10 min read2 viewsinvesting

Odd Lots: Neither Guns Nor Butter Can Be Cut, So the Leg That Gives Is Money

A half-finished data centre corridor at dusk, a three-legged surveyor's tripod standing on a cracked slab in the foreground with one leg sunk into fresh concrete, the corridor running away into the depth toward transformer gear and a thin pale sky

Notes on the Odd Lots episode of 16 November 2026 (recorded on the morning of 12 November), where Joe Weisenthal and Tracy Alloway talk with Tian Yang, who runs the VPX Fund at Variant Perception, about his impossible trilemma: sovereign mobilisation, domestic cohesion and sound money are hard to hold together, and historically the leg that gives way is sound money. On the day of recording, Jensen Huang's departure from NVIDIA was confirmed, the stock fell as much as 14% intraday and the market cap dropped below $4 trillion, while the S&P 500 was down 0.17%. Tian Yang hangs the whole argument on one assumption: that a lot more compute still buys a lot smarter a model. Educational listening notes only — no investment advice, no price targets, and the judgements here may be wrong.

  • impossible trilemma
  • podcast-notes
  • sovereign risk
  • inflation
  • gold
Contents
  1. ”Sovereignty,” in plain words
  2. Why the leg that gives way is money
  3. The company at the centre falls 14%, the index loses 0.17%
  4. Inflation is hot and the gold coins are falling
  5. Where this argument breaks
  6. A few other things worth remembering
  7. Further reading
  8. The one thing worth taking with you

A half-finished data centre corridor at dusk, a three-legged surveyor's tripod standing on a cracked slab in the foreground with one leg sunk into fresh concrete, the corridor running away into the depth toward transformer gear and a thin pale sky

When the world follows the way, war horses are sent back to manure the fields; when it does not, war horses are foaled on the frontier.

—— Laozi, Tao Te Ching, chapter 46 (Spring and Autumn period; my own translation)

In the Odd Lots episode of 16 November 2026, recorded on the morning of 12 November, Joe Weisenthal and Tracy Alloway talk with Tian Yang, who runs the VPX Fund at Variant Perception, about the impossible trilemma he has written up: sovereign mobilisation, domestic cohesion and sound money are hard to hold together at once, and historically the leg that gives way is sound money. The day they recorded, Jensen Huang’s departure from NVIDIA had just been confirmed, the stock was down as much as 14% intraday with the market cap below $4 trillion, by 9:55 a.m. it had recovered to down 7%, and the S&P 500 was off 0.17%. The whole argument hangs on one assumption: that a lot more compute still buys a lot smarter a model. He says the evidence so far still points to yes, and if that line stops, everything built on top of it stops too. Every number is as stated at the time of recording.

”Sovereignty,” in plain words

Tian Yang says one theme sits behind everything he has watched in markets these last few years, and he calls it the primacy of sovereignty. In plain words: sovereignty is the ability to say no to a rival, and to say it you cannot depend on that rival for anything that matters. Hence technology, food and energy get pulled onto the mobilisation list one by one.

His example starts in China. Worried that seed technology sat in Western hands, China went after the Syngenta deal in the 2010s, the US foreign investment review signed off — and that turned out to be the last thing the West tolerated, followed by the ZTE sanctions, the Huawei sanctions, the whole run of them. So this works both ways, and the feeling on the international side is concrete: the other side can cut you off.

What makes the definition useful is that you can check it. “Who can cut you off right now” is a fact you can go and verify; a policy slogan is not. As for Jensen Huang, the analogy Tian Yang reached for first was Oppenheimer — a pivotal figure, a national hero to begin with, who later leaves the stage once his loyalty is questioned. He put it carefully: we do not know what Huang divulged. What he cares about is the pressure at the margin, because the sovereign demands total trust in strategic assets, and this asset is a nuclear bomb plus commercialisation potential.

Why the leg that gives way is money

The derivation runs in three steps. One: with a foreign rival, mobilisation and defence spending are locked in, with no room to negotiate. Two: sovereignty needs domestic cohesion, and in the age of social media society is polarised with no time to resolve it, so you buy cohesion with fiscal deficits and keep doling out the butter. Three: with both of those locked in, history says the leg that gives way is sound money.

A triangle whose three corners are sovereign mobilization, appeasing the home front, and sound money; the first two corners are drawn with heavy solid outlines and labeled "locked," while the third is drawn in grey dashes with its edges broken open, showing that when the three cannot be had together, the leg that gives way is sound money.

His main comparison is Lyndon Johnson in the 1960s. The Great Society and Vietnam ran at the same time, and Johnson began by saying both could be paid for without tax rises; as the deficit widened there came the famous anecdote, with Fed chair William McChesney Martin hiking rates against his wishes and being summoned to the Texas ranch for a face-to-face confrontation. Then the London gold pool broke in 1968. Gold was pegged at $35 an ounce, but the foundations under the peg were dismantled a layer at a time: the gold cover requirement was repealed, so less gold was needed to back the currency, and as policy turned more inflationary, countries like France began selling dollars and pulling gold out of the pool, until a free market price appeared that had broken away from the official one. That was 1968; Bretton Woods itself did not end until Nixon in 1970 and 1971.

A flat official-price line marked $35 an ounce runs to the right, where a free-market price line branches off and climbs away; beneath the flat line stand three props, two of which have become empty grey dashed outlines while one remains solid, showing that the price only broke away after the supports holding it were removed one by one.

So he reads the gold rally that began in 2023 as the same kind of warning: deficit worries that will not go away, butter and entitlements that cannot be cut, mobilisation still being added to. His claim goes one layer further than “printing money causes inflation,” and the layer is physical. In the globalised era the price of money could do the rationing, because things flowed to wherever they were needed; now that borders block the physical movement of things, the price of the things themselves has to do the rationing. That is also why the Iran war shows up in his report as a catalyst — wars are hard to stop once started, and this one forces a lot of physical constraints to hit at the same time.

The company at the centre falls 14%, the index loses 0.17%

Joe asks it straight: we have spent years describing this company as the beating heart of the market’s most important industry, its future has never been more uncertain, and the index has not even given up a quarter of a percent.

Three downward bars hang from the same zero line: on the left the core company's intraday drop of 14 percentage points is the longest, in the middle the same stock after the move narrowed is about half as long at 7 percentage points, and on the right the broad index is only 0.17 percentage points, so thin it is barely visible.

Tian Yang’s answer is financing. AI as a whole is a marathon, because of the sheer sums that have to be raised over the next few years — and yet every marginal raise is a sprint. The circular financing arrangements around NVIDIA look to him like the marks a sprint leaves: a normal raise lets everyone come in, do their analysis and take their time, whereas now the next tranche has to be assembled on a deadline so the last set of commitments can be met, which is why people inside the ecosystem step in themselves as the anchor and hand the market its confidence. He compares it to government debt: running a deficit is a marathon, and every auction is a sprint.

A very long horizontal line stands for the marathon, and packed along it are many upward spikes standing for one sprint after another, each spike labeled underneath with the next financing round or the next auction, while the right end of the line never closes off.

So why is the index fine? The Oppenheimer analogy has a second layer here: the man left, the H-bomb got built anyway, and the race still has to be won. His read is that if the financing market struggles, governments step in and backstop it — and once governments are in, the winners and losers get reshuffled. A domestic critical-resource provider or a domestic equipment maker goes to the front of the line and may even end up the monopoly supplier, while the fifth frontier lab that cannot get enough compute is a position he would not buy. It swings the other way too: let Anthropic’s listing go well, let OpenAI close its next private round, and SoftBank’s collateral is money good again.

Inflation is hot and the gold coins are falling

Tracy puts her own position on the table: she felt good stacking gold coins early in the year, and this year gold has not responded to inflation, it rallies when rate expectations fall, it does not behave like a safe haven, and as she looks at the screen it is down.

His answer separates the time horizons. The 2023-to-2025 rally corresponds to the London gold pool breaking, that first warning; after a warning comes an adjustment period, and this generation’s Martin may be Warsh — at the margin he hiked, inflation came down, credit had a wobble, things normalised, and yet the broad direction of policy still moves the same way. On top of that, the move early this year carried a meme-trading flavour. He says the long-cycle analogy comes with plenty of counter-trend moves: higher rates are a headwind for gold, while the way gold has behaved tells you the underlying pressure is still there and may take a while to truly explode.

Where this argument breaks

He offers four exits himself, and this is the part of the episode most worth copying down.

The first is an arms-control-style agreement: a US-China AI safety summit, and that leg comes down. He thinks this is the most likely circuit breaker, though the path to it is ugly — escalation has to go far enough that severe physical constraints show up on both sides before anyone admits something must be done. The second is domestic politics turning serious about the deficit. He sees signs: a strong third-party emergence like Ross Perot in the early 1990s, and now even France talking about its deficit. Usually, though, it takes a big crisis first — Reagan and Thatcher did it, and that was after the horrors of the 1970s, with a population willing to accept pain. The third is the compute-to-intelligence slope flattening out, which he says would be a good sign: AI safety is arriving, the thing is finished.

The fourth is the counterintuitive one, and I rewound it. Joe asks: if productivity keeps rising, can’t we have guns and butter forever? Tian Yang says the first-order effect is exactly that — debt constraints loosen, the can gets kicked down the road — but it does not force you to reform, and the live question is what will force anyone to fix the guns-and-butter imbalance. He points to Spain and Portugal in the Age of Exploration: indebted empires suddenly flooded with gold and silver, and the result was that their existing institutional biases got amplified and delayed, research he attributes to Daron Acemoglu. He also cites work estimating that a technology on the order of electrification is worth at most 1.5% to 2% of GDP a year, smaller than the hole to be filled. Tracy asks the same question from the other side: if the sovereign picks winners, doesn’t that invite overinvestment? He says that is the end game — the early mobilisation phase builds duplicates of everything and bottlenecks everywhere, hence inflation; once the overbuild is done and arms control arrives, the tail turns deflationary.

A curve rises into a hump above the zero line and then falls through it, with the first half labeled duplicate build-outs and bottlenecks everywhere, matching inflation, and the second half labeled build-out finished and arms control arriving, matching deflation.

The other path, the one Tracy herself has written about, also gets measured here: financial repression. Suppress rates, run inflation at 4% to 5%, amortise the debt away. He thinks it works, with one side effect that gets less airtime — capital controls. For it to work you cannot let the money flow out first, and he says he cannot find a story in which that is a good thing for markets. The mild-looking solution carries its own condition that asset owners will find uncomfortable, a different way of going wrong from “inflation gets out of hand.”

A few other things worth remembering

  • China has the same trilemma and chose sound money and guns, suppressing household consumption. Tracy says one of her realisations of 2026 is that everyone talks about China wanting to boost domestic consumption and she does not think it does. And when Beijing names chokepoint technologies and leaves out e-retail, stocks and financing adjust, while the US is still early in that adjustment.
  • Tian Yang says the Buffett-style analysis he learned across his whole career assumes the regulatory and geopolitical environment is stable, so all the attention goes to return on investment. Now there is a second question — will you get your money back — and that is a 1-0 event: a US investor in China gets cheap valuations if nothing happens, and money that cannot be repatriated if something does.
  • Joe’s phrasing stayed with me: a marathon is 26.2 miles, this thing has no finish line, so it is a Sisyphean marathon made of one inflationary sprint after another.

Further reading

  • The Odd Lots episode of 16 November 2026, the source for this piece, recorded on the morning of 12 November
  • Tian Yang’s impossible trilemma report at Variant Perception
  • Tracy Alloway’s Odd Lots newsletter piece “The Other Way to Reduce Government Debt,” on financial repression
  • Tracy’s 2021 piece “The Chokepoint Economy”
  • Peter Turchin on long cycles and elites giving something up, and Daron Acemoglu on institutions and technology

The one thing worth taking with you

When three things are all declared non-negotiable, the one that actually gives way is the one with nobody behind it to call and complain.

The trilemma holds up for reasons that have nothing to do with which argument is strongest. Behind mobilisation stands the national security story; behind butter stand the votes; behind sound money there is no constituency that will come after you at the next election, because its victims are spread out a decade from now. When Johnson said both could be paid for, nobody raised a hand for the price of gold.

Here is something I have tried that works outside investing too. Write down the three things you have committed to this month that you have called non-negotiable — the project you are staying late for, the thing you promised a child, the fixed workout slot, the weekly call home. Next to each one write a name: if this slips, who comes looking for you. The ones with a name survive; the one with a blank is the thing you will quietly give up this month. I did this once, and the blank was the workout slot, which I had already skipped for three weeks.

Three rows sit side by side: the left column lists three things you say you cannot break, and the right column holds name boxes for who comes after you; the first two rows have names filled in, while the third row's box is an empty dashed outline marked as already skipped for three weeks.

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.

Comments

Loading comments…

Sign in with Google before posting. Only your name and profile picture are shown.