ChinaTalk

A sharp Mandarin-language tour through China’s AI scene, from why AI writing struggles to be funny to benchmarks built around memes and everyday users. The conversation explores failed agent startups, world-model hype, Chinglish branding, and whether Chinese or American tech culture exaggerates more. It also examines AI’s manufacturing-like foundations, growing mass-market demand, and the compute demands of agents.

Tony Stark, a national security commentator; Bryan Clark, a retired Navy submariner and naval strategist; and Justin McIntosh, a former Green Beret and defense technology specialist, dissect the expanding Iran conflict and the prospect of intervention in Cuba. They examine carrier strain, Pentagon dysfunction, espionage and Nvidia’s China ties, Afghanistan’s enduring myths, and the growing risk of letting AI shape decisions about war.

Barnett Koven, Accenture Federal Services’ Managing Director of AI and Data, brings a quantitative researcher’s perspective to government technology. He explores AI-powered export-control enforcement, PLA shell-company networks, and the risks of bureaucratic caution. The conversation also examines classified AI security, rising compute demands, and the Genesis Mission’s ambitious plan to build a secure backbone for scientific research.

Meg Rithmire, a Harvard Business School professor and author of Precarious Ties, examines China’s troubled economy. She discusses how Anbang, HNA, and Wanda looted financial systems, why industrial policy inherited corruption, and how households experience China’s uneven dynamism. She also makes the case for allowing BYD and CATL to build in America under strict safeguards, and explores corruption, institutional trust, and AI-powered investigative research.

Michael Kofman, a leading analyst of Russia’s military, joins retired Green Beret Justin McIntosh and defense commentator Tony Stark. They examine Ukraine’s lessons for U.S. forces: America’s leverage and Europe’s limits, why drones demand new units and logistics, the collapse of nighttime dominance, deception and decoys, mechanized maneuver, perilous river crossings, and what Congress should prioritize.

Nick Brown, a physicist, software engineer, and co-founder of the Center for Technology and Statecraft, joins technology policy researcher Aqib Zakaria. They explore why DUV lithography remains crucial to China’s AI-chip ambitions, the surge in Chinese imports of ASML systems, and how the MATCH Act could restrict exports and servicing. The conversation also examines allied coordination, possible Beijing retaliation, and the race toward indigenous lithography.

Travis Metz, DIU’s acting director and former Drone Dominance leader, joins former Green Beret Justin McIntosh and ex-submariner Bryan Clark. They explore scaling U.S. drone production, lessons from Ukraine, and surviving electronic warfare. Supply-chain dependence on Chinese components, Indo-Pacific challenges, and autonomous systems that could transform procurement, training, and force structure are also in focus.

Richard Bian, Ant Group’s AI product leader and former Microsoft and Square engineer, explores why AGI may be inevitable and what its arrival could mean. He discusses the Hugging Face hack, open-weight models as a product strategy, AI’s gross-margin challenge, better evaluations, tiny private models, and Ant Ling’s vision for finance, health, and developer ecosystems.

Saif Khan, a technology policy researcher and former U.S. official, discusses launching the Center for Technology and Statecraft. He explores DUV lithography as a pivotal lever in U.S.-China AI competition, the rise of robotics, and how rigorous research becomes policy. They also examine AI’s impact on government, work, cultural learning, childhood, and the possibility of insect-like machine societies.

Jerusalem Demsas, a journalist and policy commentator focused on housing and institutions, joins Matt Yglesias, a political writer known for public-policy and economic analysis. They roam from AI’s democratic legitimacy and impact on Congress to China’s censored models, automation’s threat to development, and contrasting optimism about disruption. They also debate corruption, AI-powered oversight, and what Trump’s summit with Xi really achieved.