THIS WEEK · 10 PODCASTS · WEEK OF AUGUST 10, 2026
This week ten conversations converge on a single question: what happens when the credit impulse that carried the AI buildout stops arriving. Ed Dowd opens with the fault line — private credit as the new junk bond market, except opaque, illiquid and about to be stress tested, with the marginal credit engine of the US economy now stalled. Glenn Schorr and Ken Worthington give Steve Eisman the sell-side counterpoint: wealth-channel inflows have gone to essentially zero, redemptions are still running above the quarterly gate, and the real reckoning arrives with a refinancing wall of software maturities in 2028 and 2029. Patrick Boyle supplies the first casualty in forensic detail — a 24-year-old with no trading experience, four times levered on a single thematic bet, losing two-thirds of a $45 billion book in a month, and the volatility-drag arithmetic that made the outcome nearly inevitable. David Rosenberg and Richard Bernstein then widen the lens: strip out AI and the economy is contracting, the credit market leads the equity market by about a year, and the way to stay invested without owning the unwind is to diversify internationally into the growth nobody is looking at. Stephanie Pomboy explains why the bill arrives in the Treasury market, where corporate borrowers are now crowding out the federal government and buybacks have quietly turned into net issuance. Steve Hanke dissects the historic US intervention in the yen, argues the fundamentals still point lower, and lays out the deadly cocktail of accelerating money supply, war and deficits that has the bond vigilantes out of hibernation. Jim Iuorio takes the same set of facts and draws the trade — the Treasury has revealed its fear of higher yields, and gold is front-running the Fed’s return to the bond market. Robert Pape provides the geopolitical engine underneath the oil price and the yield curve: an escalation trap in which Iran gains leverage with every reversal, and a Strait of Hormuz that cannot realistically be reopened before the midterms. Peter Zeihan, Marko Papic, Matt Gertken and Jacob Shapiro map the ten-year horizon on which all of this sits — deglobalization, demographics, the democratization of military force, and the breaking of the growth loop that every modern economic model assumes. And Marc Faber closes the edition with the prescriptive frame that ties the week together: this is the first phase of the piercing of the greatest global investment mania, and the goal now is not making money but losing the least. Each summary is designed to be immediately actionable — whether you are allocating capital, running a business, or simply trying to understand the forces reshaping the world around you.
THIS WEEK’S LINEUP
EP 1 Private Credit Is the New Junk Bond Market — Ed Dowd — Phinance Technologies Founding Partner, Former BlackRock Portfolio Manager — Watch Full Video
EP 2 Private Credit’s Clock Is Ticking — Glenn Schorr & Ken Worthington — Evercore ISI & J.P. Morgan Senior Research Analysts — Watch Full Video
EP 3 Lacking in Situational Awareness — Patrick Boyle — Palomar Capital Founding Partner, Visiting Professor of Finance, King’s College London — Watch Full Video
EP 4 What Ends the AI Trade — and What They Own Instead — David Rosenberg & Richard Bernstein — Rosenberg Research Founder & Richard Bernstein Advisors Founder — Watch Full Video
EP 5 Rising Bond Yields Are Everyone’s Problem — Stephanie Pomboy — MacroMavens Founder — Watch Full Video
EP 6 The Yen Bailout and the Return of the Bond Vigilantes — Steve Hanke — Johns Hopkins University Professor of Applied Economics — Watch Full Video
EP 7 Gold Is Front-Running New Money Printing — Jim Iuorio — TJM Institutional Services Managing Director — Watch Full Video
EP 8 The Iran War Is a Trap and America Has No Way Out — Robert Pape — University of Chicago Professor — Watch Full Video
EP 9 Four Ex-Stratfor Analysts on How the World Ends — Peter Zeihan, Marko Papic, Matt Gertken & Jacob Shapiro — Zeihan on Geopolitics, BCA Research & The Bespoke Group — Watch Full Video
EP 10 The First Phase of the Greatest Investment Mania Is Being Pierced — Marc Faber — Gloom, Boom & Doom Report Editor and Publisher — Watch Full Video
Full summaries with actionable insights and investment focus for each podcast follow on the pages below.
EP 1 - Private Credit Is the New Junk Bond Market
Ed Dowd — Phinance Technologies Founding Partner; Former BlackRock Portfolio Manager
Ed Dowd, speaking on Eurodollar University, makes the cleanest structural case of the week: private credit did not grow alongside the junk bond market, it took share from it, absorbing the riskier paper into a vehicle where managers mark their own books and there is no public quote to watch. His conclusion is that the asset class is the new junk bond market, except opaque, illiquid and about to be stress tested — and because private credit and private equity became the marginal credit engine of the US economy in 2024 and 2025, a stall there transmits directly into consumer, commercial and real estate lending. Dowd is explicit that he is not calling a top, but his read on the AI credit impulse, the semiconductor cycle and the underlying labor data leaves him expecting a drawdown far larger than a routine correction.
Actionable Bullet Points
The Risk Did Not Disappear, It Moved Somewhere You Cannot See It: Dowd’s core claim, drawn from six months of research, is that the public junk bond market now looks like higher-quality credit precisely because the bad credits migrated into private vehicles where there is no mark-to-market discipline, no public quote and no observable spread widening (1:33). The only trackable variables left are fund inflows, outflows and the equity of the listed managers (2:10). He describes the practical consequence with a specific example: loans in a BlackRock fund marked at a hundred cents on the dollar and then written to zero a month later, with the fund’s head subsequently fired (3:15). In the old regime a deteriorating credit walked down from 100 to 95 to 90 to 70 in public view; now it is lights out until the write-down lands. Treat the absence of visible spread stress in private credit as an information failure, not a clean bill of health.
The Marginal Credit Engine Has Stalled — and Credit Is Never Contained: Dowd’s firm concluded in its US economy report that in 2024 and 2025 the marginal credit of the country came from private credit and private equity, funded by commercial banks lending to those non-depository institutions (5:38). That makes the current flow reversal systemically important rather than sector-specific: he expects knock-on effects in consumer loans, commercial and industrial loans and real estate loans, and notes JP Morgan is already scouring its private credit books and reducing credit lines (5:27). The variable that matters most is not redemptions but new inflows — if new money stops arriving, the credit creation engine simply freezes, and Dowd emphasises that is a second-derivative change, which is all it takes (11:20). CLOs are next on his list, with forced selling arriving once rating agencies downgrade, though he expects institutions to sell before the agencies act (21:41).
The Signals to Watch: Triple-C Spreads, Three-Month Bills, and the Dollar: Dowd gives a concrete monitoring list. First, watch whether spreads widen in the higher-quality junk that remains public, and whether triple-C paper keeps widening — he notes Jeffrey Gundlach and other fixed income managers already flagging deterioration there (2:52). Second, watch the front end: he believes the T-bill market leads the Fed, and a mysterious plunge in three-month bill rates ahead of a Fed move would signal something is wrong in credit (7:12). He expects the Fed to be cutting within six months regardless. Third, watch the dollar, which he believes put in an important low in January and is technically trying to trend higher on a global dollar shortage — unambiguously negative for credit, since dollars are the lifeblood of the global system (7:38). He also flags a poorly subscribed Amazon investment-grade deal, covered only 1.6 to 1.8 times, as evidence of cold feet spreading (11:35).
The Last Credit Impulse Went to the Chip Makers — and Raised the Cost of the Buildout: Dowd was surprised by the scale of the final six-month impulse, in which the hyperscalers spent aggressively and effectively handed the money to the semiconductor industry — the pick-and-shovel beneficiaries, exactly as Cisco was for the broadband buildout while WorldCom, Qwest and Winstar went the way of the dodo (9:24). He points to a Morgan Stanley estimate that of the $1.5 trillion in external financing data centres require, $750 billion would come from private sources (9:53). The self-defeating part is the cost: memory is now roughly 30% more expensive, so the AI ecosystem raised the price of its own buildout and lowered its own ROI (27:51). Micron went from a $60 billion market cap thirteen months ago toward a trillion, on 86% gross margins against a historical 30 to 40% (27:28). Dowd’s judgement is that this is where the cycle breaks.
Equity Is Late, and the Drawdown Math Is Severe: With roughly 40 to 45% of total market cap now AI or AI-adjacent and valuations back at levels where the historical ten-year total return including dividends is zero, Dowd argues the arithmetic implies a large drawdown between now and then, and he thinks it is closer than further away (12:54). His number is 40 to 50%, with the reminder that a 50% loss requires a 100% gain to recover (13:14). Semiconductors are 19% of the S&P 500 and have never not had a boom-bust cycle, with historical drawdowns of 50 to 80% — the dot-com bust took the sector down 80%, and Korea, effectively two semiconductor stocks, is already down 30% (13:38). He notes he will not call a top without ten to twelve weeks of price structure, by which point the market will already be down 20 to 25%. Meanwhile the real economy is weak: he cites the quarterly census of employment and wages showing non-farm payrolls eight standard deviations too high in 2024 and four in 2025, alongside rising consumer, auto and now early foreclosure delinquencies (15:04).
Investment Focus
Dowd’s framework is the week’s most direct link between an opaque asset class and a broad market repricing. The investment template: (1) stop treating tight public credit spreads as a signal of system health — the risk migrated into vehicles that do not print a spread, so track fund flows, manager equities and triple-C paper instead; (2) monitor the second derivative of private credit inflows rather than the headline redemption numbers, because the freeze happens when new money stops, not when old money leaves; (3) use the front end of the curve as the early warning — an unexplained drop in three-month bill rates ahead of the Fed is the tell that credit is seizing; (4) respect the semiconductor cycle rather than the AI narrative, since 19% of the index sits in a sector that has never avoided a boom-bust and whose last cycle drew down 80%; (5) recognise that Dowd expects the eventual policy response — rate cuts within six months and recapitalisation at lower prices — to produce the real winners after the unwind, not the frontier model companies leading it now.
EP 2 - Private Credit’s Clock Is Ticking
Glenn Schorr & Ken Worthington — Evercore ISI & J.P. Morgan Senior Research Analysts
Steve Eisman convenes two sell-side analysts who cover the alternative asset managers, brokers and banks from the inside — Glenn Schorr of Evercore and, in his first appearance on the show, Ken Worthington of J.P. Morgan — and the result is the most granular picture available of where private credit actually stands. Their view is more constructive than Eisman’s, but the facts they supply are not: wealth-channel inflows into direct lending are effectively invisible, redemption requests still exceed the quarterly gate in most funds, private equity has underperformed public markets for the first time since 2008-09, and a $270 billion wall of sponsor-held software maturities in 2028 and 2029 means the refinancing negotiations begin within the next two to three quarters. Recorded on 30 July, the conversation also captures a remarkable set of bank earnings and a hedge fund liquidation on the same day.
Actionable Bullet Points
The Wealth Channel Has Gone Quiet — Inflows Near Zero, Redemptions Still Above the Gate: Schorr’s description of demand for direct lending products in the wealth channel is that you can almost not see the line, with gross sales inflows he calls infinitesimal (15:39). Redemption requests have eased from the March and April peak — it is the same people asking rather than new people asking — but in most places they remain above the 5% per quarter limit (15:56). Worthington adds the distribution framing that makes this manageable rather than fatal: the majority of these assets sit with institutions, where the business remains steady, while the retail and wealth slice was tiny to begin with (21:14). What has broken is the growth algorithm. Three years ago retail money was flowing into private real estate; rates rose and it rotated into private credit, which was growing like a weed on the promise of private equity returns at much lower risk; that has now dried up to essentially zero (22:05).
The Refinancing Wall Is the Real Test — and the Clock Starts Now: Eisman’s position is that the market is roughly a year away from finding out what it needs to know, because these loans do not come up for refinancing until then (24:19). Schorr sharpens the timing: with $270 billion of software maturities held by financial sponsors due in 2028 and 2029, normal-course refinancing conversations begin somewhere in the next two to three quarters (28:59). The current state is a stalemate — lenders asking sponsors for fresh equity, sponsors declining, very few keys actually handed back and very few checkbooks actually opened (28:13). What breaks the standoff is capital that has been raised specifically for this: opportunistic credit funds willing to take over the paper or layer in credit at 200 to 300 basis points higher yield, with tighter terms, no liability management exercises and no EBITDA adjustments (29:28). Schorr is explicit that if the underlying companies really are worth half, those worse terms eventually bring markdowns not yet seen (30:06).
Eisman’s Challenge: It Is Not the Cash Flow, It Is the Valuation: Eisman grants the managers their strongest point — that portfolio companies are still cash flowing, still have margins and still have growth — and argues it does not matter (27:07). His mechanism is the public comparable: with ServiceNow down more than 50% from its peak in what he calls the SaaS apocalypse, any private-equity-owned software company has to be worth roughly half of where it was bought, because that is what the public market has done (24:37). At refinancing the lender looks at a company that is performing fine and still demands more equity, because the collateral value has halved. Worthington’s house view is the counterweight: absent recession, as long as middle-market companies are doing generally well, credit quality can deteriorate back toward the mean without widespread default outside software and SaaS (25:21). The practical exercise is therefore manager-by-manager software exposure — Blue Owl is the name most often cited, with technology-focused direct lending products and a stock that has already suffered (26:29).
Returns Are Already Compressing, and the Marks Follow Gradually: Worthington notes that returns in these funds have come down substantially — from mid-teens-plus in direct lending to high single digits and mid single digits, with this year looking weaker again — and that this is already flowing through to valuations, gradually (30:23). Direct lending, the part of the market under discussion, is a little over half of private credit (26:18). On the private equity side the picture is mixed in a way that should not be possible: public markets have doubled in four years, yet private equity has underperformed for the first time since 2008 or 2009, and monetisations are bifurcated (16:52). KKR reported record monetisations the day of the recording, while others will post very limited realisations despite a better IPO market, a better M&A market and record equity highs — which raises the question of whether the assets are good or the entry multiple was simply too high with zero rates in 2020 and 2021 (23:24).
The Banks Are Printing Money — and Every Line Is Levered to the AI Trade: Schorr’s bank checklist is close to as good as it gets: investment banking up 38%, trading up 47%, 7% organic growth in asset and wealth management, expenses under control, strong positive operating leverage and high returns on already-high capital bases (46:29). The weak spot is the core business — take deposits, lend money, keep the difference — where competition on the deposit side is pushing cost of funds up. The forward risk is concentration of a different kind: trading is historically down 17% in the second half versus the first, and banking activity now depends on whether the big AI financing of the day is happening (47:38). Schorr’s rule is that trading does well when participants disagree, and the current backdrop — rate cuts priced out and hikes priced in, geopolitical risk and a constant debate over software and AI — is exactly that environment; equity revenues were up around 68% on volumes up 9 to 10%, with margin balances up about 50% (48:47).
Investment Focus
This is the week’s best inside-out picture of private credit, and it is more nuanced than either the bull or bear caricature. The investment template: (1) separate the institutional book from the wealth channel — institutional flows remain steady while retail direct lending has gone to roughly zero, so headline redemption numbers overstate the systemic problem and understate the growth problem; (2) underwrite managers on software and SaaS exposure specifically, because that is the one place both analysts concede widespread stress is likely; (3) put the 2028-29 sponsor maturity wall on the calendar and watch the next two to three quarters of refinancing negotiations, since that is when the stalemate resolves and the markdowns begin; (4) treat compressing fund returns — mid-teens to mid-single-digits — as the slow, already-visible version of the repricing rather than waiting for a single event; (5) recognise that bank earnings this good, driven by trading and banking rather than lending, are a levered bet on continued disagreement and continued AI financing volume, which makes them a coincident indicator rather than a defensive holding.
EP 3 - Lacking in Situational Awareness
Patrick Boyle — Palomar Capital Founding Partner; Visiting Professor of Finance, King’s College London
Patrick Boyle’s forensic account of the collapse of Situational Awareness is the week’s most useful case study, because it converts every abstract warning about the AI trade into arithmetic. Leopold Aschenbrenner, 24, with no prior trading experience, raised money from Silicon Valley on the strength of a 165-page essay, levered it roughly four times through Wall Street prime brokers, and built what was presented as a hedged long-short book but was in fact the same directional bet placed twice. When the AI trade wobbled in July, both legs lost money simultaneously, margin calls arrived during his wedding weekend, and the fund lost about two-thirds of a $45 billion book in a single month. Boyle’s deeper point is not about one manager: it is that leverage applied to a concentrated, high-volatility theme marches the typical outcome toward zero even when the average expected return looks spectacular.
Actionable Bullet Points
It Was Never a Hedged Book — the Risk Was Rotated, Not Reduced: Boyle uses Matt Levine’s framing to explain what a long-short equity fund is supposed to do: strip out market risk by pairing good longs with bad shorts, so a market-wide fall cancels out and you are left exposed only to your stock selection (18:01). Aschenbrenner did something else. He was long the perceived AI winners — SK Hynix, SanDisk, Bloom Energy — and short the perceived losers, mostly software firms like Adobe (19:08). Both sides depended on the identical theme, so the portfolio had not reduced its risk, it had rotated it and intensified exposure to one idea (19:26). Boyle’s analogy is betting $100 that one team wins and then, as a hedge, another $100 that their opponents lose. When the market began to doubt the AI timeline, the longs fell and the shorts rallied at the same moment. Interrogate whether a stated hedge is genuinely uncorrelated or simply the same view expressed twice.
The Leverage Was Applied to Exactly the Wrong Kind of Asset: Boyle points out that leverage normally has a logic — you borrow against something boring and stable to turn a tiny return into a decent one. Aschenbrenner levered the most volatile stocks on the planet roughly four times, borrowing three to four times investor capital from prime brokers including Goldman Sachs and J.P. Morgan (20:34). The same trade was crowded globally: South Korean retail investors, locally known as ants, had borrowed heavily on margin to buy Samsung and SK Hynix, so when the tech rally wobbled in July their margin calls forced selling that drove SK Hynix down while the software names on the short side rallied (22:01). His book took on water from both sides at once, prime brokers demanded collateral, and rivals who saw the positions understood immediately these were directional bets in distress and could trade ahead of the forced selling (23:00).
Volatility Drag Is the Mechanism, and It Scales With the Square of Leverage: This is the section worth committing to memory. Because gains and losses are asymmetric — a 50% fall requires a 100% gain to recover — the compounding rate you actually earn is approximately your average return minus half your volatility squared (32:55). Crucially, expected return scales linearly with leverage while the drag scales with its square: double the leverage and you double the return but quadruple the drag (33:34). Boyle runs the numbers. A concentrated AI basket with a 15% expected return and 40% volatility carries an 8% drag and compounds at a perfectly healthy 7% a year unlevered — a good business you could run for decades (33:57). Lever it four times and the headline average return quadruples to 60%, but the drag grows sixteen-fold from 8% to 128%, turning a typical compounding return of plus 7% into minus 68% a year (34:19). The bad outcome was baked into the arithmetic from the start.
The Average Return Is Real — Almost Nobody Gets to Reach It: Boyle is careful to concede that the 60% expected return is not a lie. There are versions of the world where the AI thesis compounds exactly as forecast and the bet pays off spectacularly. The problem is how little probability lives there: the mean is propped up by a thin sliver of outcomes far out in the right tail, while the median — the outcome sitting in the middle of everything that could happen — is a heavy loss (35:28). No investor receives the average; each gets the single path they land on. And leverage removes even the theoretical consolation, because a large enough downswing triggers a margin call and the prime broker closes the position at the bottom, before any of the average-rescuing outcomes can arrive (36:48). You are left with all of the drag and none of the jackpot. Boyle adds the sentiment corollary: a huge short-term return like 400% is usually the last thing you see before the whole thing comes apart (38:04).
Concentration Risk Extended to the Investor Base — and to the One Asset That Saved Him: The investors were mostly Silicon Valley insiders whose startup equity, options, salaries and property already ride the same wave, so the fund did not diversify their risk, it stacked more of the exposure they already had (37:10). Boyle’s point is that the purpose of an outside investment is to own something that moves when the rest of your life does not. The final irony is that the fund survived only because part of the book was too illiquid to sell — a multi-billion dollar private stake in Anthropic that could not be margin-called because there is no daily price to mark against (28:10). Citadel won an auction against Jane Street and Millennium for the collapsing public book, with Jane Street bidding on the remains of a fund it had itself invested in (25:44). Aschenbrenner emerged with roughly $8 to $10 billion (29:23), and days later wired $400 million into a single private startup (40:08).
Investment Focus
Boyle turns the AI unwind into a repeatable risk framework rather than a morality tale. The investment template: (1) stress-test any “hedged” portfolio for whether the longs and shorts share a single underlying driver, since a thematically paired book is one bet placed twice and loses on both legs simultaneously; (2) apply the volatility drag formula — average return minus half volatility squared — before accepting any levered return projection, remembering that drag scales with the square of leverage while return scales linearly; (3) distinguish the mean from the median when evaluating skewed strategies, because the advertised average is carried by tail outcomes almost no investor path reaches; (4) treat leverage on high-volatility assets as a survivorship problem rather than a return problem, since margin calls close the position before the good outcomes can arrive; (5) audit your own correlation to the position — if your job, your equity and your outside investments all depend on the same theme, the fund is not diversification, it is concentration with extra steps.
EP 4 - What Ends the AI Trade — and What They Own Instead
David Rosenberg & Richard Bernstein — Rosenberg Research Founder; Richard Bernstein Advisors Founder
Two former Merrill Lynch colleagues who called the housing bubble together reunite on Excess Returns and disagree productively. Richard Bernstein argues bubbles are inherently inflationary because they misallocate capital — data centres are being built while housing starves for capital, exactly as the late 1990s starved the energy sector — and he believes the Fed should be hiking. David Rosenberg argues the opposite on inflation: with demand running below potential supply, unit labour costs at 0.5% and the trimmed mean decelerating, the disinflationary forces are building. Where they converge is more useful than where they part. Both see a bubble, both see the credit market leading the equity market, and both arrive at the same prescription — stay in equities, but internationally diversified, because the growth story outside the United States is converging and almost nobody is looking.
Actionable Bullet Points
Strip Out AI and the Economy Is Already Contracting: Rosenberg’s numbers are the sharpest framing of the week. Roughly 50% of business capex is now AI or AI-related and growing at about 18% in real terms, while the ex-AI half is running negative year over year (20:12). After the third quarter, the four-quarter average of real GDP growth will be 1.4% against the Fed’s own estimate of potential growth of 2.0% — demand below potential supply, which puts downward pressure on underlying inflation (9:34). Second quarter GDP printed 1.5 (22:26), and he expects a fourth quarter run rate below 1.5%; without the AI boom, he says, we would probably be in recession (24:11). Non-residential construction is negative year-on-year and housing is contracting — neither was true in the late 1990s, when strength was genuinely broad-based and nobody talked about a K-shaped economy (23:15). Do not mistake headline GDP for a broad expansion.
The Credit Market Leads the Equity Market by About a Year: Rosenberg returns to the lesson he and Bernstein learned together: the equity market figures it out roughly a year after the credit market does, and the problems in ABS spreads and mortgage bonds were visible long before equities peaked in October 2007 (15:14). He sees the same sequence now — tech stocks hitting new highs while their financing costs rise, CDS spreads widen dramatically and credit spreads widen (15:05). His conclusion is that the Fed does not have to do anything, because the credit market will lead the rolling over of the AI trade on its own (15:34). On inflation he is unmoved by the narrative: the Dallas Fed trimmed mean, which Kevin Warsh once preferred, is running 2.2% against 2.7% a year ago and is at its lowest trend in five years, over 90% of growth is coming from productivity and unit labour costs are running 0.5% year over year (10:52, 12:29).
Bernstein’s Misallocation Trade — Be the Provider of Scarce Capital: Bernstein’s framework from the tech bubble is that bubbles are inflationary because they grossly misallocate capital, and the money always leaves somewhere. In 1998-2000 it left the energy sector so completely that US refineries were literally exploding on 1970s technology, which is why the trade was to be the provider of scarce capital to energy — and energy was one of the best performing sectors for the following decade (16:27). He argues we are there again: compare construction spending on data centres with construction spending on residential housing and the misallocation is visible, with housing unaffordable precisely because there is no new supply, the cost of capital is too high and starts are performing miserably (17:46). His broader claim is that the Fed has never appreciated that abnormal pricing in the financial economy, taken to an extreme, is as damaging to capital allocation as abnormal pricing in the real economy.
Gold: Real Rates, Not the Dollar Narrative, Explain the Correction: Rosenberg’s explanation for gold’s poor year is mechanical. The run-up in Treasury yields has not been driven by inflation expectations — TIPS breakevens are around 2.2%, not far off target — but by the term premium and the real rate, and the gold price has an almost perfectly inverse correlation with real rates (26:02). Add the flight into dollars for liquidity during the Iran war and gold got hit by both forces at once. He puts it in context: this is roughly the tenth mini bear market since the 1999 bottom, and gold did not rally after Lehman either, it sold off first (27:07). The secular driver is unchanged — supply grows 1 to 1.5% a year while demand runs 2 to 3%, with central banks the marginal buyer since 2010, and the gold share of global reserves at 25 to 30% against a 1980 peak of 70% (29:01). His peak forecast remains $6,000 an ounce (33:24). Bernstein does not trade gold at all, holding it as a spare tire against unforeseen uncertainty, and flagged the sudden flood of silver questions a year ago as the sentiment signal not to add (34:04).
The Non-US Trade Nobody Is Looking At — and the Sentiment That Makes It Possible: Bernstein calls it one of the craziest things he has seen: for the fifteen to eighteen years before 2000, venture capital so completely outperformed non-US stocks that the non-US line was barely visible on the chart, and over the last five years non-US stocks have outperformed venture capital — and the response he gets is that the data must be wrong (36:33). His firm now holds the biggest non-US overweight in its history, having launched in 2009-10 wildly bullish on the United States with virtually no emerging market exposure (39:11). The reason is that the growth story is converging: of roughly 200 companies worldwide with projected long-term earnings growth of 25% or more, spread across every region, only one is a member of the Magnificent Seven (40:45). Rosenberg supplies the valuation case — the Shiller CAPE has crossed 41, the highest since 2000, a level reached only 1.5% of the time in US history versus 25% in Canada and far more elsewhere — and the correlation case: unlike the late 1990s when only media and telecom moved with tech, today utilities, financials and industrials are AI derivatives, leaving only health care and consumer staples uncorrelated (42:10, 46:47).
Investment Focus
This is the week’s most complete answer to the question of how to stay invested without owning the unwind. The investment template: (1) use the credit market as the timing signal for the AI trade, since CDS and credit spreads have historically led equities by about a year and are already widening for the hyperscalers; (2) treat ex-AI capex, non-residential construction and housing as the true read on the economy, all of which are negative year-on-year and imply recession without the AI contribution; (3) build the international overweight around growth rather than value — the standard case for non-US is cheapness and dividends, but the actionable insight is that the majority of the world’s fastest-growing companies now sit outside the US index; (4) prefer markets with low AI representation, since Korea and Taiwan carry more concentration risk than the US and much of Southeast Asia, Japan and Latin America carry far less; (5) own gold as a spare tire rather than a trade, positioning for the mean reversion in central bank reserve shares while accepting that real rates, not the dollar debasement narrative, drive the near-term price.
