PMI data released Tuesday underlined the bind everyone is now in.
The S&P Global US Composite PMI fell to 51.4 in March, down from 51.9 in February — the lowest reading since April last year, and marking consecutive months of slowing growth. The services sector dragged, business confidence deteriorated, and — here is the detail that matters — input costs surged at the fastest pace since August 2022. Selling prices followed. This is the statistical fingerprint of an energy price shock bleeding into the general price level. It's happening fast. It happened last time too, in 2022, and the Fed waited eight months to acknowledge it and spent the next two years cleaning up the resulting mess.
Jerome Powell is not waiting this time. The dot plot still signals one cut in 2026. Markets are now pricing the October meeting at a 45% probability of a hike. February's PPI came in at +0.7% month-on-month — more than double the estimate, the largest monthly increase in over two years. The PCE print, which would ordinarily land March 27th, has been rescheduled to April 9th. The irony: at the exact moment the market most needs an inflation anchor, the BEA moved the goalposts forward by two weeks.
So we are navigating a war, an oil shock, a private credit dislocation, an equity market that has closed below its 200-day moving average three sessions running, and a Fed that cannot cut — and our primary inflation hedge has lost nearly a fifth of its value.
The bull case for gold, to be fair, is not dead. Central banks bought 230 tonnes of the stuff in Q4 2025 and have no structural reason to stop. Global gold ETFs posted $18.7 billion in inflows in January alone, a record month. The physical market — actual metal, actual buyers — held up through the futures flush. Physical premiums stayed elevated even as paper gold cratered. There's a difference between a crowded long getting liquidated in the futures market and a fundamental re-rating of the asset. The former is what happened. The latter hasn't happened yet.
But here is the question nobody is asking clearly enough: if gold can't hold $5,000 while crude is approaching $120, the Strait of Hormuz is closed, inflation is re-accelerating, and the Fed is paralysed — when, exactly, is it supposed to work?
The answer, uncomfortable as it is, may be: only when rates fall. Which makes gold less an insurance policy against chaos and more a leveraged bet on monetary policy direction. Insurance that only pays out when the firemen cooperate is a product with a significant asterisk.
The VIX closed above 26 Tuesday. The S&P 500 gave back Monday's 2% relief rally in a single session, crude rebounded 4% on the open after Iran denied that ceasefire talks were happening at all, and Broadcom fell 0.5% in premarket on supply chain constraints at TSMC. The AI infrastructure trade — which had been the single most reliable equity thesis of the past three years — is now running into physical capacity limits at the foundry level. Demand for Blackwell chips is outstripping what TSMC can manufacture. The bottleneck is no longer compute demand; it's wafer starts.
Meanwhile Estee Lauder is in merger talks with Puig at a $40 billion valuation and the stock dropped 7% on Monday because apparently even a luxury consumer staples company doing M&A can't get a bid in this tape.
The great trades of the 2023–2025 cycle — long gold, long AI capex, long private credit yield, short vol — are all simultaneously under stress. Not because any of them were wrong. Because an oil shock is the one exogenous force capable of disabling all of them at once. Higher energy costs hit middle-market borrowers in credit portfolios. Higher inflation kills the case for rate cuts that underpinned gold. Higher operational costs compress the margins of the hyperscalers that justified AI multiples. And higher uncertainty spikes vol.
One event. Four unwinds.
The Strait of Hormuz may reopen tomorrow. Iran may re-enter negotiations. Crude may give back $20 in a week and all of this resolves into a footnote about a Q1 wobble. That is genuinely possible. Markets are right to trade some probability of deescalation.
But the architecture has been exposed. The portfolio construction that felt robust in January relied, in ways that weren't fully articulated, on a stable energy supply and a Fed that had room to move. Both of those assumptions are currently suspended.
Your hedge didn't hedge. But more precisely: it hedged against the wrong version of this crisis.
