Six days ago, the Committee for a Responsible Federal Budget without fanfare posted a one-page blog that should have been the lead headline on every network in America. It wasn't. The page carried a single, devastating number: in fiscal year 2026, the United States government spent $1.1 trillion servicing the interest on its debt.

That is 3.4% of GDP — the highest share ever recorded in American history. It is more than the Pentagon's budget. More than Medicare. More than the government spent on children last year.

Combined.

Think about that for a second.

Here is what the Treasury confirmed on October 1st. The federal deficit for the fiscal year that just closed came in at $2 trillion, or 6.2% of GDP, up roughly $210 billion from fiscal 2025. Revenues rose to $5.4 trillion. Outlays climbed to $7.4 trillion. Debt held by the public now sits at $32.3 trillion. That is exactly 100% of GDP, the highest ratio in any American fiscal year other than 1945 and 1946, when the country was finishing a world war.

We are not finishing a world war.

The 10-year Treasury yield traded above 5.3% intraday this week — its highest level since 2002. Every tick higher on the long end adds billions to next year's interest bill, and the Treasury has to roll roughly $9 trillion of debt over the next twelve months to pay for maturities and new deficits. The math compounds in one direction.

Most investors are still running portfolios built for a 2% world. They own duration. They own long-dated corporate credit. They own the quiet assumption that the Federal Reserve will ride in — eventually — and cut rates back to zero. I urge you to reconsider. The Fed does not control the long end of the curve. Bond vigilantes do. And the vigilantes have been voting with their feet since the summer.

Here is the "so what." Gold closed yesterday at $4,152.95 an ounce. That is up 53.8% in twelve months. The move is not retail speculation. The People's Bank of China just extended its gold-buying streak to twenty straight months — its reserves are now 2,346 tonnes, roughly 8% of total holdings. Poland is buying. Turkey is buying. The Czech National Bank is buying. Central banks that have never been in the gold market are steadily building positions. In other words: the world's biggest institutional allocators are telling you, in the plainest possible language, what they think of a $32 trillion debt stack financed at 5%.

If you do not own gold, I urge you to reconsider your allocation. Not next quarter. Now. A 5% position in physical bullion is not a trade. It is a hedge against the arithmetic I just walked you through. Warren Buffett has famously dismissed gold for decades. But as Buffett himself says, price is what you pay, value is what you get. And when the price of credit is rising faster than the government can tax, the value of money-that-cannot-be-printed goes up.

Keep in mind: the fiscal picture does not fix itself. The CBO's own projections put interest payments above $1.5 trillion by 2028. Nothing in Washington — not the tariff receipts, not the DOGE savings, not the growth optimists — is on a scale that bends that line.

Plan accordingly.

Trump told BRICS to stop dumping the dollar — or face 100% tariffs. They panicked and bought more gold than ever.

Dear Friend,

Trump told BRICS nations: leave the dollar and face 100% tariffs.

They didn't get scared. They panicked.

Not into Treasuries. Into gold.

Poland added 102 tonnes. China has bought for 18 straight months. Central banks accumulated 244 tonnes in Q1 2026 alone.

When the US froze Russia's $300 billion overnight, every central bank asked: "Could that happen to us?"

The dollar isn't a currency anymore. It's a weapon. And gold is the only asset with zero counterparty risk.

Trump's "America First" policy is forcing the world into "Gold First." He's not stopping the reset. He's hitting the gas.

Dylan Jovine found the one stock at the epicenter of this $14 trillion repricing.

See the stock central banks are indirectly betting on >>

Two weeks ago in Busan, Donald Trump and Xi Jinping sat down and cut a deal that moved exactly $30 billion of trade and left the $1 trillion question wide open.

Here is what was agreed on September 24th. The United States cut tariffs on Chinese small appliances, toys, holiday decorations, and kids' car seats. China cut tariffs on American agricultural goods, seafood, wood products, cosmetics, and medical devices. Beijing also committed to importing 10 million metric tons of U.S. coal in 2027 and 2028. Both sides called it a de-escalation. For the categories named, it is.

But read the White House statement carefully. On rare earths, the only commodity in this trade war with genuine geopolitical teeth, the readout says only that both sides "continue to work on U.S. concerns regarding supply chain shortages related to rare earths and other critical minerals, with the goal of ensuring shipment levels return to appropriate levels."

That is diplomatic for we did not solve it.

Here is why it matters to you. In July, China's commerce ministry activated a new enforcement mechanism (Announcement No. 26 of 2026) specifically rewarding Chinese citizens who report suspected violations of the rare earth export regime. In June, Shanghai customs placed the chairman of a precision optics firm with an 11.7 billion renminbi market cap under compulsory measures for mis-declaring germanium content on export forms. In May, two Japanese nationals were detained in Dalian on allegations of smuggling export-restricted material. This is not a trade negotiation. This is a national security apparatus.

Most people have moved on from the April 2025 export control shock. They assume the problem is solved because the shelves at Walmart are still full. They are wrong. China still controls the processing bottleneck for roughly 90% of the heavy rare earths that make precision-guided munitions, F-35 engines, and MRI machines possible, and the Busan deal did not touch that leverage.

If you own defense primes trading at peace-time multiples, you are long an input chain Beijing can turn off.

On Monday, October 5th, OPEC+ ministers voted to add another 137,000 barrels per day of production for November. It is the third consecutive monthly increase. Brent closed yesterday at $101.43 a barrel, down $4.38 from the day before, and down nearly 4% on the week.

The headline reads like a bearish story. The reality is subtler.

RBC Capital Markets analysts flagged what the ministers would not say in public: most non-Saudi producers in the cartel are already running flat-out. The number on paper is 137,000 barrels. The actual barrels that reach the market will be meaningfully smaller, perhaps half of that, possibly less. The real story is Saudi Arabia reclaiming market share from U.S. shale and from a Russia that cannot ship through sanctions as easily as it once could.

And one more thing — the dark fleet continues to move. Iranian and Russian crude is reaching Chinese and Indian refineries through the same tankers, flags, and ship-to-ship transfers the Treasury's OFAC designated six months ago. The sanctions regime leaks. The price absorbs the leak. That is why Brent sits at $101 and not $75.

Here's the idea. You are watching a market where OPEC+ has one hand on the throttle and Washington has the other hand on the sanctions lever. Neither side wants a crash. Neither side wants a spike. The band is $90 to $110, and the risk is a Hormuz incident that nobody plans and nobody can control.

Berlin wants to spend €109 billion on core defense next year. Paris wants its military budget to double by 2030. The European Commission has stapled an €800 billion mobilization target (Readiness 2030) on top of everything. The money is on paper — the problem is where it comes from.

France carries 119% debt-to-GDP and a 5.4% deficit — and the French government has announced €54 billion of planned savings for 2027 that have not yet been voted. Italy carries 135% debt-to-GDP. Greece carries 151%. What Brussels will not say out loud is that roughly 70% of the euro area's incremental defense spending comes from Germany alone. The weaker balance sheets simply cannot pay for the ambition.

Here is what that means for your European exposure. The industrial defense primes (Rheinmetall, BAE, Leonardo, Thales) have order books that extend into the next decade. The sovereigns funding those orders do not. If you own European credit, you are financing the gap between what Brussels announces and what Berlin ultimately agrees to pay for.

Horse, meet water.

On April 24th of this year, a Chinese research lab you may not follow closely, DeepSeek, released its V4 model. The release was not a surprise. DeepSeek has been producing competitive open-weights models for eighteen months. What was a surprise was the hardware.

V4-Pro has 1.6 trillion parameters. V4-Flash has 284 billion. Both support one-million-token context windows. V4-Pro beat Google's Gemini 3.1 Pro on the CorpusQA benchmark at a million tokens, 62.0% to 53.8%. Those are headline numbers. They are not what matters.

What matters is this: the model was trained and runs entirely on Huawei Ascend chips. No Nvidia. No TSMC-fabricated U.S. silicon. Chinese design, Chinese fab, Chinese software stack, end to end.

Jensen Huang said it himself in a widely circulated interview this spring:

That day arrived. The market has not fully absorbed what it means.

Here is the framework. For four years, Washington's China AI strategy rested on a single bet: by denying Chinese labs access to high-end Nvidia GPUs (H100s, Blackwells, the follow-ons), the U.S. could keep the frontier of AI inside American data centers. The policy was built on an assumption that Chinese chips could not catch up. That assumption is now empirically wrong. Wei Sun at Counterpoint Research put it in one sentence: V4 "allows AI systems to be built and deployed without relying solely on Nvidia."

Let me put this in plain English. If a Chinese lab can train and serve a frontier model on domestic silicon at a tenth of OpenAI's inference cost (DeepSeek charges $0.28 per million output tokens on V4-Flash, against GPT-4's $30), then two things happen simultaneously.

First, Nvidia's China revenue, which once ran above 20% of total, has a permanent ceiling. The customer that cannot buy from you learns to build its own. That is not a cyclical story. That is Lindy's Law working against the moat.

Second, the global AI buildout bifurcates. One stack is Nvidia-plus-TSMC-plus-American-hyperscalers. The other is Huawei-plus-SMIC-plus-Chinese-cloud. Each stack serves roughly half the world's compute demand, each at different price points. The economics of the American stack have been priced for monopoly rents. Monopoly rents require a monopoly.

Third, and this is the piece that matters for your portfolio, the capital expenditure cycle gets ugly. Hyperscalers have committed roughly $600 billion of 2026-2027 capex on the assumption that frontier inference stays expensive. If a Chinese open-weights model running on Chinese chips undercuts that pricing by 90%, the ROI math on those data centers deteriorates very quickly. The capex does not stop. It slows. The slowdown gets priced into the second-order plays first: specialty power providers, cooling vendors, data-center REITs, and the smaller fabless names that depend on hyperscaler orders.

So why does this matter to you? Because the AI trade, the single most crowded trade in global equities today, has been constructed on an export-control assumption that is now provably false. The companies that benefit from the knowledge monoculture of a single stack are overvalued relative to a world in which two stacks compete. The companies that profit either way, the hyperscalers themselves, are the safest bet. The pure plays on the Nvidia moat are the ones to trim.

I urge you to look through your AI exposure line by line this week. Ask one question of each position: does this company's value proposition still hold if a credible Chinese alternative exists at a tenth of the cost? If the answer is no, do what Howard Marks calls second-level thinking. The first level says AI is up and to the right. The second level says the moat is wider on one side of the Pacific than the market is pricing.

Think about what that means for your allocation.

The April 8th U.S.–Iran ceasefire remains fragile. Both sides have violated it repeatedly: Israeli strikes on Lebanon, Iranian attacks on Gulf shipping, continued U.S. interdiction of tanker traffic. If you own energy equities on the assumption of a stable $95 to $105 Brent band, watch for the first incident in the Strait that moves tankers to insurance-prohibitive rates. That is when the equity trade becomes the volatility trade.

The last remaining U.S.–Russia nuclear arms control treaty expired on February 5th of this year. Moscow has told Washington through back channels there will be no replacement in 2026. The infrastructure of nuclear restraint built across sixty years is now effectively dismantled. If you have equity exposure to defense primes with a nuclear-modernization pipeline (Northrop, General Dynamics, BWX Technologies), the next five years are a secular tailwind.

On November 4th, Treasury releases its quarterly refunding schedule. The size of coupon issuance and the mix of bills-to-coupons will tell you whether Secretary Bessent plans to term out the debt stack at today's 5.3% ten-year, or whether he keeps shoving issuance into the bill market and praying short rates come down. The announcement is the single most important fiscal event of Q4.

The People's Bank of China just logged its twentieth consecutive month of gold purchases. The pattern is now predictable enough that gold traders position for it. If you want a leading indicator of de-dollarization intent, watch the monthly PBOC reserve release. The streak will break eventually. The question is what message it sends when it does.

The cartel meets again in early November. Watch whether the Saudis maintain the monthly cadence of production increases or discreetly pause them. A pause means they see demand weakness the macro data has not yet caught. A continuation means they are confident in the band. As a paid-up subscriber, you will see our note the morning after.