Venture capitalist and “All In Podcast” co-host Jason Calacanis joined host Adam Curry on episode 1905 of the Agenda Show, an installment titled “Nerdballs” published on September 20, 2026, delivering a pointed critique of the AI hype cycle, Silicon Valley power struggles, and the media narratives shaping the 2026 midterms. The episode presents a skeptical counterpoint to mainstream AI coverage, with Calacanis arguing that leading AI safety firm Anthropic, the maker of the Claude family of models, has adopted a quasi-religious belief in its own mission.

“They actually think that they are creating a god. They’re not. They’re creating software,” Calacanis said, comparing the company’s effective-altruism roots—a philosophy centered on guarding against catastrophic AI risk—to the doomed engineers in the film “Prometheus.”

Curry played clips of University of Washington professor Pedro Domingos describing a reported Anthropic “funeral” for a retired Claude model, an anecdote Calacanis cited as evidence of what he views as a dangerous, cult-like corporate culture. The conversation also addressed Microsoft AI CEO Mustafa Suleyman’s newly introduced code of conduct and his public criticism of Anthropic’s “model welfare” doctrine—the idea that advanced AI systems might warrant moral consideration—along with a reported Hugging Face security incident and growing calls for an AI “kill switch,” a fail-safe mechanism to shut down frontier systems.

For Calacanis, the underlying story is regulatory capture—the dynamic in which incumbent firms shape the rules meant to govern their own industry. argued that frontier labs are lobbying Washington to gate open-source competitors such as DeepSeek, GLM, and NeMoTron, contending that AI safety rhetoric can serve as a competitive moat that stifles innovation and concentrates power among a handful of firms. How open-source models fare in any resulting U.S. rules is one thread to watch as that clash plays out.

The discussion extended beyond AI policy. The pair examined the political weaponization of data centers by candidates such as Texas Democrat James Talarico, President Trump’s revocation of press credentials for CNN, MSNOW, and Politico, and rising diesel prices tied to Ukrainian strikes on Russian refineries.

Calacanis shared backstage details from the All In Summit at the Shrine Auditorium, which drew 3,000 attendees and featured surprise call-ins from President Trump, Vice President JD Vance, Elon Musk, and Nvidia CEO Jensen Huang. He recounted Huang fumbling the speakerphone button when Trump called mid-panel, and described pressing Meta’s Dina about the company’s record $17 billion settlement during a data-center segment.

Calacanis also traced the origins of “All In” to a pandemic-era poker group with David Sacks, Chamath Palihapitiya, and David Friedberg, and noted that his program “This Week in Startups” sells out its ad inventory a year in advance at $3,000 per spot. Curry pivoted to Bitcoin, predicting a $430,000 peak, while Calacanis discussed parenting in the smartphone era. The pair reflected on the late John C. Dvorak, whose memory drives the show’s ongoing Order of the Ant tribute campaign.

By framing AI development as a battle over narrative and regulation, the episode highlights the growing tension between open-source advocates and closed-model labs. It also underscores how media deconstruction can reveal the financial and political incentives behind tech coverage as the 2026 midterms approach. For listeners, the takeaway is that the AI debate is as much about power and money as it is about technology—and Suleyman’s code of conduct, the kill-switch debate, and any movement on open-source rules are markers of where that fight heads next.