Cargo Culture - Ed Zitron
An ISOC LIVE Summary
Ed Zitron’s Where’s Your Ed At - June 23 2026
Ed Zitron’s “Cargo Culture” is a sweeping critique of the modern AI industry and Silicon Valley’s broader economic ideology, arguing that today’s technology sector has become a “cargo cult” — a system that imitates the appearance of past innovation without reproducing its underlying substance. Zitron contends that the AI boom is not driven by genuine breakthroughs or sustainable economics, but by speculative consensus, ritualized behavior, and institutional desperation for new growth narratives.
Loops, Agents, and Token Consumption
The essay opens with criticism of the emerging concept of “loops,” in which AI systems recursively prompt or manage other AI systems in continuous workflows. Zitron argues that Anthropic figures such as Claude Code lead Boris Cherny are promoting loops less because they provide proven value and more because they dramatically increase token consumption and therefore revenue. He further asserts that NVIDIA CEO Jensen Huang has suggested traditional prompting is obsolete in favor of continuously-running autonomous agents.
Zitron frames this as symptomatic of a broader ideological culture inside Silicon Valley where skepticism toward AI is treated as heresy. He satirizes what he describes as a consensus-driven ecosystem in which executives, investors, and commentators reinforce one another’s claims about AI inevitability while dismissing economic concerns or practical limitations.
Silicon Valley as a Cargo Cult
The core thesis of the essay is that the tech industry has “run out of ideas.” Zitron argues that Silicon Valley increasingly imitates the superficial characteristics of earlier innovation waves — such as the dot-com era, smartphones, or cloud computing — without producing technologies of equivalent utility or economic viability. He describes this as “cargo cult” behavior: repeating rituals associated with previous success in the hope that prosperity will reappear.
As an example, Zitron critiques Snap’s new augmented reality “Specs” glasses, portraying them as an expensive, aesthetically unappealing product designed around imagined use cases disconnected from ordinary human needs. He contrasts these kinds of speculative gadgets with the actual frustrations users have with modern technology, such as advertising overload, degraded user experiences, dark patterns, scams, and bloated interfaces.
He links these trends to what he calls the “Rot-Com Bubble,” in which large technology firms have exhausted the major growth vectors of the Internet era and are now searching desperately for the “next iPhone.” According to Zitron, AI has become the latest vessel for this search because it offers a narrative of infinite future growth despite unclear economics.
AI Infrastructure Spending and the Illusion of Hypergrowth
A major section of the essay focuses on the enormous capital expenditures associated with AI infrastructure. Zitron argues that hyperscalers such as Microsoft, Google, Amazon, and Meta are investing hundreds of billions of dollars into GPUs and data centers not because of demonstrated profitability, but because they lack alternative growth strategies.
Zitron compares these expenditures unfavorably to earlier eras of innovation. He notes that the original iPhone reportedly cost around $150 million to develop ($271 million in today’s money), whereas modern AI companies raise or spend tens of billions of dollars annually without achieving comparable utility or profitability. He argues that AI differs fundamentally from prior technology revolutions because it depends on enormous continuing infrastructure costs while lacking clear paths to sustainable returns.
The essay repeatedly contrasts AI hype with the practical successes of earlier Internet businesses. Zitron emphasizes that companies such as Google, Instagram, or Amazon Web Services emerged from relatively modest investments combined with clear product-market fit and strong utility. By contrast, he portrays current AI spending as speculative overcapitalization disconnected from actual user demand.
Decline of the Software and Venture Capital Model
Zitron argues that the software industry itself is entering structural decline. He discusses the so-called “SaaSpocalypse,” claiming that declining growth rates among SaaS firms reflect broader market saturation rather than revolutionary AI-driven disruption. He cites reports showing deteriorating growth efficiency, stagnating revenues, and declining investor returns across public software companies.
He extends this critique to venture capital and private equity, arguing that both industries became addicted to assumptions of perpetual software growth during the low-interest-rate era. Zitron highlights declining venture returns, increasing numbers of “zombie unicorns,” and extended holding periods for private equity-backed firms. He argues that investors increasingly recycle capital among themselves through secondary transactions rather than generating genuine value.
According to Zitron, venture capital no longer rewards genuine risk-taking or invention. Instead, firms overwhelmingly fund companies that resemble previous successes or align with current hype cycles, particularly AI. He uses the example of AI infrastructure startup Baseten, whose business model ultimately channels investor money back to the major cloud providers supplying GPU infrastructure.
Executive Culture and “Founder Mode”
The essay criticizes modern technology leadership as excessively mythologized and detached from productive work. Zitron discusses Airbnb CEO Brian Chesky and the idea of “founder mode,” which argues that CEOs should stay deeply involved in operational details. Zitron contends that this philosophy has been transformed into another Silicon Valley mythology centered around copying symbolic behaviors associated with figures such as Steve Jobs rather than understanding the actual historical circumstances behind Apple’s success.
He argues that many modern executives seek simplified formulas or rituals for success instead of acknowledging the role of timing, talent, infrastructure, and luck in previous technology revolutions. This obsession with symbolic imitation extends, in his view, to AI hiring frenzies and celebrity researcher acquisitions.
OpenAI, Anthropic, and the Economics of AI
Zitron portrays OpenAI and Anthropic not as startups but as quasi-infrastructure companies whose operations depend entirely on hyperscaler support and continuous investor funding. He argues that their valuations and spending patterns would be regarded as irrational in any other industry.
The essay questions the profitability of AI inference and suggests that some companies may obscure actual operating costs through accounting practices. Zitron specifically challenges reported OpenAI financials, arguing that sales and marketing expenses appear implausibly large relative to the company’s visible operations and may conceal inference costs or subsidized user activity.
He contends that the AI sector survives largely through subsidies, discounted access, and speculative belief rather than economically sustainable demand. In his view, many organizations adopt AI because executives fear being left behind, not because AI products demonstrably improve productivity or profitability.
Media, Mythology, and Symbolic Capital
A recurring theme throughout the essay is the role of media and narrative creation. Zitron argues that technology journalism and financial analysis have become complicit in sustaining AI hype by constantly searching for the “next big thing.” He claims that journalists, investors, and executives alike rely on symbolic comparisons to previous technology revolutions rather than rigorous economic analysis.
The essay describes AI as the culmination of Silicon Valley’s transformation into a system optimized for symbolic capital — narratives, valuations, and growth expectations — rather than practical innovation. Zitron argues that AI companies are judged primarily by their resemblance to previous success stories rather than by present-day financial or technological realities.
Conclusion: Demanding Proof Rather Than Faith
Zitron concludes by arguing that AI advocates should no longer be allowed to justify present losses with historical analogies or speculative future promises. He insists that AI must prove its value and sustainability in the present rather than relying on mythology about previous technology booms. He challenges readers to reject this “cargo cult” mentality and evaluate AI on measurable utility, sustainable economics, and real-world outcomes rather than hype, symbolism, or speculative inevitability.
FACT CHECK
Two events, blended. Zitron treats Cherny’s loops remarks as one occasion. They’re two: the “are loops for real?” exchange was at Meta’s @Scale conference (Friday), while “I don’t prompt Claude anymore… my job is to write loops” comes from Cherny’s remarks at Anthropic’s developer conference, reported via CNBC.
Jensen Huang attribution is unsupported. Zitron asserts Huang “intimated that the age of prompting models is over” but cites no quote or source. The widely-reported loops framing traces to Cherny, OpenAI engineer Peter Steinberger (”you should be designing loops that prompt your agents”), and Google’s Addy Osmani — not Huang.
The $130,000/month figure is Zitron’s, uncited. His claim that Anthropic lets Cherny “burn upwards of $130,000 a month in tokens” has no linked source; treat as assertion, not established fact.
“$200 billion on GPUs” includes capacity. Zitron’s own wording is “$200 billion just on buying GPUs and building capacity” — infrastructure broadly, not GPUs alone.
iPhone cost. The ~$150 million development figure is nominal; Zitron gives $271 million inflation-adjusted, which is the figure that actually grounds his comparison to SoftBank’s OpenAI investment.
RESOURCES
Cargo Culture — Ed Zitron’s original essay (Where’s Your Ed At, Jun 2026)
The AI world is getting ‘loopy’ — TechCrunch on Cherny’s @Scale loops remarks
Ed Zitron — author, host of the Better Offline podcast
How Boris Uses Claude Code — Cherny’s own documented loop setup (/loop, /schedule)
Loop Engineering — Addy Osmani’s (Google) essay naming the pattern
Anthropic — maker of Claude and Claude Code
OpenAI — maker of ChatGPT and Codex
Snap — maker of the $2,195 Specs AR glasses Zitron critiques
Baseten — AI inference startup cited in the venture-capital section
Founder Mode — Paul Graham’s essay underpinning the Chesky discussion