10
Sep 2026
Norway Is Selling Treasuries. China Is Buying Gold.
The 10-year Treasury yield has touched 4.85%, its highest since late 2023, and the 30-year has pushed past 5.3%. Brent crude settled above $101. Gold is trading near $4,400 an ounce, roughly 20% below the record it set in January.
The reflex reading is that this is bad news for gold. Higher yields mean you get paid to hold government debt, which raises the opportunity cost of owning an asset that pays nothing.
That reflex is watching the wrong number.
The nominal yield tells you very little
What you actually earn is the yield after inflation: Real yield ≈ nominal yield − inflation expectations
Run today's numbers. A 10-year at 4.85% against CPI of 3.4% in the last reading leaves you roughly 1.4 points of real return. Thin, but positive — and that is the entire foundation of the argument that bonds now beat gold.
Now finish the calculation the way a taxpayer has to. Treasury interest is taxable as ordinary income at the federal level. At a 37% marginal rate, a 4.85% coupon nets 3.06%. Against 3.4% inflation, the real after-tax return is negative. The investor being urged to sell gold and go collect a yield is, once the IRS is done, paying for the privilege of lending to the government.
And the direction of travel isn't obviously friendly. Producer prices rose 4.7% year over year in the last report, with August consensus running higher still. That is not the profile of an inflation problem that has been solved.
Watch who is stepping back from the auction
Norges Bank Investment Management was founded in the early 1990s to invest Norway's oil and gas wealth. It now manages the world's largest sovereign wealth fund, roughly $2.3 trillion, and U.S. Treasuries make up the bulk of its bond holdings.
NBIM has proposed cutting the government share of its benchmark bond index from 70% to 50%. Treasuries, as its largest government bond position, would take the biggest reduction — about $80 billion of the roughly $215 billion held at the end of June, according to Reuters. Another $20 billion of Japanese government bonds would go, alongside a reduced euro-area allocation.
Be precise about what this is and isn't. NBIM has framed it as a search for return and diversification rather than a verdict on the United States. It is also a proposal rather than a transaction: it goes to Norway's finance ministry in January, then through the annual white paper process and a parliamentary hearing, with execution unlikely before 2027.
Which is exactly why it's worth reading. This is not a political gesture. It is one of the largest, most patient and least emotional investors in the world concluding that U.S. government debt no longer pays enough for the risk it carries. Mohamed El-Erian's read was that the size matters less than the signal — traditional buyers are becoming less reliable.
Norway isn't the only one. Japan has been repeatedly selling Treasuries to support a weak yen. Different motives entirely, same effect on the auction. When two of the largest traditional holders are reducing at the same moment for unrelated reasons, the identity of the marginal buyer becomes a live question. And that buyer wants a better price. Higher yields are what a better price looks like.
Rhetoric doesn't pump oil
Treasury Secretary Scott Bessent has been warning investors that they are making a mistake bidding up oil and driving Treasury yields higher, on the basis that he has information about government policy plans they don't have.
The market has been unimpressed. In August the Treasury said it would at least double its debt buybacks. This week it attached a number: $6 billion, triple the standard $2 billion operation, though Wall Street had been positioned for $7 to $8 billion. Yields rose anyway, to the highest level in nearly three years. As some analysts have put it, the market called his bluff.
The skepticism has a simple basis. None of this touches what is actually pushing yields up. A $6 billion buyback is a rounding error against roughly $8.4 trillion of government securities rolling over between now and year-end, in a market where corporate borrowers — including more than $1.5 trillion of new AI-related issuance — are competing for the same dollars. Behind that sits a deficit stacked on a national debt above $40 trillion, and a war in Iran keeping a bid under crude.
That is the same limitation monetary policy runs into. Higher rates work on demand. They make mortgages, corporate credit and consumer borrowing more expensive, which slows spending — a real tool that does real work. But rates don't pump oil. They don't build houses. They don't end a war. And they don't close a fiscal gap; they widen it, because the government has to refinance its own debt at the new rate.
Is the administration prepared to spend multiples of what the market expects? Has fiscal discipline simply left the building? Time will tell. Gold has been a port of safety through this kind of saber-rattling for centuries.
Who owes you what
There is a second question worth asking about where your money actually sits, and it surfaced this week from an unlikely direction.
Bill Dudley, the former president of the Federal Reserve Bank of New York, argued in a Bloomberg article that the U.S. process for resolving failed banks remains inadequate, and that regulators should not be loosening capital requirements and supervision before they fix it. His evidence is recent. The failure of Silicon Valley Bank's parent cost the FDIC nearly $20 billion, recovered through higher deposit insurance assessments on every other bank. Switzerland's decision to fold Credit Suisse into UBS rather than resolve it left the country with a far more concentrated banking market.
His prescription is to institutionalize bail-ins — a regime in which failure is contained rather than transmitted, and losses land on the failed bank's own investors instead of on other banks or on the government and its taxpayers.
On the policy merits he is right. Bail-ins are better than bailouts. But read it as a depositor rather than as a regulator and it lands differently. A bail-in framework is a formal acknowledgment that when an institution fails, someone inside it absorbs the loss. A deposit is not a vault holding your money. It is a claim on a bank's balance sheet, and the live conversation in Washington is about clarifying where in that line you stand.
Gold doesn't pay a coupon. That's the point.
The standard criticism is true. Gold pays no dividend, no coupon, no interest. But a bond and an ounce of gold are not competing products and comparing them on yield is a category error.
A bond is a promise. Someone owes you money, you are compensated for the risk they don't pay, and you are repaid in the same currency whose purchasing power is the thing in question. Gold is the absence of a promise. No issuer, nothing to default on. What it offers instead is scarcity, liquidity, and a long record of holding value against currencies that didn't.
Which is why gold isn’t just an inflation hedge. It is a hedge against the monetary and fiscal response to problems governments cannot grow or tax their way out of. CPI is the symptom. The balance sheet is the disease.
Stackable, not hackable
The obvious modern objection is that scarcity now comes in digital form. This month was a reminder of what that convenience costs.
The Bitcoin-linked Liquid Network lost roughly $320 million to an attack. Bitcoin's own protocol was not broken — it rarely is. What failed was the layer built around it: the wallets, custody arrangements and transaction infrastructure that holders actually have to depend on. That is the pattern, not the exception. Roughly $1.4 billion has been taken across about 250 attacks so far in 2026, and attacks on cross-chain bridges have gone from three incidents in 2025 to 26 this year, according to DefiLlama.
A one-ounce coin in a safe or an allocated vault has no software dependency, no bridge, no counterparty, and no update that can quietly introduce a flaw. That is not an argument that digital assets are worthless. It is an argument about which risks you are actually taking. Bitcoin shares gold's scarcity story. It does not share gold's physical simplicity.
Look at who is buying
The strongest tell is not in the commentary. It is in the flows.
The People's Bank of China added 650,000 ounces of gold in August, its largest monthly purchase since 2023, extending its buying streak to 22 months. It did that while prices were rising.
And the institutions that trimmed gold during this year's retreat have been putting it back on. Amundi, Europe's largest asset manager, has been buying on the expectation that gold returns to $5,000 an ounce by year-end. Pictet, Robeco and Fidelity International have added back as well. Bullion rose almost 10% in August as the so-called debasement trade revived.
Their reasoning is not complicated: money is moving out of the dollar and into hard assets because confidence in U.S. fiscal credibility has slipped. Not all of them think the alarm around the dollar is warranted. Most of them agree that a steady drift toward more diversified portfolios gives gold a durable base to build on.
The bottom line
Higher yields are a genuine near-term problem for gold. But they are only a durable problem under one scenario: inflation is beaten, real yields stay high, and the fiscal picture stabilizes.
That is not the scenario the bond market is pricing. It is not the scenario Norway's sovereign wealth fund is repositioning for. And it is not the scenario the People's Bank of China has been buying into for 22 consecutive months.
Gold can fall further from here. It can always fall further. The question is whether the reasons to own it have changed. Debt hasn't improved. Deficits haven't closed. The political appetite for austerity hasn't appeared. Meanwhile the largest and most patient pools of capital in the world are voting with their allocations.
If yields are rising because lenders want more compensation for deficits, debt and currency risk, then the same force holding gold below its January high is the reason to own it over the next decade.
Weakness in that environment isn't a warning. It's an entry point.
Past performance is not indicative of future results.
