Water Has No Constant Shape: Bubbles Don't Burst, They Roll

The shape of an army is like water:
water avoids the high and rushes to the low.
So an army has no constant formation, and water no constant shape.
—— Sun Tzu, The Art of War, “Weaknesses and Strengths”

What this episode is about

MacroVoices is Erik Townsend’s weekly macro podcast, aimed at professional and sophisticated private investors. This one was recorded on 6 August 2026, with Viktor Shvets, global strategist at Macquarie Capital, back on the show after roughly two years.

Two years ago the host asked him whether central banks had become too political. Two years later the president has installed his own preferred Fed chair — so the opening is essentially a session of marking the old answer.

The episode covers the Fed, inflation, Hormuz, gold, social polarisation, the AI bubble and China. Scattered on the surface, but the skeleton is one sentence: the backdrop is disinflationary, the volatility is manufactured by us, and the real risk has been expelled from the cycle into places the central bank cannot see.

He has a name for the shape of this world: the Fujiwhara effect — two hurricanes that, once close enough, orbit and reinforce each other. His two hurricanes are the highly disruptive information age and deep financialisation.

Original episode: MacroVoices #544, “Viktor Shvets: How Markets Survive Disruption” (2026-08-06)

The main points

1. The Supreme Court left the Fed standing alone, and the exception doesn’t hold together intellectually. Compared with 2024, he says, the big change is how the Court has settled the question of executive power: a hundred years of independent bureaucracy and independent institutions has effectively been gutted — with one carve-out, for the Federal Reserve. He is blunt that the carve-out cannot be justified: there is no argument for why the Fed should be independent while the SEC, the EPA or others should not. The result is that the Fed is the last independent institution in the United States, but its independence rests on a ruling with no stated logic rather than on a coherent framework.

2. He classifies the new chair not as a hawk or a dove but as a politician. Most of what Warsh wants, he says, few people would object to — communicate less, return risk to the market, shrink the Fed’s footprint. The problem isn’t whether it’s desirable; it’s that it can’t be done. Take communication: the less you say, the more others say, so you get higher volatility and shadow chairs still driving the conversation. And the debate about the role of monetary policy and how far financialisation should go? That should have happened in the 1980s and 1990s — that train left the station long ago. What matters more is the committee: the chair is one voice out of nineteen and one of twelve votes at any time, the dots show a wide spread on where neutral even is, and the new chair hasn’t yet accumulated the standing his predecessors had. That’s fine when nothing is happening. Three, six, nine months from now, in an emergency, you need a chair who can turn the whole ship.

3. Disinflation is the backdrop; the inflation waves are ones we generate. His view hasn’t changed: technology-driven disinflation is the dominant long-term trend — that’s what he takes from Musk’s line about young people not needing to save for pensions, that every product eventually meets a marginal cost curve heading toward zero. So where does inflation come from? From our reaction to events: COVID, Russia-Ukraine, the Iran war, chaotic trade and immigration policy. He runs a counterfactual worth writing down: under a different president, tariffs would not have vanished (the previous administration never removed them) and immigration would still have tightened, but not at that intensity — and it isn’t obvious there would have been a war with Iran at all. On a three-month moving average, inflation was already close to 2% before all this. In other words: stop making waves, and disinflation takes over on its own.

4. He hands over two windows for telling when transient becomes permanent — the question the new chair actually has to answer. COVID was transient. The Iran war will prove transient unless another one is started. The 2025 tariffs will prove transient unless more follow. The danger is that enough transient episodes stacked on top of each other become the default assumption businesses and households operate on: firms raise prices because they expect other firms to, workers organise for higher wages because they expect others to. How would you know? First, consumer and business surveys — small-business expectations for prices over the next three months, wage expectations — which currently show no evidence of unanchoring at all. Second, breakevens and zero-coupon inflation swaps: one- and two-year tenors did spike toward 3–5% during the Iran war, but are now back below 2%, while five-year-five-year-forward and ten-year have sat at 2.2–2.4% the whole way through, essentially unmoved. He applies the same logic to the neutral-rate argument: one camp says the real neutral rate is near zero, the dots say one to one and a half — and both can be right at once, because risk has been expelled from the economic cycle into polarisation, politics, geopolitics, climate, healthcare and technology, and a central bank’s ability to assess those factors, their timing and their magnitude is close to zero.

5. He sorts two new wars using two old ones, and explains along the way why gold didn’t follow the script. Aerial campaigns never succeed, he argues — they pull a country together, even where people intensely dislike the regime they live under. And Iran is not Venezuela or Panama; it is an ancient civilisation that killed three Roman emperors in its time, more than a million square kilometres, with brutally complex society and geography. He files it as a Vietnam: no clear objectives, no exit, no definition of victory — one side’s aims shifting almost weekly, the other’s systematic and stable. Russia-Ukraine he files as Korea: extremely brutal, relatively short, ending at a line. He also notes that Hormuz was never Iran’s only choke point. As for gold, which fell hard during the conflict against most people’s expectations, his explanation is real rates plus the fact that gold isn’t transportable: even with the United States starting the war, investors reaching for liquidity went back to the two deepest pools, the dollar and Treasuries. He thinks that’s temporary — in a scenario where the monetary system genuinely melts down, his answer is still gold, not digital assets.

6. Polarisation has a thermometer, and the US reading is the highest and fastest in the history of that database. He uses V-Dem, run out of Sweden, with data back to the early 1900s, scoring whether groups can still talk to each other, pass legislation, form coalitions — from minus four to plus four. Developed countries are traditionally negative: the US in the 1990s around minus one, Canada and Europe minus one and a half to minus two — more polarised, but not dramatically so. By 2023 the US had crossed into positive territory; for 2025-26 the reading is plus 2.3. And underneath that reading sits the real cause: you don’t have to lose your job to feel you’re falling behind — it’s marginal utility and marginal rewards declining. The wage share of gross domestic income is the lowest since 1947; the profit share is the highest ever at 16-17% (he notes Buffett’s line that anything above six or seven is too high). The top 0.1% — 135,000 households — average nearly $200 million in net worth and control almost 15% of national wealth, against 8% in the 1980s. So the killer isn’t unemployment; it’s usefulness declining drip by drip, like a water torture. He quotes the Palantir CEO to the effect that the average person may end up with a little more money while anyone in AI will be a hundred or a thousand times richer, and says: that’s the problem — it opens a gap no amount of effort closes. Hence a counterintuitive generational observation: those born after about 1985 are not becoming more conservative with age, they’re becoming more left, because they see no way out and are willing to gamble. His three exits: violence, redistribution, or productivity fast enough to eat the problem — and while he believes in productivity long term, he doesn’t believe it arrives in the next five to ten years.

7. Rolling bubbles: AI is everything, and the hard part is which derivative you’re standing on. Asked whether AI is a bubble, he asks back: what is AI? Data centres? Chips? LLMs? Robotics? Automation? 3D printing? Quantum? The answer is that AI is everything, and within a couple of years the distinction between tech and non-tech disappears. So he looks at the stack instead. The bottom derivative is the commodities the technology needs — energy, copper, nickel, cobalt, rare earths, lithium — and he isn’t a believer, because humans are very good at finding this stuff and even better at using technology to optimise it away: LLMs today use less than a third of the energy we thought they’d need three years ago, in four more years maybe a tenth, and once quantum arrives “a lot of data centres will become playgrounds for children.” The second layer is infrastructure, data centres and chips — and here he does not think we’ve peaked: hyperscaler growth rates decelerate, but China, Europe and India pick up the slack. Further along, the models themselves start to commoditise and cannibalise (he expects the future to be open-weight), and what takes the baton are the businesses that need costs near zero before they can work: humanoid robotics and automation, biotech, the metaverse (far too early then, plausibly right now), 3D printing (invented over a century ago, same story), and quantum in five to seven years. That’s the rolling bubble — not one bubble bursting but one after another, until metal bashing, barber shops, pharma, insurance and banks each take their turn. The conclusion stings: concentration of returns stays high — ten stocks delivering half the performance — so the index itself may not move much while the rotation inside it is violent. And the hardest line for an investor: you may have identified a theme with ten or twenty years in it and still be positioned on the wrong derivative of that theme.

8. China: yes and no at the same time. China is far stronger than most assume — the United States has already lost the entire electrification stack, not just EVs and batteries but solar and wind — and from electrification it’s one step to robotics and automation, which China is also taking. Even in science and technology, traditionally an American lead, the gap is closing. The other side is a 45% national savings rate sustained for three decades: too much investment, too much reliance on exports, and what he calls the fastest capital misallocation in human history — eleven to twelve trillion dollars invested every year, close to triple Japan’s GDP, until the planet is simply too small for those exports. Fixing it means raising consumption, which runs into three things: unwillingness to give up central control; a classical view in which consumption is not an independent variable but a product of investment; and geopolitics. He closes with the sharpest line in the episode: China is building the world without earning a return on equity; America is earning the return on equity while losing the world.

Where my thinking went

1. The valuable thing here isn’t his forecast — it’s the set of windows he hands you for marking your own answers.

He makes plenty of directional calls, and none of them are checkable. “Disinflation wins long term” is a claim you can never date as wrong. What you can actually use are the two windows in point four: survey-based price and wage expectations, and the term structure of inflation breakevens.

What makes them good is that they carry their own falsification condition. He isn’t saying inflation will fall; he’s saying if short-tenor swaps are back below 2% while five-year-five-year sits unmoved at 2.2–2.4%, the market has not repriced this as permanent — and conversely, if the long end starts drifting, his whole transient framework should be thrown out. A judgement that can’t say what would change its mind is a position, not an analysis.

I’ve been narrowing down to the first kind. Directional claims always sound smarter, but they can’t be marked — and what can’t be marked never compounds into judgement.

2. “Right theme, wrong derivative” belongs on the checklist for any bottleneck analysis.

When I map an industry I walk up the supply chain looking for the layer that breaks first when demand doubles. This episode adds a dimension I hadn’t been handling head-on: how long that bottleneck survives.

That’s exactly his point about the commodity layer — the choke points are real, but short-lived, because humans find things well and optimise usage better. The evidence isn’t hypothetical: today’s models draw less than a third of the energy projected three years ago. A bottleneck dissolved by engineering and a bottleneck filled by new capacity decay at completely different speeds, and only the second is what conventional supply analysis is built to handle.

So every link I label a bottleneck now gets one more question: is it constrained by physics, or by the current engineering approach? If it’s the latter, its half-life may be shorter than the time I’d need to build a position — the map isn’t wrong, it expires, and walking with an expired map is worse than walking with none.

As for his call that the infrastructure layer hasn’t peaked: that’s something to check, not to quote. His reasoning is that hyperscalers decelerate while China, Europe and India compensate. Every link in that chain is independently verifiable, which means I should verify it rather than borrow it.

3. “You should have no confidence in what you’re doing” — the point isn’t chasing the next bubble, it’s admitting you’ll hold through the de-rating.

His advice is: stay agile, watch for new breakouts, minimise exposure to past winners being de-rated. The first two read like encouragement to chase, but the third is the one that matters, because it names a mistake I actually make — the positions hardest to cut are the ones that proved you right.

If the rolling-bubble picture holds, then “the theme is intact but the returns in this layer are done” becomes the normal case rather than the exception. And in that case nothing bad enough happens to make you sell. The position simply, quietly stops going up — and not going up triggers nobody’s stop-loss.

The episode also hands over a clean example of noise versus structure. The war headlines were the noise: oil and gold that week moved against most people’s scripts, and he says outright that he’s stopped watching the oil price. The structure is the wage share at its lowest since 1947, the profit share at a record, and a polarisation reading that has crossed zero and set a database high. The first moves in front of you daily; the second moves once a year — and only the second changes the tax code, the regulation and the cost of labour a decade out. Attention gravitates to the one that flickers. That’s the part I have to work against.

Worth a look

• Original episode: MacroVoices #544, “Viktor Shvets: How Markets Survive Disruption” (2026-08-06, all major podcast platforms)
• The guest mentions two books of his own, The Great Rupture (2020) and The Twilight Before the Storm (2024), on how the information age and deep financialisation reinforce one another
• The polarisation section draws on the V-Dem (Varieties of Democracy) database, run out of Sweden and publicly available, with a series going back to the early 1900s
• Inflation breakevens, zero-coupon inflation swaps and small-business price-expectation surveys are all public — you can track that set of windows yourself
• The three lines from Sun Tzu’s “Weaknesses and Strengths” are my own footnote while listening, not part of the episode

---

Disclaimer: This is a personal set of listening notes and study material, published for educational purposes. It is not investment advice, an offer, or a solicitation. Institutions, figures and market observations referenced here come from the publicly available episode and public sources; no specific security is recommended and no price targets or entry/exit guidance are given. Investing carries risk — form your own judgement based on your financial situation and risk tolerance, and consult a qualified professional where appropriate. Copyright in the quoted material belongs to the original programme; please listen to the original and support the creators.

Raw Markdown

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.