The CapEx Train Keeps Rolling (This…

1. Key Themes

AI Infrastructure Capital Constraints Are the Next Bottleneck

The conversation has moved beyond compute and power scarcity to a third, less-discussed constraint: capital. The newsletter frames this as the critical near-term risk to the AI buildout.

"Everyone knows we are short of AI compute. Everyone knows that we may very soon be short on power. What happens, however, if we are short on capital? If AI is as valuable as it seems, then it should pay for itself, but that hasn't happened yet."

Nvidia Is Engineering Its Way Around the Capital Problem — With Real Risk

Nvidia is not passively waiting for customers to find funding; it is actively creating new financial mechanisms, which both extends the runway and concentrates risk.

"This week Nvidia announced a new funding mechanism to tap long-duration capital; this sort of financial engineering, along with Google leading the way in tapping equity, may build the bridge to sustainable AI revenue. It also expands the blast radius of a bubble in the service of Nvidia's threatened margins."

AI-Generated Content Regulation Is Advancing, With Philosophical Pushback

The EU is mandating watermarking of all AI text outputs, and Thompson considers this directionally wrong — not just technically flawed, but philosophically misguided.

"The EU has mandated that providers of AI systems mark all their outputs, including text, as AI-generated... Ben thinks all of this is a terrible idea, first and foremost for philosophical reasons."

Network Effects as a Moat: Geography Still Matters

Urban economic clustering (New York finance) vs. industry dispersal (Hollywood) offers a live case study in when network effects hold and when they break down.

"New York is a case study in the power of network effects, while Hollywood offers a lesson in the conditions that allow those effects to be broken."

2. Contrarian Perspectives

AI CapEx Will Find a Way — But the Bridge Is Risky Financial Engineering

The consensus assumes either (a) AI revenue will organically catch up to infrastructure spend, or (b) a bubble will pop. Thompson's framing suggests a third path: structured financial products and equity mechanisms filling the gap — though this expands systemic risk rather than eliminating it.

"This sort of financial engineering...may build the bridge to sustainable AI revenue. It also expands the blast radius of a bubble in the service of Nvidia's threatened margins."

The implication: Nvidia is not just a chip company but increasingly a financial intermediary bearing bubble risk to protect its own margins.

AI Watermarking Is Worse Than Useless

While regulators and many in the public view watermarking as a reasonable transparency measure, Thompson argues the opposite — that it is bad on principle, not just in execution.

"Anthropic is adding watermarking in response to the E.U.'s AI law. It's a terrible idea, first and foremost for philosophical reasons."

The deeper point, referencing Thompson's prior work on idea propagation across centuries, is that policing the origin of ideas and text runs against how human knowledge has always spread.

Generational Attitudes Toward AI Content Will Diverge Sharply

Thompson suggests the current anxiety about AI-generated content is a generational artifact, not a permanent societal norm.

"The likelihood that my son will likely feel differently about AI-generated output than I do."

This implies investors and builders designing for today's user anxieties may be solving a transitional problem rather than a durable one.

3. Companies Identified

Nvidia

Description: Leading AI chip designer
Why mentioned: Announced a new funding mechanism to provide long-duration capital to customers; framed as both a margin-defense move and a risk-expansion strategy
Quote: "Nvidia announced a new funding mechanism to tap long-duration capital...It also expands the blast radius of a bubble in the service of Nvidia's threatened margins."

Google

Description: Alphabet's core technology and cloud business
Why mentioned: Cited as a pioneer in tapping equity markets to fund AI infrastructure buildout
Quote: "Google leading the way in tapping equity...may build the bridge to sustainable AI revenue."

Apple

Description: Consumer hardware and software giant
Why mentioned: Earnings noted as constrained not by market demand or memory but by chip shortages
Quote: "Apple's earnings (and stock) are limited not by memory but rather chip shortages."

Amazon

Description: E-commerce and cloud infrastructure leader
Why mentioned: Earnings discussed alongside Andy Jassy's market analysis, suggesting strategic commentary on AI infrastructure demand
Quote: "More on Amazon's earnings and Andy Jassy's market analysis."

Anthropic

Description: AI safety-focused frontier model company
Why mentioned: Implementing EU-mandated AI watermarking, which Thompson critiques as misguided
Quote: "Anthropic is adding watermarking in response to the E.U.'s AI law. It's a terrible idea."

Paramount

Description: Major Hollywood film and media studio
Why mentioned: Used as an example of Hollywood's geographic and industrial fragility, with David Ellison threatening relocation
Quote: "Hollywood has unfortunately seen much of its filmmaking industry outsourced to neighboring states and foreign countries (as David Ellison threatens to relocate Paramount's operations)."

SK Hynix

Description: South Korean semiconductor manufacturer
Why mentioned: Featured in Asianometry coverage focused on HBM (High Bandwidth Memory), a critical AI infrastructure component
Quote: "SK Hynix and the HBM Revolution" (episode title)

OpenAI

Description: Leading frontier AI lab
Why mentioned: Identified as one of only two major players opposing open-weight AI models
Quote: "Why everyone but OpenAI and Anthropic wants open weight models."

4. People Identified

Ben Thompson

Description: Founder and primary author of Stratechery
Why mentioned: Author of the primary analysis on Nvidia's capital strategy and critic of AI watermarking regulation
Quote: "Nvidia is finding new ways for its customers to raise money, and it's expanding the risk of the AI buildout significantly."

Andrew Sharp

Description: Co-host of Sharp Tech, author of Sharp Text
Why mentioned: Contributed analysis on AI writing, NYC network effects, and Hollywood's industrial dispersal
Quote: "Caustic rhetoric from Mayor Zohran Mamdani is probably not enough to convince finance execs to leave New York City."

Zohran Mamdani

Description: Mayor of New York City; DSA-aligned politician
Why mentioned: Used as a case study in the limits of political rhetoric against entrenched economic network effects
Quote: "Mamdani exemplifies an irony of the DSA movement, generally."

David Ellison

Description: CEO of Skydance Media; oversees Paramount
Why mentioned: Threatening to relocate Paramount's operations, illustrating Hollywood's weakening geographic network effects
Quote: "David Ellison threatens to relocate Paramount's operations."

Andy Jassy

Description: CEO of Amazon
Why mentioned: His market analysis of AI infrastructure demand cited as notable in the context of Amazon's earnings
Quote: "More on Amazon's earnings and Andy Jassy's market analysis."

John Gruber

Description: Author of Daring Fireball; co-host of Dithering podcast
Why mentioned: Co-hosts Dithering with Thompson; discussed watermarking and frontier AI models
Quote: (Episode titles) "Immersive Baseball and Frontier Models" and "Watermarks"

Jon Yu

Description: Creator of Asianometry
Why mentioned: Produced analysis on subsea cable intelligence and SK Hynix's HBM dominance — both relevant to AI and geopolitical infrastructure
Quote: "The Subsea Cables Are Listening" and "SK Hynix and the HBM Revolution"

5. Operating Insights

Financial Engineering as a Strategic Weapon for Platform Companies

Nvidia's move to create its own capital mechanism is a masterclass in vertical integration beyond the product layer. When customers can't afford your product, don't wait for banks — become the financier. The risk is real, but so is the lock-in.

"Nvidia announced a new funding mechanism to tap long-duration capital...this sort of financial engineering...may build the bridge to sustainable AI revenue."

Geography Is Not Dead — Know Whether Your Industry Has Network Effects or Not

Operators deciding where to locate (or whether to fight political headwinds) should distinguish between industries with deep local network effects (finance in New York) and those where production can fragment (Hollywood). The strategic response is completely different.

"New York is a case study in the power of network effects, while Hollywood offers a lesson in the conditions that allow those effects to be broken."

Open-Weight Model Strategy Is Becoming a Competitive Divide

The alignment of virtually every major AI player except OpenAI and Anthropic behind open-weight models signals a growing strategic schism. Operators building on AI infrastructure should factor in which side of this divide their key vendors fall on.

"Why everyone but OpenAI and Anthropic wants open weight models."

6. Overlooked Insights

Subsea Cables as AI/Geopolitical Infrastructure

Jon Yu's Asianometry episode on subsea cables is briefly listed but potentially significant. As AI workloads become globally distributed, the physical internet backbone — and who controls or monitors it — becomes a strategic and national security issue that most AI investors underweight.

"The Subsea Cables Are Listening" (episode title, Asianometry)

Apple's Constraint Is Supply, Not Demand

Buried in the earnings note is a signal that Apple's growth ceiling is currently self-imposed by chip availability, not consumer appetite. This has implications for semiconductor suppliers and suggests upside optionality if supply normalizes.

"Apple's earnings (and stock) are limited not by memory but rather chip shortages."