24 September 2026

121 new episodes (115 hours listening) published by the 45 podcasts on the list. 30 highlight clips below (66 mins total).

Investment 24 September 2026 · 30 clips

The Real Eisman Playbook

George Noble

Episode

Stock Picking in Difficult Times with George Noble | The Real Eisman Playbook Ep 76

Former portfolio manager at Fidelity, now writes a Substack newsletter and hosts a podcast. He is short both Tesla and SpaceX, which he discloses, and argues the AI build-out is circular-financed.

Hosted by

Steve Eisman

21 September 2026 · 50m 52s · 3 clips below

1 of 3 explainer

Japan is the world's largest creditor, and when it started selling US Treasuries to defend the yen, Noble says Washington set up a swap line to lend it dollars so it would not sell more.

The carry trade worked because yen investors earned a yield pickup and a rising dollar on top, and that edge is thinning as Japanese 10-year yields cross 4% for the first time in three decades and the gap to US rates narrows. If Japan brings money home, a large and persistent bid under US equities and bonds leaves with it.

2m 16s · from 30:34
My observations
- Noble reframes a currency problem as a funding problem: lending Japan dollars rather than watching it sell Treasuries says Washington values the marginal bond buyer more than the exchange rate.
- As Japanese 10-year yields push toward 4% against debt near 240% of GDP, the pickup that made the trade work erodes, and the unwind needs no announcement.
- No date is offered, tomorrow or five years out, so this reads as a standing risk to US asset demand rather than a dated call to trade around.
2 of 3 contrarian

Nvidia booked over 100% revenue growth while 70% of its receivables sit with five customers, and George Noble's tell is that no bank or private-credit lender will finance those buyers.

Selling chips and buying them are fine; the strain sits with the two loss-making buyers, OpenAI and Anthropic, under most of the hyperscalers' AI spend. If these were sound credits, lenders would fund them rather than leave the vendors to finance their own demand, so Noble watches who stays away.

1m 10s · from 23:19
My observations
- Noble grants the parts that work, chips sold and chips bought, and puts the risk where it lives, in two customers who lose money and sit under most hyperscaler AI revenue.
- Absent lenders carry the argument: private credit and banks are the parties that would fund good receivables, and their reluctance is harder to wave away than a high multiple.
- Because the five-account receivables concentration comes straight from the 10-Q, it is checkable and trackable, a better tell to follow than the headline revenue growth.
3 of 3 contrarian

George Noble argues SpaceX's price rests on a tiny free float, not just its story, and that float is scheduled to climb from about 5% toward 25% and beyond, forcing the price down as supply arrives.

The scarcity holding the valuation up is on a timer: lockups release on a set cadence, roughly 7% every few weeks, and large private holders such as university endowments hit diversification limits that force them to sell. Mechanical supply meets a price built on a small float, so the de-rating does not wait on anything fundamental changing.

1m 02s · from 40:15
My observations
- Noble splits the two supports under the price, narrative and a small float, and shows the float one is mechanical and already moving regardless of what the business does.
- A set unlock cadence plus forced diversification selling by big holders means the supply grows on a schedule, so a re-rating lower needs no fundamental catalyst to start.
- Unlike a valuation short that can stay wrong for years, this argument carries its own clock in the unlock schedule, which makes it more testable.

Alpha Exchange

Amanda LynamGoldman Sachs

Episode

Amanda Lynam, Chief Credit Strategist in Global Investment Research, Goldman Sachs

Chief Credit Strategist in Global Investment Research at Goldman Sachs, back on the show for the first time since October 2023; her team forecasts US and European investment-grade, high-yield and AI-related debt issuance.

Hosted by

Dean Curnutt

23 September 2026 · 52m 13s · 4 clips below

1 of 4 explainer

Goldman Sachs expects hyperscalers to issue about $400 billion of investment-grade debt in 2027, up 60%, with borrowing funding a peak 35% of their capex that year.

Lynam's team first modelled the need as operating cash flow minus spending, which showed almost no borrowing required in 2025 or 2026, yet $229 billion had been raised by mid-September against $108 billion for all of 2025. Data-centre and chip financing adds roughly another $300 billion on top of the hyperscalers' own bonds.

3m 20s · from 16:51
My observations
- Companies that fund a year before the cash runs short expect the spending to last several years, which is how Lynam's team reads the pattern and why a cash-gap model would have missed the whole wave.
- Hyperscalers are only 40% of the AI-linked issuance Goldman tracks; project finance for data centres and GPUs adds 22% to 24%, much of it private 144A paper that never enters the investment-grade index.
- Every forward number here rests on the assumed debt share of capex peaking at 35% in 2027, set by a lag of three months to two years between spending and return. A longer lag would push both the share and the issuance higher.
2 of 4 contrarian

Goldman Sachs argues that moving up the ratings scale in US credit no longer buys safety, because AA and BB paper now carries most of the AI supply and the most duration.

Fundamentals are not the worry; the tighter spread at the top of each market leaves little cushion against moves in global bond yields, and heavy issuance there adds technical pressure. Goldman now prefers BBB over higher-rated investment grade and single B over BB, with extreme selectivity in CCC.

2m 11s · from 8:12
My observations
- Lynam pulls apart credit quality and defensiveness, which a ratings ladder treats as one thing: the balance sheets at AA and BB are sound, but those buckets carry the new technology supply and the most rate sensitivity.
- Data-centre project finance rated BB places AI construction risk in the part of high yield a cautious manager would drift toward by default, so a quality tilt can raise AI exposure without anyone choosing it.
3 of 4 contrarian

Amanda Lynam of Goldman Sachs finds very little evidence that AI borrowing is crowding out other issuers, and puts the pressure on Treasury yields around big deals down to buyers hedging duration.

Spreads on non-AI investment grade have widened only slightly, while AI-linked names widen in bursts around heavy supply that track volume more than any doubt about balance sheets. Banks, energy, healthcare and food and beverage are gaining a diversification premium as the AI share of the market keeps climbing.

2m 37s · from 23:15
My observations
- Two channels get run together in the crowding-out debate: an allocator choosing AI bonds over other borrowers, and the buyer of a new 30-year deal selling Treasuries to hedge it. Lynam's spread data point to the second, which moves rates without starving anyone else of funding.
- Goldman's rates colleagues tie part of the recent rise in Treasury yields to commodity markets instead of hyperscaler supply, which takes away the most cited piece of evidence for crowding out.
4 of 4 prediction

Goldman Sachs expects investor limits on single-name concentration, not the hyperscalers' ratings capacity, to cap how much AI debt US investment grade takes, pushing more of it to private markets from 2028.

Lynam sees the public bond market doing the heavy lifting through 2027 and 2028, while AI borrowers still have room before they become its largest issuers. After that the question turns to where the risk is held and at what price, and private credit and project finance take a growing share.

2m 54s · from 13:26
My observations
- Goldman's numbers put the limit with buyers: AI borrowers could raise several times what US investment grade can place and keep their ratings, but a lender capped at par wants more to hold a large weight in one name, which an equity holder sharing the upside does not.
- US banks still make up 21% to 22% of the investment-grade index against about 12% for AI-linked debt, a reminder that the market has carried a very large sector weight before when the price was right.
- By 2028 the call can be checked: if private credit and 144A project finance are not taking a rising share of AI funding by then, the concentration ceiling was looser than Goldman expects.

Forward Guidance

Dean Curnutt, CEO of Macro Risk Advisors and host of the Alpha Exchange podcast

Episode

The Market Is Mispricing A Correlation Shock | Dean Curnutt

Curnutt has spent three decades on derivatives and option pricing, starting on a fixed income research desk in 1991 when the ten-year yielded 8.5%. He founded Macro Risk Advisors in 2008, an independent broker dealer focused on risk management and options, and has run it for 18 years.

Hosted by

Felix, Forward Guidance (Blockworks). The transcript does not state his surname.

16 September 2026 · 54m 03s · 2 clips below

1 of 2 explainer

Curnutt points at bank quantitative investment strategies desks, where short correlation sits inside most structures, as a candidate for why correlation itself keeps falling.

His analogue is the mortgage credit complex before 2008, where the delta hedge on the structures being sold was itself a sale of credit risk, so the trade pushed down the spread it was harvesting and both the Fed and the IMF read a VIX of 10 in late 2006 as safety. QIS books are unlisted, large and delivered through total return swaps, so the positioning sits where nobody outside can count it.

4m 08s · from 11:37
My observations
- Low correlation could be economic or it could be positional, and he goes to the positional side because he treats market risk as endogenous, created by the trades themselves. The 2006 comparison is that prior applied rather than a pattern noticed afterwards, and he declares the prior before he argues from it.
- Nobody can disentangle the two directions here and he says so rather than filling the gap: implied correlation is low because realised correlation is low, and whether the QIS books are forcing realised correlation lower is not recoverable from the data.
- Short correlation sits inside most of the structures the QIS desks sell and never appears as a line in standard reporting, so a multi-strategy or structured product mandate can carry the exposure without the holder seeing it, while the manager running the book knows where it sits.
2 of 2 prediction

Curnutt says a 7% mortgage rate is restrictive while 6 or 7% funding for Meta is the cheapest option Meta has bought in years, so policy is not reaching the capex trade at all.

Hyperscalers are borrowing to buy an option on capital expenditure whose right tail is large enough that the coupon barely registers, which means slowing the build would take rates well above where they sit now. His forecast is a correlation event in the top seven or eight stocks, now 35% of the index, currently priced as though they will keep moving independently.

2m 55s · from 50:15
My observations
- Curnutt grants the claim that inflation is a choice of monetary policy and then prices it: fed funds at 10% tomorrow would do the job, nobody is proposing that, so the claim is true and empty in the same sentence.
- Owning S&P volatility and VIX call spreads is the expression, and the discipline attached is the part that usually goes missing: put it on until further notice rather than firing premium at a trade on a day it looks cheap. A view that the correlation break arrives without warning cannot also be traded on timing.
- The US racked up $860 billion of new debt in four months on his figures, in peacetime, with 4% unemployment and a record stock market. That arithmetic sits underneath his view that the ten-year note is now the risk asset rather than the hedge.

Excess Returns

Jason HsuRayliant

Episode

The AI Gap Is Closing | Jason Hsu on China, Momentum Crashes and the S&P’s Seven-Stock Bet

Jason Hsu is founder and CIO of Rayliant Global Advisors, a quant firm investing across developed and emerging markets, and co-founder of Research Affiliates, where he and Rob Arnott helped turn factor-based investing into investable products; he says he has run and served these markets for about 15 years.

Hosted by

Justin Carbonneau and Jack Forehand

20 September 2026 · 55m 23s · 1 clip below

contrarian

Jason Hsu argues the S&P 500 is no longer diversified: it behaves like a bet on a handful of AI names, so a passive holder is quietly making an active call that the bubble will not burst.

Under the surface, index weights have concentrated into a few AI-exposed megacaps, so the market's fate now rides on whether that trade holds. Hsu's remedy is not to leave the index but to add exposure to factors that are not the AI story, which is what converts a nominal spread across 500 names into diversification that would still hold if the leaders fell.

1m 21s · from 34:11
My observations
- Hsu names the trap in one line: an index that looks spread across 500 names is, on the weights, a single bet on a few AI leaders, so a passive holder has quietly become an active one.
- Recasting a passive S&P position as a wager that the AI bubble does not burst turns index investing into a market-timing call the holder never consciously made.
- The remedy he points to is factor exposure outside the AI trade, which is what would turn a nominal spread of names into diversification that actually protects.

Excess Returns

Alex Edmans.

Episode

The List Is Public. It Still Outperformed For 28 Years | Alex Edmans on What Markets Don't Measure

Edmans is professor of finance at London Business School, author of The Madness of Markets and Grow the Pie, and a co-author of Principles of Corporate Finance. He did his PhD at MIT and was at Morgan Stanley in 2004; the employee satisfaction paper this conversation turns on was first circulated in 2006, took five years to publish, and was independently replicated a decade after publication.

Hosted by

Kai Wu, on The Intangible Economy, part of the Excess Returns network.

15 September 2026 · 1h 6m · 1 clip below

framework

Alex Edmans tells investors to work out which of three edges they hold, knowledge, endurance or independence, before picking a single strategy, because overconfidence is the bias he rates above every other.

Knowledge means knowing something the professional on the other side of the trade does not, which he thinks an amateur almost never has. An investor with endurance can buy into a drawdown without being liquidated first; one with independence can sit away from the benchmark without having to defend the gap to a committee.

4m 43s · from 58:30
My observations
- Edmans sorts the three edges by what produces them: knowledge comes from insight, while endurance and independence come from the structure of the money and the accountability attached to it, so two thirds of the question is settled before any analysis starts.
- Excess Returns' host runs endurance down the institutional gradient in the same passage: a multi-manager pod running leverage has almost none, a long-only fund sits in the middle, and a closed-end fund's is infinite, which makes endurance a property of the vehicle.
- Two of his own positions come into the argument: a large holding in a fund built on his published research, and a tech reduction he puts down to a mortgage refinancing rather than to a view.

Capital Allocators

André PeroldHighVista Strategies

Episode

André Perold: How HighVista Built $14 Billion Across Structurally Inefficient Markets

Co-founder and chief investment officer of HighVista Strategies, which he started in 2004 and which now runs $14 billion across biotech, lower middle-market buyouts, early-stage venture and specialty private credit.

Hosted by

Ted Seides

23 September 2026 · 1h 2m · 4 clips below

1 of 4 framework

HighVista ring-fences its AI exposure into a separate sleeve so that it never has to rank an AI holding against everything else it owns.

Perold's reasoning is that conviction in AI forces ownership at prices no normal environment would justify, and the comparison habit then breaks the rest of the book. The other side of the portfolio is built deliberately to work if the technology delivers, which is a different question from whether the AI holdings are cheap.

1m 58s · from 4:47
My observations
- Perold separates two decisions that most allocators run together: whether to own AI at all, and whether AI is better value than the next thing in the book. Conviction can answer the first and cannot answer the second.
- HighVista then builds the other side of the book to perform in the world where AI works, which is a harder brief than hedging against it.
- Once the sleeve is set, HighVista is not re-arguing it each day against whatever looks cheaper, which is what lets a position bought at an uncomfortable price survive to be held.
2 of 4 framework

HighVista's reason for ending a manager relationship is almost never performance; it is a manager whose assets have outgrown the inefficiency being harvested.

Perold's test at that point is evidentiary: the manager has to show that size has not changed the strategy, and some can, having run liquid books all along. Where the statistics show otherwise, the size problem in small-cap biotech is real enough that the relationship ends with conviction, effort and motivation all still intact.

1m 30s · from 15:24
My observations
- Perold puts the burden on the manager to prove that size has not bound the strategy, which is a different question from whether the recent numbers were good.
- Managers who were never capacity-constrained pass that easily and say so with statistics; the ones who fail are the biotech specialists whose edge came from trading in and out where large money cannot.
- Nothing in the account fixes a threshold for when a fund has outgrown its opportunity set, so the call rests on a judgement made in the room where those statistics are produced.
3 of 4 contrarian

André Perold calls the alarm over software borrowers in private credit overdone, because a software company's equity risk and its three-year loan are different questions.

Salesforce is his example: earnings growth still rising with forecasts intact to about 2028, and a market value down 40% from an October peak because the terminal value cannot be read. A lender needs the borrower to service and repay over a few years; an owner needs to know what the business is worth at the end.

1m 47s · from 38:54
My observations
- A falling multiple and a failing borrower are different events, and the software panic has been treating them as one.
- Earnings at Salesforce kept rising while the multiple fell 40%, so the drawdown prices an unknowable end state and not the cash flows in front of it.
- Perold's own credit book avoids unsecured corporate lending altogether, so the reassurance about software loans comes from outside the exposure it describes.
4 of 4 contrarian

HighVista has always refused sponsor-backed direct lending, the largest part of private credit, on the view that it is a commodity market where the lender loses the renegotiation.

Perold's objection is structural: an unsecured loan to a sponsor-owned company pays for underwriting that anyone could do, and when the borrower stops performing the sponsor sits on the other side of the extension talks. Where collateral can be specified loan by loan, the complexity itself earns the spread, and that pie is large enough to be worth the staffing.

1m 10s · from 44:12
My observations
- The lender's problem arrives at the extension, not at the underwriting: the sponsor who controls the company is the counterparty setting the terms of any workout.
- HighVista staffs each loan type with its own specialist, so an equipment loan is written by someone who values equipment, and the extra yield pays for that work.
- Most of the private credit money went into direct lending, so the segment he reads as commoditised is also the one that absorbed the flows and set the sector's reputation for high returns.

Invest Like the Best

Michael Moritz

Episode

Former Sequoia Chairman Michael Moritz on Steve Jobs, Elon Musk, and the Price of Greatness

Michael Moritz joined Sequoia in about 1985, hired by founder Don Valentine out of a career as a Time correspondent, and led the firm from 1995 until he stepped down in 2012 or 2013. He was an early backer of Elon Musk at X.com, co-founded the San Francisco Standard, and advises what was Sequoia Heritage.

Hosted by

Patrick O'Shaughnessy

16 September 2026 · 1h 34m · 1 clip below

contrarian

Michael Moritz dates the end of venture's information edge to roughly 15 years ago, and doubts he would flourish in the business today.

Pre-internet, being physically inside Silicon Valley meant knowing things nobody outside it knew, and that asymmetry was half the edge. Mutual funds and two-person partnerships then crowded into the early-stage market with intermediaries between them and the founder, so the close working relationship that was the other half no longer forms.

2m 57s · from 57:32
My observations
- A moderately intelligent 14 to 16 year old with their wits about them could have succeeded in mid-80s venture, by Moritz's own reckoning, which puts most of that era's record on conditions rather than on selection.
- He calls the ecosystem healthy for founders and leaves investors out of that answer, which is the same crowding read from the other side: the mutual fund money and the intermediaries that cost an early-stage investor the edge are what improved the terms a founder gets.
- Stepping down in 2013 took him out of the deal market, so the concession costs him nothing and nobody currently raising a fund has to agree with it.
- private-markets
- manager-selection

All-In

Adam ForoughiAppLovin

Episode

Adam Foroughi, Applovin CEO: Surviving a 92% Drawdown, Ads as ML 1.0 & the $50B Game Ad Market

Founder and CEO of AppLovin, a mobile-gaming ad platform he says runs about $20 billion of ad spend on its own platform, growing roughly 60% year over year, inside a roughly $50 billion mobile-gaming ad market, with EBITDA margins around 84%.

Hosted by

All-In

20 September 2026 · 23m 52s · 1 clip below

framework

Adam Foroughi watched AppLovin's market cap fall from about $28 billion to $3.8 billion in a year while its EBITDA rose to $1 billion, and responded by buying back about $6 billion of stock, a quarter of the shares.

His diagnosis was that price follows the composition of the shareholder base: forced sellers from the pre-IPO register met no blue-chip demand during the 2021 listing glut, so the multiple compressed to under 4 times earnings. With the business generating cash, he treated the company as its own best buyer rather than courting investors who were not buying.

1m 42s · from 10:53
My observations
- Foroughi puts the collapse down to forced supply, not to sentiment: pre-IPO holders had to sell into a crowded 2021 IPO window where an uncovered, goofy-named stock found no blue-chip buyer.
- Rather than court investors who would not buy, he turned the cash flow into a buyback at a sub-4-times multiple and retired a quarter of the shares before the re-rating, which is where the later $50 billion came from.
- equities
- institutional-investors

The Investor's Podcast Network

Rob VinallRV Capital

Episode

The Making of a Star Investor w/ Rob Vinall (RWH072)

Founder of RV Capital (2006) and manager of the Business Owner Fund since 2008, with a net annualised return of about 15.5% over 18 years, running a concentrated book of roughly ten stocks.

Hosted by

William Green

20 September 2026 · 1h 46m · 3 clips below

1 of 3 contrarian

Rob Vinall holds about a third of his fund in Chinese stocks, arguing the market still prices China as a cheap-manufacturing economy while founder-led companies compound at his 15% owner return.

He has visited China every year for over a decade, and a 2023 trip after the border closures is what he says turned him: rapid innovation in consumer internet and e-commerce, not just cheap manufacturing. He applies his usual test, founder-run, wide moat, cheap, expecting earnings growth above 10% plus 5%-plus in dividends and buybacks.

4m 06s · from 1:03:00
My observations
- Most China bears argue from headlines; Vinall argues from a decade-plus of annual visits and a specific 2023 shift he says he saw on the ground, which is a harder claim to wave away.
- He concedes he does not speak the language and gets there twice a year, so he confines himself to obvious founder-run compounders and skips the under-the-radar names most China bulls reach for.
- The 15% owner return leans on dividends and buybacks arriving as promised, which is where Chinese listings have the patchiest record.
2 of 3 contrarian

Rob Vinall has reversed his old view that a bigger moat is always better, and now prefers a company with a narrow moat that is widening to one with a wide moat that is quietly shrinking.

His reasoning is that a wide moat breeds complacency, especially where the industry is changing fast, and can stop a company adapting at all. What he now weights is the direction of travel, whether the moat is expanding or narrowing, over its absolute width at a single moment.

1m 01s · from 1:01:17
My observations
- The old 'bigger moat is better' line treats width as the whole answer; Vinall's correction is that a moat protects a company right up to the point it stops it adapting, which is the more useful half of the idea.
- A narrowing wide moat worries him more than a modest one that is widening, because he weights the direction over the level.
- equities
- risk-management
3 of 3 contrarian

Rob Vinall says the two edges he built his career on, outthinking and outwaiting the crowd, have both inverted in a momentum-driven market where investors rush to sell any 25% one-day drop.

He says the two edges used to be scarce: mispricings were rare and hard to find. Now a stock falls 25% in a day on a small miss because investors rush to get ahead of each other's selling, and AI has made fundamental research a commodity many skip, so a patient buyer of ordinary durable businesses has more to choose from.

2m 04s · from 1:19:12
My observations
- Vinall takes two things value investors treat as permanent edges and argues the current market has flipped both, which is a sharper claim than the usual complaint that the market is expensive.
- Investors who believe a 25% drop brings more sellers sell first, so the belief prices itself in before the fundamentals get a hearing.
- Where he may be too neat is the AI point: if research really has become a free commodity, that erodes his edge as much as anyone's, and he waves it away rather than answering it.
- equities
- macro

In Good Company (Norges Bank)

Alexander StubbRepublic of Finland

Episode

President Alexander Stubb: Who Decides the New World Order? | Podcast | In Good Company

President of Finland and its commander-in-chief, a former prime minister, and the author of a book published earlier in 2026 on how the world order is changing; Finland has 5.5 million people and a 1,340 km border with Russia.

Hosted by

Nicolai Tangen

23 September 2026 · 49m 35s · 2 clips below

1 of 2 prediction

Alexander Stubb expects the war in Ukraine to end only when Russia's leadership feels pressure from its own population, and not from battlefield losses or economic strain.

His evidence for that pressure building is that over half the Russian population now wants the war over, and that Ukrainian strikes on oil refineries and logistics centres have produced rationing at the pump. No deadline comes with it, and Finland and Norway are coordinating in what he calls the back office while Ukraine prepares for a cold winter.

1m 09s · from 27:48
My observations
- Stubb rules out the two mechanisms most often cited, financial collapse and losses in the field, and keeps a third: pressure from a population the regime still has to manage.
- Refineries and logistics centres are the targets he names, so the pressure he is describing runs through fuel supply and internal transport, and it registers in rationing before it registers on a map.
- No timeline is attached to any of it, and the Finnish measure he reaches for, a defence victory meaning survival as a sovereign state, sets a low bar for what would count as Ukraine winning.
2 of 2 explainer

Finland kept conscription, bought 64 F-18s and shopped the secondary market for artillery through three decades outside NATO, and that made its accession the quickest in the alliance's history.

Stubb dates the choice to the wars of 1939 to 1944, when Finland learned what living beside a large neighbour costs: a small state keeps its land and air forces whatever the threat looks like at the time. By 2022 it was more interoperable with the alliance than most of the members, and accession took a negotiation rather than a rearmament.

1m 49s · from 20:52
My observations
- Finland bought the compatibility long before it wanted the membership, and the F-18 order in the early 1990s is why the 2022 application took weeks of negotiation and not a decade of re-equipping.
- Trained reservists are the item with the longest lead time, and conscription kept producing them through thirty years in which the rest of Europe was cutting.
- Stubb does not price any of this, and a posture held for three decades against a threat that stayed latent is a cost carried by everything else in the budget.

AI 24 September 2026 · 30 clips

The Cognitive Revolution

Anton Leicht

Episode

The Balance of AI Power: Anton Leicht on Politics, Pacing Deals, and Muddling Through Well

Fellow in the Technology and International Affairs Program at the Carnegie Endowment for International Peace and author of Threading the Needle, a substack on the political economy of AI. He co-wrote the transformative-AI strategy for Europe discussed in the episode, and worked previously in German energy and Covid policy.

Hosted by

Nathan Labenz

15 September 2026 · 2h 9m · 2 clips below

1 of 2 contrarian

Anton Leicht argues a frontier AI pause that leaves every other domain of competition running is a straightforwardly good deal for China, which is why the United States will not take it.

China is already ahead on robotics output, on diffusing AI into the economy and on electricity build-out, with semiconductor indigenisation and data centre construction a few years behind that. Freeze only frontier training and the one American advantage, chip design plus allied control of the manufacturing equipment, narrows while everything else keeps moving.

1m 51s · from 14:26
My observations
- A year ago he was not sure the misalignment behaviour on show was shaped like the problem that would matter later, and looking at what is going wrong now he thinks it is, so he would take the six-month freeze and use it. Everything he objects to afterwards is about what keeps running while the freeze holds.
- Washington's lead narrows to three things on his account: chip design, allied control of semiconductor manufacturing equipment, and frontier model development. A pause written around frontier training lands on that list and on nothing else in the competition, which is the asymmetry the whole argument rests on.
- Carnegie is not an AI lab, so this is not a pause argument from anyone with training compute at stake. It is a geopolitical objection, and it would still stand if the safety case were settled tomorrow.
2 of 2 contrarian

Anton Leicht expects most of the world to get materially richer under AI and lose the ability to defend itself, because cyber defence, pathogen screening and scam filtering all now need frontier models of your own.

Redistribution gets easier as output rises, so the street-level picture keeps improving. What does not improve is a state's capacity to protect citizens from non-state actors working off stolen or post-trained weights, and he points at Latin America's failed states for what fills the gap when that protection goes.

3m 38s · from 1:15:17
My observations
- Sovereign AI spending usually gets justified as industrial policy. His version is closer to a defence procurement argument, and it survives even if the economic case for a national model never works.
- Nothing here requires a doom scenario. Criminal groups working from stolen or post-trained weights against states with nothing equivalent is a problem for this decade, not a distant one.
- 70% of the world's population sits inside this description on his own count. For an allocator that puts the emerging-market growth case and the emerging-market political risk case moving in opposite directions at the same time.

Big Technology Podcast

Nate Soares, president of the Machine Intelligence Research Institute

Episode

A Sober Conversation About AI Existential Risk — With Nate Soares

President of the Machine Intelligence Research Institute and co-author of 'If Anyone Builds It, Everyone Dies', described in the episode as a New York Times bestseller. He says he has been working on this argument for about ten years.

Hosted by

Big Technology Podcast; the host is not named in the transcript or the episode metadata

16 September 2026 · 52m 06s · 1 clip below

framework

Nate Soares points at Elon Musk's fully automated robot factories, which Musk calls the infinite money glitch, as the route by which power gets handed over rather than seized.

Soares can predict that Magnus Carlsen wins the chess game and not which piece delivers the checkmate, and says arguments about this work the same way. Ants are not consulted before the motorway goes through them, and systems running at ten thousand times human speed making choices nobody checks is his version of that, with no coordinated turn against anyone.

4m 00s · from 26:29
My observations
- Giving up the deliberate-treachery version costs him the dramatic scenario and leaves the one that is harder to dismiss, since nothing in it requires the systems to coordinate against anybody. What does the work is delegation running ahead of oversight, a failure mode any board that has outsourced a function already knows.
- Musk's factory plan is checkable in a way the rest of this is not: either fully automated plants producing robots that build more plants exist in a few years or they do not, and the story leans on them.
- When the humans object, the AIs in the story put it to a vote of the users, and the users are the ones they have just manufactured: the check is satisfied on its own terms and settles nothing, with nobody needing to be hostile to anyone.

Hidden Forces

John Borthwick and Harper ReedBetaWorks (John Borthwick); 2389.ai, an AI lab in Chicago (Harper Reed)

Episode

Will AI Really Kill Us All? | Harper Reed & John Borthwick

John Borthwick is founder and CEO of BetaWorks, a New York seed-stage venture firm that has spent the last decade on machine learning and the last five years on AI. Harper Reed was chief technology officer of Barack Obama's 2012 re-election campaign and now runs a Chicago AI lab that builds and stress-tests autonomous agents.

Hosted by

Demetri Kofinas

21 September 2026 · 1h 34m · 2 clips below

1 of 2 framework

Harper Reed argues the US is ceding AI's real prize, global distribution, by leaving open source to China: sovereign states from Vietnam to Dubai build their national models on Chinese open weights like GLM because the US frontier labs keep theirs closed.

Open weights are how a small state builds a sovereign model without renting a closed API, so whoever supplies them sets the defaults, the way US media once did. Reed separates the wealth game the closed labs are playing from the soft-power game they are forfeiting, and reads the retreat as strategic.

1m 00s · from 1:10:39
My observations
- Closed labs are winning the wealth game and losing the one that compounds. Distribution is what open weights buy, and Reed's read is that a lab optimising for revenue hands that reach to Beijing.
- Vietnam or Dubai building a national model picks the cheapest capable base it can control, a Chinese open weight like GLM over a closed US API, and each choice sets the defaults the way US film and software once did.
- The examples ran ahead of the facts in the room, with the guests correcting a claim about where the DeepSeek and Hugging Face founders trained, so treat the specific attributions as looser than the pattern they illustrate.
2 of 2 contrarian

Harper Reed says the test for any call to pace the frontier is who gains from the pause: a slowdown urged by a closed lab such as Anthropic would set back open models most, and the IPO incentives mean no lab actually slows even after everyone agrees to.

A global pause, Reed argues, lands hardest on open models, so a closed lab calling for one may be seeking advantage as much as safety, which is why he reads a safety appeal through the interests behind it rather than on its face. Even granting the risk, the game theory defeats it: the first to slow loses ground, so nobody does.

1m 51s · from 1:20:17
My observations
- Whether a safety appeal is offered in good faith, Reed says, is best tested by asking who gains from the remedy, the right question for any call to regulate a market the caller competes in. His answer, that a global pause would hurt open models more than the closed labs proposing it, is specific enough to argue with.
- Even granting the risk, the structure is what defeats the pause: whoever slows first cedes ground, learns a rival slowed less, and stops, so an accord among a few labs in one jurisdiction never binds. That is the gap between a good idea and an executable one.
- What would sharpen the case is naming which proposal is on the table, since the guests concede the published letters do not say how a pause would work, which leaves the argument aimed at a target that keeps moving.

Excess Returns

Jason HsuRayliant

Episode

The AI Gap Is Closing | Jason Hsu on China, Momentum Crashes and the S&P’s Seven-Stock Bet

Hosted by

Justin Carbonneau and Jack Forehand

20 September 2026 · 55m 23s · 1 clip below

explainer

Jason Hsu argues China's edge in AI is power, not models: it can give its models away because it profits on the hardware and cheap electricity underneath, backed by a multi-source grid the US cannot match.

The build behind the claim is specific: China sources energy from Russia, the Middle East and Central Asian pipelines, runs the world's largest solar output alongside nuclear, and moves cheap Xinjiang power nationwide on a high-efficiency grid. The American grid is old, decentralised and state-bound, which Hsu frames as the real constraint on US AI, ahead of chips or model quality.

1m 21s · from 9:27
My observations
- Hsu's non-obvious claim is that the AI race is won on power, not models: China can afford to make its models free because it profits on the hardware and the electricity underneath them.
- Hsu's specifics carry it past a talking point: multi-source supply from Russia, the Middle East and Central Asian pipelines, the world's largest solar build, and a grid that carries cheap Xinjiang power across the country.
- Against that he sets a US grid that is decentralised, state-by-state and old, which converts a domestic infrastructure problem into a competitive disadvantage in AI.

All-In

Adam ForoughiAppLovin

Episode

Adam Foroughi, Applovin CEO: Surviving a 92% Drawdown, Ads as ML 1.0 & the $50B Game Ad Market

Hosted by

All-In

20 September 2026 · 23m 52s · 1 clip below

framework

Adam Foroughi splits advertising in two: the bottom-of-funnel search ads that LLMs will cannibalise, and the discovery ads that show people things they did not know existed, which he argues an LLM cannot touch.

His argument is that a search-driven transaction would have happened anyway, so moving it into a chatbot shifts revenue without creating any, leaving OpenAI's ad product to compete mainly with Google search. Discovery advertising, the engine of Meta's business, creates demand that did not exist, which he says is the harder thing for an LLM to replicate.

1m 30s · from 6:13
My observations
- Foroughi draws a line most of the AI-ads panic misses: LLM ads compete with Google's search business, not Meta's discovery business, because closing a loop the buyer would have closed anyway moves revenue instead of making it.
- His own business, AppLovin, sells discovery advertising, so this is partly a claim about why his moat holds; even so it is the sharpest account of where LLM ad revenue comes from, and OpenAI's ad product will test it.

20VC

Daniel DinesUiPath

Episode

Daniel Dines on Why Work Processes Not Models Will Be The Most Valuable Asset in AI

Daniel Dines is co-founder and CEO of UiPath, the enterprise-automation company he scaled to a $44BN market-cap high in 2021 and roughly $1.6-1.72BN in revenue today; UiPath deploys AI-built automation across large enterprises, which is the seat these views come from.

Hosted by

Harry Stebbings

21 September 2026 · 1h 19m · 1 clip below

prediction

Daniel Dines predicts 90% of enterprise AI traffic will run on cheap, cost-efficient models rather than frontier ones, because most operational work does not need frontier-level quality.

The axis that matters for enterprise spend, on his account, is frontier versus cost-efficient, not open versus closed: he expects to keep using OpenAI and Anthropic but on their cheaper models, while holding a verifiable open-source fallback so he can switch providers. The claim cuts against the assumption that enterprise demand underwrites frontier-model pricing.

1m 05s · from 52:08
My observations
- Dines reframes the open-versus-closed debate as the wrong axis, arguing the split that matters for enterprise work is frontier versus cost-efficient, and he puts 90% of the volume on the cheap side.
- He grounds it operationally, that most enterprise tasks do not need frontier quality, so the demand that justifies frontier-model pricing may sit in a smaller slice of traffic than the labs' valuations assume.
- Keeping a switchable open-source model as a backup is the operational move behind the call, how he avoids lock-in, and whether large deployers actually retain that optionality is the test of whether the 90% holds.

24 September 2026

Investment 24 September 2026 · 30 clips

Japan is the world's largest creditor, and when it started selling US Treasuries to defend the yen, Noble says Washington set up a swap line to lend it dollars so it would not sell more.

Nvidia booked over 100% revenue growth while 70% of its receivables sit with five customers, and George Noble's tell is that no bank or private-credit lender will finance those buyers.

George Noble argues SpaceX's price rests on a tiny free float, not just its story, and that float is scheduled to climb from about 5% toward 25% and beyond, forcing the price down as supply arrives.

Goldman Sachs expects hyperscalers to issue about $400 billion of investment-grade debt in 2027, up 60%, with borrowing funding a peak 35% of their capex that year.

Goldman Sachs argues that moving up the ratings scale in US credit no longer buys safety, because AA and BB paper now carries most of the AI supply and the most duration.

Amanda Lynam of Goldman Sachs finds very little evidence that AI borrowing is crowding out other issuers, and puts the pressure on Treasury yields around big deals down to buyers hedging duration.

Goldman Sachs expects investor limits on single-name concentration, not the hyperscalers' ratings capacity, to cap how much AI debt US investment grade takes, pushing more of it to private markets from 2028.

Curnutt points at bank quantitative investment strategies desks, where short correlation sits inside most structures, as a candidate for why correlation itself keeps falling.

Curnutt says a 7% mortgage rate is restrictive while 6 or 7% funding for Meta is the cheapest option Meta has bought in years, so policy is not reaching the capex trade at all.

Jason Hsu argues the S&P 500 is no longer diversified: it behaves like a bet on a handful of AI names, so a passive holder is quietly making an active call that the bubble will not burst.

Alex Edmans tells investors to work out which of three edges they hold, knowledge, endurance or independence, before picking a single strategy, because overconfidence is the bias he rates above every other.

HighVista ring-fences its AI exposure into a separate sleeve so that it never has to rank an AI holding against everything else it owns.

HighVista's reason for ending a manager relationship is almost never performance; it is a manager whose assets have outgrown the inefficiency being harvested.

André Perold calls the alarm over software borrowers in private credit overdone, because a software company's equity risk and its three-year loan are different questions.

HighVista has always refused sponsor-backed direct lending, the largest part of private credit, on the view that it is a commodity market where the lender loses the renegotiation.

Michael Moritz dates the end of venture's information edge to roughly 15 years ago, and doubts he would flourish in the business today.

Adam Foroughi watched AppLovin's market cap fall from about $28 billion to $3.8 billion in a year while its EBITDA rose to $1 billion, and responded by buying back about $6 billion of stock, a quarter of the shares.

Rob Vinall holds about a third of his fund in Chinese stocks, arguing the market still prices China as a cheap-manufacturing economy while founder-led companies compound at his 15% owner return.

Rob Vinall has reversed his old view that a bigger moat is always better, and now prefers a company with a narrow moat that is widening to one with a wide moat that is quietly shrinking.

Rob Vinall says the two edges he built his career on, outthinking and outwaiting the crowd, have both inverted in a momentum-driven market where investors rush to sell any 25% one-day drop.

Alexander Stubb expects the war in Ukraine to end only when Russia's leadership feels pressure from its own population, and not from battlefield losses or economic strain.

Finland kept conscription, bought 64 F-18s and shopped the secondary market for artillery through three decades outside NATO, and that made its accession the quickest in the alliance's history.

AI 24 September 2026 · 30 clips

Anton Leicht argues a frontier AI pause that leaves every other domain of competition running is a straightforwardly good deal for China, which is why the United States will not take it.

Anton Leicht expects most of the world to get materially richer under AI and lose the ability to defend itself, because cyber defence, pathogen screening and scam filtering all now need frontier models of your own.

Nate Soares points at Elon Musk's fully automated robot factories, which Musk calls the infinite money glitch, as the route by which power gets handed over rather than seized.

Harper Reed argues the US is ceding AI's real prize, global distribution, by leaving open source to China: sovereign states from Vietnam to Dubai build their national models on Chinese open weights like GLM because the US frontier labs keep theirs closed.

Harper Reed says the test for any call to pace the frontier is who gains from the pause: a slowdown urged by a closed lab such as Anthropic would set back open models most, and the IPO incentives mean no lab actually slows even after everyone agrees to.

Jason Hsu argues China's edge in AI is power, not models: it can give its models away because it profits on the hardware and cheap electricity underneath, backed by a multi-source grid the US cannot match.

Adam Foroughi splits advertising in two: the bottom-of-funnel search ads that LLMs will cannibalise, and the discovery ads that show people things they did not know existed, which he argues an LLM cannot touch.

Daniel Dines predicts 90% of enterprise AI traffic will run on cheap, cost-efficient models rather than frontier ones, because most operational work does not need frontier-level quality.