PodBrowser
Impact Theory

Livestreamed Executions, Wall Street AI Panic, and the War for Worker Power | The Tom Bilyeu Show

Friday, 9 October 2026 · 4 min read · Listen to the episode ↗

Tom Bilyeu condemns Hegseth’s reported plan to livestream a firing-squad execution as spectacle rather than somber justice. The discussion turns to AI market jitters: Oracle, NVIDIA, Broadcom, Micron, and AMD shed a combined $370 billion in market value in one session amid doubts about data-center demand and power. It also weighs the administration’s freeze on Green Card access for eight major tech employers and whether H-1B dependence weakens worker bargaining power.

Tom Bilyeu said Hegseth had committed to livestream an execution, and described a plan to take a man who murdered multiple people to the crime scene and execute him by firing squad. Bilyeu planned to watch but objected to making the killing a spectacle, calling it vengeance and “bread and circus.” He argued executions should be somber rather than fuel bloodlust, and worried that the broadcast would make America look bad. He also criticized Hegseth’s “warriors” framing as excessive, contrasting it with toughness grounded in training, honor, and service. Bilyeu praised Marco Rubio’s speech for pairing strength with moral clarity.

OpenAI confirmed, according to the discussion, that Iranian operatives used ChatGPT in a propaganda operation targeting U.S. media. The banned account cluster translated prompts into English, tailored drafts to outlets’ submission rules, and wrote pitch emails under fake Western freelance-journalist identities; nearly 100 pieces about the U.S.-Iran war appeared in real U.S. outlets. OpenAI rated the operation four out of six on the Brookings breakout scale. The speakers warned that AI could make state-backed propaganda harder to distinguish from reality, recommending primary sources and strong opposing views. One proposed credibility profiles displaying signals such as account age, posting region, bot interactions, and undisclosed AI use without requiring users to reveal their identities.

The discussion portrayed AI valuations and data-center demand as facing a more cautious market. OpenAI told investors its annualized revenue was approaching $50 billion at the end of September, while Axios and Reuters had reported an estimate of about $70 billion 10 days earlier. Bilyeu said the higher figure reflected investors comparing OpenAI with Anthropic, not a $20 billion fall in OpenAI sales. Oracle, NVIDIA, Broadcom, Micron, and AMD lost a combined $370 billion in market value in one session; NVIDIA fell 2.9%, Broadcom 4.4%, Micron 4.8%, and AMD 3.9%. Oracle had risen more than 5% after the $70 billion report, then fell 5.5% during the correction.

Bilyeu said data-center investors were uncertain about real demand, power access, and whether projects could meet payment obligations. He said Oracle invoked force majeure after a New Mexico facility could not obtain power, meaning it would not make associated payments. An Australian data-center IPO, cut from an initial $11 price to $8, was withdrawn amid weaker demand, though the company could try again. The speakers described an investment cycle in which infrastructure costs arrive before expected revenue, while longer private-company timelines can shift more risk to public investors at IPO. They called the IPO run-up frothy but cautioned that continuing technological progress meant the correction was not necessarily an immediate bubble burst; in real time, it is difficult to know whether a pullback is healthy or the start of a collapse.

The administration froze Green Card program access for eight major users of foreign tech labor: Microsoft, Adobe, Cognizant, Infosys, Tata, Wipro, HCL, and Capgemini. J.D. Vance said the message was to hire Americans rather than lay them off and replace them with foreign workers, and argued that H-1B workers’ reliance on employers can depress wages. Bilyeu said Microsoft had just under 4,000 H-1B employees while laying off Americans; other figures cited 6,000 Microsoft layoffs and 6,300 H-1B visas, with a filing reportedly showing 80% of the visas were renewals. He stressed that these figures do not prove Americans were laid off to make way for H-1B workers. He argued that limiting paths from H-1B employment to permanent status could weaken workers’ bargaining power, while acknowledging that describing highly skilled workers as “indentured servants” overstates their situation if another employer can take over their visa process.

Bilyeu argued that policy should prioritize middle-class jobs and rising purchasing power, not just nominal wages, as globalization and technology have weakened workers’ bargaining power and shifted more productivity gains to executives. He supports incentives to hire Americans but said employers who cannot find needed U.S. talent may send jobs abroad; his company, Impact Theory, also needs some specialized workers outside the country. He said a federal court struck down the administration’s proposed $100,000 H-1B fee as an illegal tax, although he had supported it. More broadly, the speakers said AI-generated content may soon be indistinguishable from, and possibly better than, human writing, with most internet content eventually generated by AI. They favored identifying AI-made material over restricting posting, while noting that creators will need enforceable copyright, trademark, or DMCA-like protection as AI makes copying and repurposing their work easier.

This summary was generated from the episode transcript and can contain mistakes.