THIS WEEK · 10 PODCASTS · WEEK OF JUNE 5, 2026
This week we bring you ten exceptional podcasts mapping the same terrain from opposite ends — the AI buildout that is either the greatest capital misallocation in history or the engine of an unprecedented global spending super cycle, the energy shock radiating out of the Persian Gulf, and the hard-asset and regional trades that sit on the other side of it all. Gary Marcus and Julien Garran deliver the definitive bear case: no technical moat, no commercial viability across the supply chain, and a Wicksellian misallocation of capital now running at two-thirds of GDP. Ed Zitron dismantles the “token maxing” era from the inside — token-based billing, the collapse of measurable ROI at Uber, T-Mobile and Walmart, and his verdict that two unprofitable companies are being marched toward IPOs that function as exit liquidity. Vincent Deluard maps a “K within a K,” the high end of the economy now splitting as college-educated millennials crack, and a liquidity tipping point where the yen, oil, and the 10-year all sit on the edge. Chris Whalen warns of an under-discussed energy shock — Persian Gulf refining destruction driving lubricant and diesel rationing and a double-digit inflation wave — even as the AI trade, in his words, “is almost done.” Robert Pape supplies the geopolitical backbone: an “age of instability” in which Iran’s leverage grows by the day as oil inventories count down toward exhaustion. Jay Pelosky offers the bull’s mirror image — a tripolar-world spending super cycle that makes this the best time to be a global investor in 25 years, with the US premium set to unravel in favor of emerging markets and commodities. Jesper Koll brings the boots-on-the-ground case for Japan’s “great awakening” — genuine pricing power, returning animal spirits, and a Bank of Japan he believes is far behind the curve. Lawrence McDonald and Rick Rule converge on the rotation into hard assets — gold miners, uranium, and energy — as inflation traps the Fed and a decade of dollar debasement runs its course. And a documentary on the $25 billion Hudson Yards closes the edition with the cautionary tale of what happens when capital tries to manufacture a neighborhood. 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 AI: The Biggest Capital Misallocation in History — Gary Marcus & Julien Garran — AI Researcher/Author & MacroStrategy Partnership — Watch Full Video
EP 2 How AI’s Push Toward IPOs Became a Death Drive — Ed Zitron — EZPR Founder & CEO, “Better Offline” Host — Watch Full Video
EP 3 Is the “Geriatric” Bull Market in Trouble? — Vincent Deluard — StoneX Director of Global Macro Strategy — Watch Full Video
EP 4 The Markets Know There’s a Problem, the Trump Admin Doesn’t — Chris Whalen — Whalen Global Advisors Chairman — Watch Full Video
EP 5 Bombing and Talking at the Same Time: Iran and the Escalation Trap — Robert Pape — University of Chicago Professor — Watch Full Video
EP 6 The U.S. Market Premium Is About to Unravel — Jay Pelosky — TPW Advisory Founder — Watch Full Video
EP 7 Japan’s Great Awakening — Jesper Koll — Japan Optimist Author, Economist & Investor — Watch/Read Full Conversation
EP 8 How to Listen When Markets Speak — Lawrence McDonald — The Bear Traps Report Founder — Watch Full Video
EP 9 Gold Will Soar Over the Next 10 Years — Rick Rule — Rule Investment Media Founder — Watch Full Video
EP 10 The Collapse of Manhattan’s Newest District — Hudson Yards — Documentary Feature — Watch Full Video
Full summaries with actionable insights and investment focus for each podcast follow on the pages below.
EP 1 - AI: The Biggest Capital Misallocation in History
Gary Marcus & Julien Garran — AI Researcher, Author & Entrepreneur; MacroStrategy Partnership
Gary Marcus and Julien Garran, joined by host George Noble and co-host Jack, deliver the most comprehensive bear case of the week: the large language model industry rests on a technology with unsolved reliability problems and no defensible moat, an ecosystem that loses money at every layer of the supply chain, and a financial structure that increasingly rhymes with Enron. Garran’s framework puts misallocated capital in the US economy at two-thirds of GDP — 23 times the level on the eve of the dot-com crash. Both men expect the unwind to be a multi-year capital-markets event, not a one-year GDP dip, and both point to the same near-term detonators: the death of token maxing, usage-based billing, and the coming IPO disclosures from OpenAI and Anthropic.
Actionable Bullet Points
There Is No Technical Moat and No Reliability — Everyone Is Running the Same Experiment: Marcus’s core technical argument is that the entire industry converged on one approach (LLMs), one method (scaling), and one data source (the internet), which means nobody owns a durable edge — a benchmark organization recently found open models only four months behind closed ones (35:09). Hallucinations, which Marcus first flagged in 2001 (3:46), remain unsolved at roughly a 20% error rate (7:53), producing what researchers now call “work slop”: output that mimics the style of competence while hiding fabricated citations and made-up numbers, costing reviewers more time than it saves (10:57). The only genuine recent advance — Claude Code — works not because of scaling but because Anthropic bolted on 500,000 lines of code and 50 tools, i.e. the neurosymbolic approach Marcus has advocated for years (49:57). The predictive implication is severe: if the right answer to AI is not the world’s biggest supercomputer, the roughly $2 trillion of infrastructure may be largely stranded. Do not underwrite the AGI-via-scale story.
The Entire Ecosystem Is Not Commercially Viable — Losses Cascade From Top to Bottom: Garran’s “golden rule” is that you cannot use a generalized LLM to build an app, product, or service that is commercial across the supply chain. The losses stack: data centers run at a loss subsidizing the frontier model labs, which lose roughly two dollars for every dollar of revenue (14:32), which in turn subsidize the front-end app developers (Perplexity, Cursor, Replit, Lovable), which subsidize the end consumer. None of it self-funds. The trigger Garran flags is precise — June 1, the day Microsoft moved GitHub Copilot from a flat fee to usage-based billing (15:31) — because that is the moment customers must ask whether the tool actually makes them money. Anthropic’s rumored profitable quarter is dismissed as an anomaly: it occurred at the height of token maxing and included an estimated ~$2 billion one-time subsidy from Elon Musk against a ~$550 million profit (18:56). Watch the usage-based repricing as the mechanism that breaks the model.
The Financial Engineering Rhymes With Enron — The Loss Lands in Grandma’s Annuity: Jack walked through Michael Burry’s analysis of a special-purpose vehicle (Valor) assembled by Apollo, XAI/SpaceX, and Nvidia to buy $5.6 billion of Nvidia chips — of which $1.9 billion came from Nvidia itself, booking revenue by buying its own product (20:06). The remaining debt was placed by Apollo into Athene, the insurance company Apollo owns, leaving the exposure off the operating companies’ balance sheets and inside retirement annuities (21:01). Layer on the S&P index rule changes that will force passive retirement funds to hold an overvalued SpaceX, and Google’s earnings beat in which roughly $30 billion of $60 billion came from marking up its Anthropic stake (1:19:51), and the picture is paper gains on illiquid shares. As Jack put it, you can fake the accounting but you cannot fake money in the bank — and the cash balances are falling. Treat the depreciation cycle as the moment these marks reverse.
Misallocated Capital Is Two-Thirds of GDP — and It Is Not Rate-Sensitive: Using the Wicksell spread (the neutral rate sits a couple of points above nominal GDP growth; the spread hit a record minus 12% in the pandemic), Garran estimates total misallocated capital — AI plus crypto, private equity, and private credit — at two-thirds of US GDP, 23 times the dot-com eve (27:06). The critical insight is that this capital did not stop being deployed when rates rose in 2022; it is sensitive only to the market’s belief that a commercial ecosystem will ever materialize. The first cracks are already visible: Oracle and data-center debt trading poorly, CDS blowing out (24:07), and bank credit desks “dancing near the door,” with Deutsche and others stepping back (29:30). When the debt markets decide not to fund the next round, the unwind begins — and Garran is explicit that this is a multi-year capital-markets upset, not a transient dip. Identify funding stress, not valuation, as the timing signal.
The Endgame Is Reflation — Position for the Resources and Emerging-Market Phoenix: Garran estimates the wealth effect plus the data-center buildout is adding roughly three percentage points to nominal GDP; if it merely stops growing, that is minus three points, and if it reverses, minus three to six — a recession by itself (1:02:24). There is no clean bailout, because GPUs will be near-worthless in a couple of years and cannot be parked on the Fed’s books the way mortgages were. The likely path is extraordinary fiscal and monetary intervention, a weaker dollar under the Miran doctrine (1:06:53), and a surge in offshore dollar liquidity that ultimately funds a major resources and emerging-market boom (1:09:56). Marcus’s caution on timing is the Wile E. Coyote problem — the fundamentals are already off the cliff, but the fall is psychological. Play it as Garran prescribes: on a relative basis now (short/avoid tech, rotate to value, resources, and EM), and on an absolute basis gradually, into the bottom (1:12:14).
Investment Focus
This is the most analytically complete bear case available this week, spanning the technical, commercial, financial, and macro layers simultaneously. The investment template: (1) reduce passive exposure to the cap-weighted index — roughly 60% of the S&P now trades in correlation with the AI trade (1:20:51), so “diversified” passive holders are running one giant concentrated bet; (2) treat OpenAI as the weak link in the chain — a management and trust problem, no remaining lead over Anthropic or Google, and less capital efficiency — and expect its IPO disclosures to be the catalyst; (3) avoid the neoclouds, the front-end app layer, and enterprise-software exposure where AI is now the competitor and the moat is non-existent; (4) own gold and silver as the reflation hedge and rotate toward value, resources, and emerging markets as the post-crisis beneficiaries; (5) monitor the concrete triggers Garran and Marcus name — usage-based billing adoption, data-center and Oracle CDS, bank credit appetite, and the OpenAI/Anthropic S-1s — as the sequence that turns the psychological standoff into a repricing.
EP 2 - How AI’s Push Toward IPOs Became a Death Drive
Ed Zitron — EZPR Founder & CEO, “Better Offline” Host
Ed Zitron, in conversation with Louis Sykes on The Tech Report, argues the AI industry has entered a phase of “token austerity” as Uber, T-Mobile, Walmart and others discover they cannot measure either the cost or the return of large language model usage. With OpenAI and Anthropic having moved enterprise customers from subsidized subscriptions to token-based billing, the real cost of AI is finally surfacing — and Zitron’s verdict is that two structurally unprofitable companies are being marched toward IPOs that function as exit liquidity. His framing: if OpenAI or Anthropic fails to go public, “it’s bedtime for this whole thing.”
Actionable Bullet Points
Token Austerity Has Begun — and the Caps Will Keep Dropping: Zitron’s central observation is that because no one can measure the cost of a single unit of AI work across harnesses like Claude Code or Copilot CLI, no one can measure its return on investment — which by definition makes it a thing without an ROI. Uber said it struggles to justify its AI spend against measurable return; T-Mobile capped usage at 2K per month and saw staff revolt when an earlier limit went as low as $30; Brex, now owned by Capital One, capped roughly 2K on OpenAI’s Codex just as OpenAI’s free-Codex trial periods expire. Zitron predicts the caps fall further — 2K to 1.5K to 1K and then a question mark — because the companies imposing them genuinely do not know the right number. Treat the spread of enterprise token caps as the leading indicator that the subsidy era is closing.
The Move From Subscriptions to Token Billing Is the Detonator: The industry, Zitron argues, conned the market by offering subscriptions that let users burn $5,000 of compute on a $200-a-month plan — meaning every journalist who reviewed AI was using a product that will not exist at real prices. His analogy: a cab service that charged point-to-point suddenly switching to per-mile billing, except the jump is less like Uber raising fares and more like a $10 ride becoming a $500 ride. GitHub Copilot users, he notes, are already finding they cannot do the things that excited them once they pay the true cost. The predictive read is that demand built on subsidized pricing evaporates when real costs arrive. Watch the enterprise billing transition as the catalyst that resets perceived value.
The Code Bases Are Built on Sand — More Code Is the Problem, Not the Proof: Zitron contends that years of pressure to “use AI to write as much code as possible” have left companies unable to simply rip the tools out, because AI wrote so much of the code that no one understands how it works — citing conversations indicating the practice made code bases worse at firms like Zillow. AI code is verbose and written without intention, and Anthropic’s own boast of shipping eight times more code is, in his framing, a liability: more code means more to review, and review at that volume requires another LLM that can hallucinate. He points to visible harms — AWS outages tied to an Amazon AI coding tool, an AI tool that deleted a user’s database, and Claude Cowork deleting a user’s photos in its first week. Expect internal uproar as firms confront the cleanup cost; do not equate lines of code with productivity.
The IPO Push Is Exit Liquidity, Not Strength: Zitron is unequivocal that OpenAI and Anthropic are unsustainable, unprofitable, and not ready for public markets — describing their press-shared financials as deliberately manipulated and non-GAAP. The reason to list now, he argues, is that both have raised at valuations near the trillion-dollar mark that make further private rounds difficult, so the public markets become the exit. He notes Daniela Amodei cited the need for “more access to capital” even after Anthropic raised $75 billion in three months, and points to a Broadcom-Google-SPV structure in which chips are routed through a special-purpose vehicle and leased to Anthropic, with Anthropic borrowing against it — and Anthropic declining to share financials with some prospective lenders. The prescriptive takeaway: treat the OpenAI and Anthropic S-1s as the moment of truth, because if either fails to list, the financing chain that depends on perpetual growth loses its keystone.
AI Is Not Causing the Job Losses — and the Narrative Is Self-Serving: Zitron argues there is no data showing LLMs are replacing jobs; the technology is weakening practitioners, not displacing them. Youth unemployment, he contends, traces to 2021–22 overhiring in the zero-rate era and a decades-long breakdown of mentorship, compounded by automated HR systems that obscure rather than enable hiring — he cites a posting demanding five years of experience with “AI agents,” a term roughly two years old. His blunt conclusion is that anyone claiming AI is taking jobs is doing so to get you to buy or invest in something they are positioned in. With Walmart already echoing the Uber cost-cap story, expect more such headlines — and discount the “AI took the jobs” framing accordingly.
Investment Focus
Zitron’s framework is the most forceful inside-out account of why the unit economics break. The investment template: (1) track the spread of enterprise token caps (Uber, T-Mobile, Brex, Walmart) as the clearest real-time signal that subsidized demand is being withdrawn; (2) treat the subscription-to-token-billing transition as the repricing event that resets how customers value the tools; (3) separate the semiconductor and infrastructure narrative from the model-company story — the former is being built on the assumption that demand never slows, which requires three or four more OpenAI/Anthropic-sized buyers that do not exist; (4) regard the OpenAI and Anthropic IPOs as exit liquidity and the single most important binary for the whole complex; (5) heavily discount the “AI is replacing labor” narrative when it comes from anyone with a position to sell.
EP 3 - Is the “Geriatric” Bull Market in Trouble?
Vincent Deluard — StoneX Director of Global Macro Strategy
Vincent Deluard, director of global macro strategy at StoneX, argues the market is at a tipping point — and, unusually for him, he is now worried about growth. His thesis: the “K-shaped” economy has developed a “K within a K,” with the high end itself splitting as the college-educated millennial cohort that has carried consumption finally cracks. He sees the big summer IPOs as a likely last hurrah, a Fed boxed in under new chair Warsh, and a liquidity tipping point where the yen, oil, and the 10-year all sit on the edge at once.
Actionable Bullet Points
The “K Within a K” — the High End Is Now Splitting: Deluard’s central new insight is that the high end of the K-shaped economy is itself fracturing. At the top, people are cashing out at the big IPOs into generational wealth while AI scientists collect $10 million bonuses to switch firms (7:13); below them sits the cohort that has actually carried the economy — college-educated millennials earning around $300,000 in big cities, white-collar, able to afford a house — and that cohort is now showing stress (7:30). The savings rate collapsed to roughly 2.5% in April (3:31), tax collection has fallen since the war (Q1 income-tax growth of close to 10% dropping toward 3–4%) (5:56), credit delinquencies outside mortgages are at record highs, and 401(k) hardship withdrawals are rising (8:17). His predictive call: the market breaks when this cohort breaks, and the tipping point arrives in summer or fall.
Permanent Stimulus Is Narrowing to Corporate Cronyism: Deluard frames the US as living in “permanent stimulus” — a 6–7% deficit even with unemployment low and stocks at records — but argues the stimulus is narrowing from the broad COVID-era payments toward a roughly $1 trillion shift from consumers to “corporate cronyism” (13:07). The three remaining channels: the investment channel, where 100% capex expensing feeds the data-center buildout that benefits few and angers many (11:00); tariff refunds, where the government is now paying out more in refunds than it collects in tariffs even as companies keep the price increases they already passed through (12:00); and a coming $1.5 trillion Pentagon blowout later this year (13:00). The prescriptive read: this redistribution upward compounds the political tension heading into the midterms.
We May Already Be in Stagflation and Not Know It: Asked whether disaving lower quintiles already mean stagflation, Deluard agrees — “we just don’t know it” (13:50). His argument is that we live in a nominal world and the post-COVID surge in nominal growth is, at least in part, an inflation illusion (15:10). He likens inflation to cocaine: the first phase feels great as margins and multiples expand because costs are recorded at historical levels, and the hangover arrives in the second wave (15:26). Inflation, he notes, has come in roughly five-year waves (1942/47/52, 1969/74/79); the current wave built in 2021–22, which puts the second-wave “hangover” arriving now (16:25). Position for the realization that reported growth has been part-inflation all along.
Warsh Inherits a No-Win Fed: Deluard expects new chair Warsh to regret having campaigned for the job, with board members ready to backstab him and Powell staying on “to enjoy the show” (17:21). Although appointed to cut rates, Warsh’s hands are tied: if two or three more bad inflation reports push the two-year yield well above the funds rate, cutting into a market that wants a hike would only steepen the curve and worsen the mess (18:24). Deluard notes new Fed chairs historically meet major sell-offs — Greenspan into the late-1980s crash, Bernanke into the GFC — and that with growth slowing and inflation rising, “whatever you do is going to be the wrong choice.” Do not position for rate cuts.
The Liquidity Tipping Point — Yen, Oil, and the 10-Year All on the Edge: Deluard identifies key levels across markets sitting simultaneously on a knife’s edge: the yen at 160, oil at $80 on the 12-month contract, and the 10-year at roughly 4.6–4.7% (19:50). The dangerous scenario is oil breaking $80, which pressures the currencies of Asian oil importers (Japan, South Korea) and forces them to sell US assets to afford energy, pushing yields higher and tipping the stock market over the summer (20:57). Against that fragile backdrop he reads the rush to list — the big three at a combined $4 trillion-plus, low float, and NASDAQ’s “3x float” rule plus fast-track index inclusion (24:00) — as eager insiders dumping shares into 401(k)s. Treat very eager sellers as a warning, not a green light.
Investment Focus
Deluard’s framework is the week’s most granular read on consumer and liquidity stress. The investment template: (1) hold cash as genuinely attractive now that cuts are off the table and one or two hikes are possible — even more so in Europe; (2) own emerging markets through old-school, brick-and-mortar exposure (Brazilian banks, commodities, and miners with real rates of 6–7%) rather than EM indices now dangerously concentrated in Korean and Taiwanese electronics; (3) look at the laggards — cheap Chinese green-technology names after a near-decade bear market — and check exactly what any EM index actually holds; (4) be patient on Brazil through the election noise, since the political-economic trend across Latin America (El Salvador, Argentina, Chile, Colombia) is favorable and “politicians are never as great as hyped or as bad as feared”; (5) watch the yen-160 / oil-$80 / 10-year-4.7% tipping levels and the IPO float mechanics as the signals that turn a few good weeks into the fall top.
