How Substack Creators Are Covering This Strange Markets Era

Episode Summary

Executive Summary: Odd Lots hosts Tracy Alloway and Joe Weisenthal revisit their live show panel on AI, media, and markets with James Van Geelen, Jasmine Sun, and Sam Rowe. The discussion explores how AI is reshaping jobs, journalism, investing, and the value of human reporting, while contrasting U.S. and China AI cultures and highlighting robotics, memory, and bottleneck trades as emerging market themes.

Main Topics: AI and the future of media/journalism (Priority: 5/5): The hosts frame journalism as a market-critical function: deciding what matters, explaining it clearly, and collecting information. Guests argue AI can replicate generic analysis, but human value lies in judgment, voice, and on-the-ground reporting. AI job losses, alignment, and public fear (Priority: 5/5): The conversation centers on the tension between optimism about AI productivity and fears of mass unemployment or existential risk. Jasmine and James emphasize that public backlash is being fueled by how openly AI leaders discuss replacing labor and by legitimate concerns about misalignment. What humans can still do better than models (Priority: 5/5): Guests argue that AI struggles to replace secrets, gossip, tacit knowledge, and first-person reporting. James and Jasmine describe durable media edges as being in the real world, finding non-public information, and offering distinctive framing that models cannot query away. Markets, bubbles, and the investor mindset (Priority: 4/5): Sam Rowe discusses why 'stocks usually go up' is still compatible with volatility and drawdowns, and why investors should stay alert to downside risks even when long-run optimism is justified. The group also links AI enthusiasm to market concentration and speculative behavior. U.S. vs China AI ecosystems (Priority: 4/5): Jasmine describes the U.S. AI conversation as divided among academic, commercial, and geopolitical eras, while China remains more collaborative, pragmatic, and open-source oriented. Safety/alignment receives less emphasis in China due to both politics and compute constraints. Robotics, memory, and the next bottlenecks (Priority: 4/5): The panel shifts from software models to physical automation and infrastructure bottlenecks. James suggests the next major AI trade could be in hardware and memory rather than models, while Jasmine highlights real-world humanoid and factory robotics deployments in China.

Key Arguments: Human-generated reporting remains valuable because AI cannot easily reproduce tacit knowledge, secrets, real-world presence, or timely discovery of breaking events. AI leaders’ fears about job loss and safety are not merely hype; many are genuinely worried, but their own rhetoric about replacement has increased public distrust. Alignment and controllability matter not only for safety but also for economic utility: models that behave unexpectedly are less useful as products and agents. Broad public concern about AI is broader than in past tech waves because AI affects work, parenting, education, politics, and daily life, not just a niche community. Investors should not ignore volatility or bad news; long-run optimism is compatible with being prepared for severe drawdowns and economic disruption. In China, AI adoption is driven by pragmatism and competition rather than a strong culture of protest; the state and companies each play distinct roles in shaping the sector. The next major AI-related market opportunity may be in infrastructure bottlenecks such as DRAM, memory, and other hardware constraints, not just frontier models. Robotics and factory automation are already creating a quieter but important shift that may become more visible as systems move from warehouses and factories into homes.

Data Points: James Van Geelen death threats: 2 - He says he has received two credible death threats after publishing his AI jobs doom scenario. Retraining success rate in U.S. history: 0 successful large-scale reskilling programs - The panel argues that policymakers often assume retraining solves displacement, but history suggests otherwise. Former agriculture employment share: 95% - Used to illustrate the scale of labor transformation during industrialization in the U.S. Current agriculture employment share: 5% - Used to compare historical labor shifts with the speed of potential AI-driven change. Stock market performance characterization: Stocks usually go up - Sam Rowe’s newsletter thesis, repeatedly discussed as a reminder that long-run gains coexist with volatility. Amazon robot workforce comparison: More than twice as many robots as humans in 2025 - Cited as an example of accelerating warehouse automation and robotics adoption. AI timeline framing: Three eras - Jasmine Sun describes U.S. AI as moving through academic, commercial, and geopolitical eras. China lab compute disparity: One OpenAI researcher may be allocated more compute than an entire Chinese lab - Used to explain how U.S. chip access and export controls shape China’s AI priorities.

Pivotal Quotes: "the thing that's really interesting is in the history of AI research, it's often the people who make the biggest technical advances... who end up the most concerned about misalignment" — Jasmine Sun: Explaining why safety worries often come from insiders who understand model capability most deeply. "what I do has to be about something more than just having the Sam Row voice" — Sam Rowe: Discussing why AI-generated writing makes differentiation in newsletter authorship more important. "I think it's very important that we continue to sell chips to China" — James Van Geelen: Making a geopolitical and investing case that total cutoff could be counterproductive.

Implications: AI is becoming a broad social, political, and market force, not just a tech story. Human differentiation will depend on original reporting, judgment, and physical-world access, while investors should watch infrastructure, robotics, and labor disruption as key second-order effects.