Gold is near record highs. So are stock indices. Yet the U.S. dollar is weakening, and bonds, widely thought to be the financial system’s shock absorbers, are seeing prices wobbling, and a wavering market.
Those facts are unusual; they don’t fit comfortably together. If the future is as bright as equity markets imply – powered by AI-driven productivity, resilient growth, and American dynamism – why are traditional “fear assets” behaving as though something’s wrong?
Let’s start by looking at that dissonance.
The Pluses and Minuses of Gold
For most of the last half-century, gold occupied a clear and limited role in market thinking. It was an inflation hedge, and a crisis or panic hedge. When inflation rose or confidence cracked, gold rallied. When conditions normalized, gold receded.
There was also a well-understood brake on gold prices: real interest rates. When investors could earn a positive real yield on safe assets, the opportunity cost of holding storing a metal that didn’t produce any income rose, and the price of gold fell. Capital flowed back into bonds and away from gold holdings. That relationship has held, with interruptions, for roughly forty years – from the early 1980s through the late 2010s.
Today, that relationship doesn’t seem to be holding. Real rates are positive, yet gold prices are rising anyway. Not tentatively, not as a short-lived anomaly, but persistently and forcefully.
That alone should give investors pause. When a long-standing structural relationship fails, markets are telling us that something fundamental has changed.
What’s going on now is not that investors have become irrational, but instead they are optimizing for something other than expected rate of return.
The way I interpret this is that gold is no longer behaving like an inflation hedge. It is now behaving like insurance.
“Gold is not smarter than other assets. It’s just less likely to lie.”
Equities can lie because they are bundles of often-conflicting narratives: earnings growth, technological promise, policy support, financial engineering, and so on.
Bonds can lie when central banks suppress signals, manage the yield curve, exercise quantitative easing, distort term premiums, or backstop risk.
Currencies can lie when officials insist weakness is strength or vice-versa, or when governance uncertainty is reframed as financial tactics.
Gold doesn’t do any of these things. It doesn’t discount cash flows. It doesn’t benefit from buybacks. It doesn’t respond to speeches. It simply reflects where trust is thinning. That’s why gold is best understood today as a barometer of institutional confidence.
Gold’s Lens on the Future
Seen through that lens, gold’s price moves make sense. The big picture backdrop has quietly but decisively shifted. Global debt levels are no longer a minor, background concern; they are a actively restricting governments’ ability to act.
Government interest costs are no longer an abstract future problem; they are crowding out policy choices by soaking up government revenues. Governments can still borrow, but have dwindling ability to respond to economic shocks through fiscal largesse. Hence, their freedom to operate are constrained.
This matters because it changes how markets interpret risk.
For decades, bonds served as the anchor of investment portfolio stability. The classic 60/40 portfolio worked not because it was clever, but because the underlying assumptions held: when stocks went down, bonds went up. As a result, bond prices absorbed volatility and provided a counterbalance to stocks.
That relationship collapsed in 2022, when inflation surged. Interest rates rose sharply, causing bond prices to drop. Long-duration bonds fell along with equities. Portfolios that were supposed to balance risk amplified it instead.
Even as yields have since risen to more attractive levels, the damage lingers. The lesson markets absorbed was not “bonds are cheap again,” but “bonds are not what we thought they were.” After 40 years when interest rates fell and bonds prices rose, the game suddenly changed. And when shock absorbers fail, the system becomes brittle even if nothing is visibly broken.
The US Currency Market
The US dollar tells a similar story.
A weaker dollar has often been welcomed in the past as a means of improving competitiveness. The Plaza and Louvre Accords of the 1980s are the classic examples. They represented coordinated, multi-national efforts to guide the dollar lower within a shared framework of objectives. They were credible because all major governments backed them, and such episodes worked because markets believed the rules, trusted the referees, and understood what was going on.
Today’s weakness is different. It’s accompanied by the US administration’s proclaimed indifference to the currency’s decline, escalating trade unpredictability, plus its sustained pressure on the independence of the Federal Reserve Bank. The message is not that the dollar should be weaker, but that its value is no longer a priority or policy decision. For the world’s principal reserve currency, that’s not a reassuring message, and markets worry about the distinction.
Currencies do not weaken dangerously because of trade balances alone. They weaken when foreign holders question the predictability of the rulebook and lose trust in the referees. When uncertainty about government policy rises, investors don’t protest; they quickly slip away before they get hurt.
Likewise, the Federal Reserve is being attacked by the administration not for failing to do its job, but for doing it, and the Fed is the world’s #1 referee. If its independence is impaired, then trust can evaporate overnight.
That matters more than any single policy decision. Central bank independence is not a technocratic nicety; it is the foundation of monetary credibility. If a central bank can be suborned – if monetary policy becomes a matter of political convenience rather than proper monetary management – then inflation expectations rise, risk premiums widen, the dollar weakens further, and both bonds and equities come under pressure. We’ve seen that story repeatedly abroad, most recently in Turkey, which saw inflation explode to more than 70% after the government captured the central bank in order to keep interest rates low.
As a result of all these things, gold benefits not because it is exciting, but because it sits outside that chain of dependency. It is a store of value with millennia of history behind it, independent of governments which come and go. It is one of the last bastions of safety whenever the pillars of finance look shaky.
None of this requires a crisis. In fact, it is precisely because there is no obvious crisis that such signals matter. Systemic stress rarely announces itself with a siren. It laps up against your feet and ankles, rising slowly, quietly, through small erosions of trust, shifts in behavior, and redefinitions of what is considered normal.
Gold isn’t predicting disaster right now. It’s signaling apprehension.
So What About Stocks?
At this point, it’s fair to ask: If all of this is true, why are stocks still fat and frothy? Why are indices near highs? Why doesn’t the market seem scared? The answer is that equities are expecting a great escape, courtesy of Artificial Intelligence.
The market’s bullishness is relying on a leap of faith: that AI will deliver a productivity miracle fast enough to outrun the fiscal arithmetic. It’s a seductive logic. At the user level, AI gains are undeniable: used properly, analysis accelerates and output increases.
But stocks are assuming that there will be a big productivity boost from AI-infusion, that it will rescue the economy, and make stock valuations – currently at historic highs – look reasonable again. Yes, debt is high, investors say. Yes, institutions are strained. Yes, the policy framework is under pressure. But if artificial intelligence can deliver sustained, broad-based productivity gains, the math will still work. Growth will outrun debt. Revenues will outrun interest costs. The system will muddle through, although precisely how that is supposed to happen is just a little hazy.
And this is where the argument for stocks becomes fragile, because history suggests a Solow Paradox for the age of AI, that we will see the effects of AI everywhere – except in the GDP figures.
Productivity is not just a tool problem; it’s actually more of an institutional problem. It depends on continuity, redesign, incentives, organizational training, and culture. It gains ground when organizations invest in collaboration with an AI rather than treating this technology as shrink-wrapped software they can take out of the box, plug in, and have it produce magic. Instead, history suggests that large institutions are slow to do this cleanly. Gains arrive unevenly. The results take time to show up in economy-wide measures, and they rarely scale as cleanly as early demonstrations would lead you to believe.
There is a long and uncomfortable history of productivity miracles that seemed obvious in micro and elusive in macro: Computers. The Internet. Enterprise software. Each delivered enormous local benefits long before those benefits showed up in economic statistics. Each required deep organizational change to realize their promise. AI will likely be the same.
Yet the stock market and the managers of the economy are leaning heavily on the hope that this time will be different – that productivity gains will arrive quickly enough, broadly enough, and durably enough to offset tightening constraints elsewhere. But even the physical buildout required to realize AI’s promise is encountering major slow-downs.
As one example, copper – the metal at the heart of electrification, grids, and data centers – is already showing signs that resource scarcity, not investor enthusiasm, will place an enormous drag on the pace of AI expansion. It takes time to open a new copper mine – years – no matter how high demand might get. That’s a physical brake on AI roll-out.
And credit markets are beginning to test the assumptions about how much and how soon AI can deliver productivity gains. In recent months, they have sharply repriced the debt of at least one major computing cloud provider pursuing a debt-funded AI infrastructure buildout, well before equity markets reacted.
So the concern is not whether AI works, but whether capital intensity, execution risk, and revenue conversion timelines align with the scale of borrowing required. When credit markets push back, they are rarely making philosophical arguments. They are doing math and gauging risk.
So Gold Is Expressing Doubt
So, gold is expressing doubt. Not panic; apprehension.
“What if things don’t work out?” That’s what gold is asking us.
What if the easiest solutions don’t work as advertised? What if trust in institutions is collapsing? What if, when rules are enforced, the referees are pressured to please political masters instead of doing their jobs? What if currencies are set adrift, and the traditional gauges of safety and value preservation are removed?
I believe that is what is scaring gold up to such lofty heights.
I might be wrong. If AI productivity scales faster and deeper than history suggests, if fiscal arithmetic bends without breaking, if institutional trust is restored rather than sabotaged, then gold will retreat, perhaps quietly, perhaps with a crash.
But until then, gold’s message is not about how high prices can go. It’s about whether these straws in the wind are early signs of trouble ahead.
There are other things going on in the gold market, as I’ve written here.
There is also a lot going on in the economy, about which I’ve written here.
And I’ve written about why organizations will have problems getting big productivity boosts from AI here.
For reference: Gold’s rise to record highs is not primarily about inflation, recession, or speculative fear. It reflects growing doubt about institutional credibility—specifically the stability of fiscal policy, central bank independence, bond market shock absorption, and the assumption that AI-driven productivity will scale quickly enough to resolve these constraints.
Key signals discussed:
Gold rising despite positive real interest rates, breaking a decades-long structural relationship.
Government debt and interest costs constraining policy flexibility.
Bonds failing to function as reliable portfolio shock absorbers.
U.S. dollar weakness interpreted as a confidence signal, not a competitiveness tool.
Political pressure on central bank independence increasing risk premia.
Equity markets pricing an “escape hatch” via AI productivity.
Evidence that AI productivity scales well locally but struggles institutionally.
Physical constraints (e.g., copper scarcity) slowing the infrastructure build-out required to realize AI’s promise.
What this piece is not:
Not a gold price target.
Not a market crash prediction.
Not an anti-technology argument.
Not a partisan political critique.
Falsifiability (“I may be wrong if…”):
AI productivity scales faster, broader, and more durably than historical precedent.
Fiscal dynamics stabilize without institutional strain.
Central bank independence and policy credibility are reinforced rather than eroded.
Interpretive frame:
Gold is treated as a trust barometer, not a fear gauge. It does not predict disaster; it prices disappointment when optimistic narratives begin to outrun institutional capacity.



Another very interesting essay with a very good summary of what gold’s function in the investment universe has been.
One aspect not addressed in the article is the role cryptocurrency has played in the disruption. Crypto has zero underlying value yet there is a segment of investors that treat it as gold. I note that the value of Bitcoin has essentially flatlined during the same period that gold values have spiked and I wonder how much of that is Bitcoin “investors” cashing out of crypto and into gold?
The AI box contains at least two separate categories. The first, which has been rolling out continuously over the last 5-7 years productively, are the hyper-focused applications of algorithms to replace human data collection and analysis—this is the job destroyer of call centres, legal clerks, greenhouse workers… a whole host of jobs. The second is the quest for the ultimate LLM that can approach general intelligence, and it is this quest that is absorbing and burning through barrels of cash with little prospect of return. As you point out there are resource constraints to the build out of data centres (copper, which you mention, but also grid power, and water, both of which are resulting in resistance to building them). I have no doubt the AI crash of this category will happen when enough people realize the revenue will never materialize to pay off the massive capital investments in this quest.
I look at Berkshire Hathaway, whose portfolio last I checked was more than 50% in cash as a signal “the end is near” for equities.
very interesting questions.
I should think, however, that Trump's overall behaviour is currently stress testing a lot of the traditional assumptions