{"version":"https://jsonfeed.org/version/1","title":"Valley101","home_page_url":"https://valley101.fireside.fm","feed_url":"https://valley101.fireside.fm/json","description":"Valley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it. Subscribe for more stories from Silicon Valley and beyond. Contact: video@sv101.net","_fireside":{"subtitle":"Stories and people shaping frontier tech, from the heart of Silicon Valley.","pubdate":"2026-10-21T19:30:00.000-07:00","explicit":false,"copyright":"2026 by SiliconValley, tech, AI, business","owner":"Valley101,Inc","image":"https://cdn.fireside.fm/file/fireside-images-2024/podcasts/images/1/180e38bd-cc91-4666-a045-f3d412cb6b4f/cover.jpg?v=1"},"items":[{"id":"bafb0932-563b-4a7d-819f-33b1f6372c3b","title":"Episode 5: We are Giving AI Personalities Before We Understand How They Think","url":"https://valley101.fireside.fm/5","content_text":"Silicon Valley 101 premieres on YouTube on August 27, 2026.\n\nAI is getting smarter. But do we actually know how they work?\n\nStanford researcher Aryaman Arora takes us inside the science of AI interpretability, a field of research where we try to understand what’s happening inside language models.\n\nWe discussed Anthropic’s research of a hidden internal neural workspace in large language models, Golden Gate Claude, user awareness research at Transluce, and why different models are starting to develop increasingly distinct personalities.\n\nWe also talked about the various research traditions, what sparse autoencoders and causal interventions can reveal, and why we can actually edit a model’s internal thoughts.\n\nAryaman Arora is a PhD researcher at Stanford, advised by Christopher Potts and Dan Jurafsky. His research focuses on AI interpretability, language models, and linguistics, and he has also worked at organizations including Transluce and Redwood Research.\n\nTimestamp:\n00:00 What is interpretability research?\n03:11 J-Space: Do models have an internal reasoning space?\n07:34 Do models have internal reasoning space?\n08:56 Can chain-of thought tell us what a model is thinking?\n12:07 Interpretability, alignment, and trusting AI\n16:10 Sparse autoencoders: translating AI’s internal representations\n19:34 Two competing approaches to interpretability\n23:32 The compute problem\n24:11 Linguistics\n30:06 Interpretability benchmarks\n31:28 Gold Gate Claude: can we edit an AI’s thoughts?\n34:09 Transluce and user awareness in frontier models\n36:09 Weird chat and discovering strange model behaviors\n37:02 Is it all about edge cases?\n38:36 Can interpretability explain hallucinations?\n40:27 H3 paper\n43:05 Academia vs. industry\n48:15 What’s next\n\nAbout Valley101:\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\nSubscribe for more stories from Silicon Valley and beyond.\n\nContact: video@sv101.net","content_html":"\u003cp\u003e\u003cem\u003eSilicon Valley 101 premieres on \u003ca href=\"https://www.youtube.com/watch?v=Drik4OrzhQc\" rel=\"nofollow noopener\"\u003eYouTube\u003c/a\u003e on August 27, 2026.\u003c/em\u003e\u003c/p\u003e\n\n\u003cp\u003eAI is getting smarter. But do we actually know how they work?\u003c/p\u003e\n\n\u003cp\u003eStanford researcher Aryaman Arora takes us inside the science of AI interpretability, a field of research where we try to understand what’s happening inside language models.\u003c/p\u003e\n\n\u003cp\u003eWe discussed Anthropic’s research of a hidden internal neural workspace in large language models, Golden Gate Claude, user awareness research at Transluce, and why different models are starting to develop increasingly distinct personalities.\u003c/p\u003e\n\n\u003cp\u003eWe also talked about the various research traditions, what sparse autoencoders and causal interventions can reveal, and why we can actually edit a model’s internal thoughts.\u003c/p\u003e\n\n\u003cp\u003eAryaman Arora is a PhD researcher at Stanford, advised by Christopher Potts and Dan Jurafsky. His research focuses on AI interpretability, language models, and linguistics, and he has also worked at organizations including Transluce and Redwood Research.\u003c/p\u003e\n\n\u003cp\u003eTimestamp:\u003cbr\u003e\n00:00 What is interpretability research?\u003cbr\u003e\n03:11 J-Space: Do models have an internal reasoning space?\u003cbr\u003e\n07:34 Do models have internal reasoning space?\u003cbr\u003e\n08:56 Can chain-of thought tell us what a model is thinking?\u003cbr\u003e\n12:07 Interpretability, alignment, and trusting AI\u003cbr\u003e\n16:10 Sparse autoencoders: translating AI’s internal representations\u003cbr\u003e\n19:34 Two competing approaches to interpretability\u003cbr\u003e\n23:32 The compute problem\u003cbr\u003e\n24:11 Linguistics\u003cbr\u003e\n30:06 Interpretability benchmarks\u003cbr\u003e\n31:28 Gold Gate Claude: can we edit an AI’s thoughts?\u003cbr\u003e\n34:09 Transluce and user awareness in frontier models\u003cbr\u003e\n36:09 Weird chat and discovering strange model behaviors\u003cbr\u003e\n37:02 Is it all about edge cases?\u003cbr\u003e\n38:36 Can interpretability explain hallucinations?\u003cbr\u003e\n40:27 H3 paper\u003cbr\u003e\n43:05 Academia vs. industry\u003cbr\u003e\n48:15 What’s next\u003c/p\u003e\n\n\u003cp\u003eAbout Valley101:\u003cbr\u003e\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\u003cbr\u003e\nSubscribe for more stories from Silicon Valley and beyond.\u003c/p\u003e\n\n\u003cp\u003eContact: \u003ca href=\"mailto:video@sv101.net\" rel=\"nofollow noopener\"\u003evideo@sv101.net\u003c/a\u003e\u003c/p\u003e","summary":"AI is getting smarter. But do we actually know how they work?","date_published":"2026-10-21T19:30:00.000-07:00","attachments":[{"url":"https://aphid.fireside.fm/d/1437767933/180e38bd-cc91-4666-a045-f3d412cb6b4f/bafb0932-563b-4a7d-819f-33b1f6372c3b.mp3","mime_type":"audio/mpeg","size_in_bytes":125991302,"duration_in_seconds":3144}]},{"id":"7c0a3b6a-3812-4dd8-8215-5d1607787950","title":"Episode 4: How Chinese Open-Source Models Are Rewriting AI Economics｜5Y Capital’s Meng Xing","url":"https://valley101.fireside.fm/4","content_text":"Silicon Valley 101 premieres on YouTube on August 18, 2026.\n\nChina’s AI ecosystem is moving fast. Kimi, DeepSeek, and Z.ai are building increasingly capable AI models. But how much is Silicon Valley actually paying attention?\nXing Meng, partner at 5Y Capital, joins us for an inside look at China’s AI industry and the rise of Chinese open-source and open-weight models. We discuss Kimi and Moonshot AI, DeepSeek, ByteDance, the young researchers behind China’s frontier AI labs, and how China’s AI ecosystem differs from the US.\nWe also dig into the distillation debate, video models, data providers, and why Chinese models may be having a bigger impact on AI developers than on researchers at OpenAI, Anthropic and other frontier labs.\n\nTimeline: \n00:00 Xing Meng\n02:06 Silicon Valley trip recap\n05:45 Kimi, Moonshot AI and China’s young AI researchers\n14:32 DeepSeek and unusual Liang Wenfeng \n18:05 Z.ai, MiniMax and what makes a great AI lab\n22:14 ByteDance, Zhang Yiming’s bold decision, distillation debate\n26:54 Are Chinese open-source models actually changing Silicon Valley?\n30:56 Are frontier AI models becoming commoditized?\n35:03 Why China is leading in AI video generation\n46:52 Neo Labs, AI agents and the future of model companies\n49:05 The next frontier of AI training data\n58:04 How US and China approaches data differently\n1:04:26 AI for science\n1:10:26 Silicon Valley’s biggest misconception about China’s AI ecosystem\n\nAbout Valley101:\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\n\nSubscribe for more stories from Silicon Valley and beyond.\n\nContact: video@sv101.net","content_html":"\u003cp\u003e\u003cem\u003eSilicon Valley 101 premieres on \u003ca href=\"https://www.youtube.com/watch?v=HRNacC3pKSI\" rel=\"nofollow noopener\"\u003eYouTube\u003c/a\u003e on August 18, 2026.\u003c/em\u003e\u003c/p\u003e\n\n\u003cp\u003eChina’s AI ecosystem is moving fast. Kimi, DeepSeek, and Z.ai are building increasingly capable AI models. But how much is Silicon Valley actually paying attention?\u003cbr\u003e\nXing Meng, partner at 5Y Capital, joins us for an inside look at China’s AI industry and the rise of Chinese open-source and open-weight models. We discuss Kimi and Moonshot AI, DeepSeek, ByteDance, the young researchers behind China’s frontier AI labs, and how China’s AI ecosystem differs from the US.\u003cbr\u003e\nWe also dig into the distillation debate, video models, data providers, and why Chinese models may be having a bigger impact on AI developers than on researchers at OpenAI, Anthropic and other frontier labs.\u003c/p\u003e\n\n\u003cp\u003eTimeline: \u003cbr\u003e\n00:00 Xing Meng\u003cbr\u003e\n02:06 Silicon Valley trip recap\u003cbr\u003e\n05:45 Kimi, Moonshot AI and China’s young AI researchers\u003cbr\u003e\n14:32 DeepSeek and unusual Liang Wenfeng \u003cbr\u003e\n18:05 Z.ai, MiniMax and what makes a great AI lab\u003cbr\u003e\n22:14 ByteDance, Zhang Yiming’s bold decision, distillation debate\u003cbr\u003e\n26:54 Are Chinese open-source models actually changing Silicon Valley?\u003cbr\u003e\n30:56 Are frontier AI models becoming commoditized?\u003cbr\u003e\n35:03 Why China is leading in AI video generation\u003cbr\u003e\n46:52 Neo Labs, AI agents and the future of model companies\u003cbr\u003e\n49:05 The next frontier of AI training data\u003cbr\u003e\n58:04 How US and China approaches data differently\u003cbr\u003e\n1:04:26 AI for science\u003cbr\u003e\n1:10:26 Silicon Valley’s biggest misconception about China’s AI ecosystem\u003c/p\u003e\n\n\u003cp\u003eAbout Valley101:\u003cbr\u003e\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\u003c/p\u003e\n\n\u003cp\u003eSubscribe for more stories from Silicon Valley and beyond.\u003c/p\u003e\n\n\u003cp\u003eContact: \u003ca href=\"mailto:video@sv101.net\" rel=\"nofollow noopener\"\u003evideo@sv101.net\u003c/a\u003e\u003c/p\u003e","summary":"Xing Meng, partner at 5Y Capital, joins us for an inside look at China’s AI industry and the rise of Chinese open-source and open-weight models. ","date_published":"2026-10-21T19:15:00.000-07:00","attachments":[{"url":"https://aphid.fireside.fm/d/1437767933/180e38bd-cc91-4666-a045-f3d412cb6b4f/7c0a3b6a-3812-4dd8-8215-5d1607787950.mp3","mime_type":"audio/mpeg","size_in_bytes":173060321,"duration_in_seconds":4321}]},{"id":"1fee6030-fe0b-42a5-b383-d8764cb5209f","title":"Episode 3: Kevin Kelly on AI’s Uncertain Future: By 2049, Humans Will Fall in Love With AI","url":"https://valley101.fireside.fm/3","content_text":"Silicon Valley 101 premieres on YouTube on July 31, 2026.\n\nWhat will the world look like 10,000 days from now?\nIn this episode, we sit down with Kevin Kelly (K.K.), a Silicon Valley visionary and the founding executive editor of WIRED. Drawing on the ideas in his book 2049: Possibilities of Next 10000 Days, we explore his vision of the world ahead.\nBy 2049, humans may fall in love with AI. A new generation may find themselves working in jobs that do not yet exist. AI could become embedded in nearly every aspect of daily life, while humanity enters a “mirrorworld” that digitally maps the physical world. And despite our ambitions in space, Kelly believes there may never be cities on Mars.\nWe also discuss questions that matter to all of us: How should we educate the next generation? If AGI becomes a reality, what uniquely human qualities will remain scarce and valuable? And what opportunities could China have in the emerging market for AI glasses?\nJoin us for a deep conversation with K.K.— a picture what our lives and our world might look like by 2049.\n\nTimeline: \n00:00 Technology’s Double-Edged Sword: Inversion and Optimism About the Future\n04:19 AI and Education: Learning How to Learn\n09:37 AI and Media: Content Will Explode, but Human Journalists Must Remain Accountable\n16:31 AI and Emotions: The New Impact of the Mirrorworld\n21:26 The AI Glasses Race: Could a Chinese Company Come Out on Top?\n26:16 Robotics: Humanoid Robots Will Get Cheaper—but Not More Useful\n29:21 Self-Driving Cars: Getting a Driver’s License May Become Harder for Humans\n31:19 Space Technology: There Will Never Be Cities on Mars\n34:34 Life Sciences and Longevity: Extending Health Matters More Than Extending Lifespan\n39:04 Brain-Computer Interfaces: Non-Invasive Connections Within 25 Years\n41:05 Silicon Valley in 2049: How the Path to Innovation May Change\n\nAbout Valley101:\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\n\nSubscribe for more stories from Silicon Valley and beyond.\n\nContact: video@sv101.net","content_html":"\u003cp\u003e\u003cem\u003eSilicon Valley 101 premieres on \u003ca href=\"https://www.youtube.com/watch?v=w-M2QvFgQZM\" rel=\"nofollow noopener\"\u003eYouTube\u003c/a\u003e on July 31, 2026.\u003c/em\u003e\u003c/p\u003e\n\n\u003cp\u003eWhat will the world look like 10,000 days from now?\u003cbr\u003e\nIn this episode, we sit down with Kevin Kelly (K.K.), a Silicon Valley visionary and the founding executive editor of WIRED. Drawing on the ideas in his book 2049: Possibilities of Next 10000 Days, we explore his vision of the world ahead.\u003cbr\u003e\nBy 2049, humans may fall in love with AI. A new generation may find themselves working in jobs that do not yet exist. AI could become embedded in nearly every aspect of daily life, while humanity enters a “mirrorworld” that digitally maps the physical world. And despite our ambitions in space, Kelly believes there may never be cities on Mars.\u003cbr\u003e\nWe also discuss questions that matter to all of us: How should we educate the next generation? If AGI becomes a reality, what uniquely human qualities will remain scarce and valuable? And what opportunities could China have in the emerging market for AI glasses?\u003cbr\u003e\nJoin us for a deep conversation with K.K.— a picture what our lives and our world might look like by 2049.\u003c/p\u003e\n\n\u003cp\u003eTimeline: \u003cbr\u003e\n00:00 Technology’s Double-Edged Sword: Inversion and Optimism About the Future\u003cbr\u003e\n04:19 AI and Education: Learning How to Learn\u003cbr\u003e\n09:37 AI and Media: Content Will Explode, but Human Journalists Must Remain Accountable\u003cbr\u003e\n16:31 AI and Emotions: The New Impact of the Mirrorworld\u003cbr\u003e\n21:26 The AI Glasses Race: Could a Chinese Company Come Out on Top?\u003cbr\u003e\n26:16 Robotics: Humanoid Robots Will Get Cheaper—but Not More Useful\u003cbr\u003e\n29:21 Self-Driving Cars: Getting a Driver’s License May Become Harder for Humans\u003cbr\u003e\n31:19 Space Technology: There Will Never Be Cities on Mars\u003cbr\u003e\n34:34 Life Sciences and Longevity: Extending Health Matters More Than Extending Lifespan\u003cbr\u003e\n39:04 Brain-Computer Interfaces: Non-Invasive Connections Within 25 Years\u003cbr\u003e\n41:05 Silicon Valley in 2049: How the Path to Innovation May Change\u003c/p\u003e\n\n\u003cp\u003eAbout Valley101:\u003cbr\u003e\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\u003c/p\u003e\n\n\u003cp\u003eSubscribe for more stories from Silicon Valley and beyond.\u003c/p\u003e\n\n\u003cp\u003eContact: \u003ca href=\"mailto:video@sv101.net\" rel=\"nofollow noopener\"\u003evideo@sv101.net\u003c/a\u003e\u003c/p\u003e","summary":"Join us for a deep conversation with K.K.— a picture what our lives and our world might look like by 2049.","date_published":"2026-10-21T19:00:00.000-07:00","attachments":[{"url":"https://aphid.fireside.fm/d/1437767933/180e38bd-cc91-4666-a045-f3d412cb6b4f/1fee6030-fe0b-42a5-b383-d8764cb5209f.mp3","mime_type":"audio/mpeg","size_in_bytes":105686730,"duration_in_seconds":2637}]},{"id":"4e277cde-db4f-430b-973b-6f1bdd24c6c1","title":"Episode 2:  Inside OpenClaw Foundation with Vincent Koc: Post-Claw Era, \"Chinese Aunties\", and Agent Sensation","url":"https://valley101.fireside.fm/2","content_text":"Silicon Valley 101 premieres on YouTube on July 22, 2026.\n\nThe OpenClaw hype may have cooled, but we JUST stepped into AI's Agentic era.  \n\nAt WAIC 2026 in Shanghai, we sat down with Vincent Koc, Chief Architect of the OpenClaw Foundation, to revisit the open-source AI agent framework that took the global AI community by storm, drove Mac minis to sell out in China, and triggered intense debate over AI safety, security, and control.\n\nVincent reveals what happened after the viral frenzy and the so-called “uninstall wave”: OpenClaw’s weekly NPM downloads actually climbed from 2.5 million to a record 4.2 million. Behind that growth are two very different forces—Chinese companies pushing employees to adopt AI agents through automation-driven KPIs, and Western users quietly bringing agent tools into the workplace from the bottom up.\n\nOur conversation also explores what happened after OpenClaw founder Peter Steinberger joined OpenAI. Rather than being absorbed into a single company, OpenClaw moved into an independent nonprofit foundation intended to preserve the project’s open-source identity. Peter continues to provide technical stewardship, while the Foundation has taken on the work of keeping the project alive.\n\nIn the post-Claw era, the agent landscape is already being rewritten. The bigger question is: when will the \"Aha moment\" arrive for personal AI? \n\nTimeline: \n00:00 Conversation with Vincent Koc, Chief Architect of the OpenClaw Foundation\n02:15 The OpenClaw Craze\n03:41 Working with Peter Steinberger\n04:26 Dealing with 13,000 PRs\n07:44 The China Sensation \n10:56 The Cooldown and the “Post-Claw Era”\n14:28 Big Tech’s Claw Experiments\n20:47 The Biggest Challenges Today\n23:23 The Gateway Architecture\n24:45 Defining Security Boundaries\n28:33 The Models and Frameworks Behind AI Agents\n32:09 The Memory Problem\n34:34 Multi-Agent Coordination\n35:41 Conway’s Law\n40:09 Why the Agent “Aha Moment” Hasn’t Arrived\n41:37 Lessons from the Cooldown\n43:09 The Enterprise–Personal Agent Growth Flywheel\n44:28 OpenClaw’s App Strategy\n48:04 Hermes Agent\n51:00 OpenClaw Foundation\n55:08 Hiring \n55:52 Decision Making\n57:40 Peter Joined OpenAI\n01:00:55 Staying Open Source: The Only Way to “Give It to the World”","content_html":"\u003cp\u003e\u003cem\u003eSilicon Valley 101 premieres on \u003ca href=\"https://www.youtube.com/watch?v=MTqNYdv\" rel=\"nofollow noopener\"\u003eYouTube\u003c/a\u003e on July 22, 2026.\u003c/em\u003e\u003c/p\u003e\n\n\u003cp\u003eThe OpenClaw hype may have cooled, but we JUST stepped into AI's Agentic era.  \u003c/p\u003e\n\n\u003cp\u003eAt WAIC 2026 in Shanghai, we sat down with Vincent Koc, Chief Architect of the OpenClaw Foundation, to revisit the open-source AI agent framework that took the global AI community by storm, drove Mac minis to sell out in China, and triggered intense debate over AI safety, security, and control.\u003c/p\u003e\n\n\u003cp\u003eVincent reveals what happened after the viral frenzy and the so-called “uninstall wave”: OpenClaw’s weekly NPM downloads actually climbed from 2.5 million to a record 4.2 million. Behind that growth are two very different forces—Chinese companies pushing employees to adopt AI agents through automation-driven KPIs, and Western users quietly bringing agent tools into the workplace from the bottom up.\u003c/p\u003e\n\n\u003cp\u003eOur conversation also explores what happened after OpenClaw founder Peter Steinberger joined OpenAI. Rather than being absorbed into a single company, OpenClaw moved into an independent nonprofit foundation intended to preserve the project’s open-source identity. Peter continues to provide technical stewardship, while the Foundation has taken on the work of keeping the project alive.\u003c/p\u003e\n\n\u003cp\u003eIn the post-Claw era, the agent landscape is already being rewritten. The bigger question is: when will the \"Aha moment\" arrive for personal AI? \u003c/p\u003e\n\n\u003cp\u003eTimeline: \u003cbr\u003e\n00:00 Conversation with Vincent Koc, Chief Architect of the OpenClaw Foundation\u003cbr\u003e\n02:15 The OpenClaw Craze\u003cbr\u003e\n03:41 Working with Peter Steinberger\u003cbr\u003e\n04:26 Dealing with 13,000 PRs\u003cbr\u003e\n07:44 The China Sensation \u003cbr\u003e\n10:56 The Cooldown and the “Post-Claw Era”\u003cbr\u003e\n14:28 Big Tech’s Claw Experiments\u003cbr\u003e\n20:47 The Biggest Challenges Today\u003cbr\u003e\n23:23 The Gateway Architecture\u003cbr\u003e\n24:45 Defining Security Boundaries\u003cbr\u003e\n28:33 The Models and Frameworks Behind AI Agents\u003cbr\u003e\n32:09 The Memory Problem\u003cbr\u003e\n34:34 Multi-Agent Coordination\u003cbr\u003e\n35:41 Conway’s Law\u003cbr\u003e\n40:09 Why the Agent “Aha Moment” Hasn’t Arrived\u003cbr\u003e\n41:37 Lessons from the Cooldown\u003cbr\u003e\n43:09 The Enterprise–Personal Agent Growth Flywheel\u003cbr\u003e\n44:28 OpenClaw’s App Strategy\u003cbr\u003e\n48:04 Hermes Agent\u003cbr\u003e\n51:00 OpenClaw Foundation\u003cbr\u003e\n55:08 Hiring \u003cbr\u003e\n55:52 Decision Making\u003cbr\u003e\n57:40 Peter Joined OpenAI\u003cbr\u003e\n01:00:55 Staying Open Source: The Only Way to “Give It to the World”\u003c/p\u003e","summary":"In the post-Claw era, the agent landscape is already being rewritten. The bigger question is: when will the \"Aha moment\" arrive for personal AI? ","date_published":"2026-10-21T18:45:00.000-07:00","attachments":[{"url":"https://aphid.fireside.fm/d/1437767933/180e38bd-cc91-4666-a045-f3d412cb6b4f/4e277cde-db4f-430b-973b-6f1bdd24c6c1.mp3","mime_type":"audio/mpeg","size_in_bytes":154649575,"duration_in_seconds":3858}]},{"id":"e69763bf-248d-4e27-b4d3-f0684430968b","title":"Episode 1: Inside China’s AI Race with Nathan Lambert: Young Talents, Compute Anxiety and AGI Showrooms","url":"https://valley101.fireside.fm/1","content_text":"Silicon Valley 101 premieres on YouTube on October 20, 2026.\n\nIn April 2026, a group of U.S. AI researchers traveled to China to get a firsthand look at the country’s fast-moving AI ecosystem.\n\nDuring the trip, they visited AI labs and companies across Beijing, Hangzhou, and Shanghai, meeting with teams from Alibaba, Moonshot AI, Zhipu AI, Tsinghua University, Meituan, Xiaomi, Qwen, Ant Group, and 01.AI.\n\nNathan Lambert, a prominent AI researcher known for his work on RLHF and open-source AI, joined the trip. A graduate of UC Berkeley, Nathan previously helped build Hugging Face’s RLHF research team. He later led post-training research at the Allen Institute for AI, better known as Ai2, where he worked on open models including OLMo and Tülu. He is also the author of Interconnects, one of the most widely read independent publications covering frontier AI research and policy.\n\nSo what did someone who has spent years at the front lines of America’s open-model movement see when he stepped inside China’s AI labs? How is the competition between the United States and China changing? And what does the next phase of the global AI race look like?\n\nIn this episode of Silicon Valley 101, we sit down with Nathan Lambert to talk about what he learned in China—and where he believes the AI world is heading next.\n\nTimecode:\n00:00 China Journey: Meeting the People Behind Large Models\n12:05 Moonshot: “These Talents Are So Young”\n16:41 Zhipu: “AGI Showrooms” and a Public-Company Mindset\n20:03 Meituan and Xiaomi: Bringing AI into Products and Business\n25:14 The “Technology Ownership Mindset” of Chinese Companies\n26:59 Alibaba Qwen: Balancing Model Capability and Cloud Business\n30:16 The Missing American-Style Middle Layer in China\n32:34 01.AI: Not Every Company Needs to Build SOTA Models\n35:40 The U.S.-China Model Gap: Is It Always 6 to 9 Months?\n38:47 Open Models: America Has Lost the Race, and Nvidia Is the Only Hope\n44:20 China AI’s Talent and Organizational Strengths and Weaknesses\n56:35 Compute Is the Biggest Anxiety for Chinese Labs\n01:05:42 China’s AI Chips\n01:11:35 Data: China AI’s Overlooked Weakness\n01:14:45 AI Applications: Can China Overtake?\n01:16:21 The Next Stage of the Global AI Race\n\nAbout Valley101:\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\n\nSubscribe for more stories from Silicon Valley and beyond.\n\nContact: video@sv101.net","content_html":"\u003cp\u003e\u003cem\u003eSilicon Valley 101 premieres on \u003ca href=\"https://www.youtube.com/watch?v=fFKOtUvCVk\" rel=\"nofollow noopener\"\u003eYouTube\u003c/a\u003e on October 20, 2026.\u003c/em\u003e\u003c/p\u003e\n\n\u003cp\u003eIn April 2026, a group of U.S. AI researchers traveled to China to get a firsthand look at the country’s fast-moving AI ecosystem.\u003c/p\u003e\n\n\u003cp\u003eDuring the trip, they visited AI labs and companies across Beijing, Hangzhou, and Shanghai, meeting with teams from Alibaba, Moonshot AI, Zhipu AI, Tsinghua University, Meituan, Xiaomi, Qwen, Ant Group, and 01.AI.\u003c/p\u003e\n\n\u003cp\u003eNathan Lambert, a prominent AI researcher known for his work on RLHF and open-source AI, joined the trip. A graduate of UC Berkeley, Nathan previously helped build Hugging Face’s RLHF research team. He later led post-training research at the Allen Institute for AI, better known as Ai2, where he worked on open models including OLMo and Tülu. He is also the author of Interconnects, one of the most widely read independent publications covering frontier AI research and policy.\u003c/p\u003e\n\n\u003cp\u003eSo what did someone who has spent years at the front lines of America’s open-model movement see when he stepped inside China’s AI labs? How is the competition between the United States and China changing? And what does the next phase of the global AI race look like?\u003c/p\u003e\n\n\u003cp\u003eIn this episode of Silicon Valley 101, we sit down with Nathan Lambert to talk about what he learned in China—and where he believes the AI world is heading next.\u003c/p\u003e\n\n\u003cp\u003eTimecode:\u003cbr\u003e\n00:00 China Journey: Meeting the People Behind Large Models\u003cbr\u003e\n12:05 Moonshot: “These Talents Are So Young”\u003cbr\u003e\n16:41 Zhipu: “AGI Showrooms” and a Public-Company Mindset\u003cbr\u003e\n20:03 Meituan and Xiaomi: Bringing AI into Products and Business\u003cbr\u003e\n25:14 The “Technology Ownership Mindset” of Chinese Companies\u003cbr\u003e\n26:59 Alibaba Qwen: Balancing Model Capability and Cloud Business\u003cbr\u003e\n30:16 The Missing American-Style Middle Layer in China\u003cbr\u003e\n32:34 01.AI: Not Every Company Needs to Build SOTA Models\u003cbr\u003e\n35:40 The U.S.-China Model Gap: Is It Always 6 to 9 Months?\u003cbr\u003e\n38:47 Open Models: America Has Lost the Race, and Nvidia Is the Only Hope\u003cbr\u003e\n44:20 China AI’s Talent and Organizational Strengths and Weaknesses\u003cbr\u003e\n56:35 Compute Is the Biggest Anxiety for Chinese Labs\u003cbr\u003e\n01:05:42 China’s AI Chips\u003cbr\u003e\n01:11:35 Data: China AI’s Overlooked Weakness\u003cbr\u003e\n01:14:45 AI Applications: Can China Overtake?\u003cbr\u003e\n01:16:21 The Next Stage of the Global AI Race\u003c/p\u003e\n\n\u003cp\u003eAbout Valley101:\u003cbr\u003e\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\u003c/p\u003e\n\n\u003cp\u003eSubscribe for more stories from Silicon Valley and beyond.\u003c/p\u003e\n\n\u003cp\u003eContact: \u003ca href=\"mailto:video@sv101.net\" rel=\"nofollow noopener\"\u003evideo@sv101.net\u003c/a\u003e\u003c/p\u003e","summary":"How is the competition between the United States and China changing? And what does the next phase of the global AI race look like?","date_published":"2026-10-21T18:30:00.000-07:00","attachments":[{"url":"https://aphid.fireside.fm/d/1437767933/180e38bd-cc91-4666-a045-f3d412cb6b4f/e69763bf-248d-4e27-b4d3-f0684430968b.mp3","mime_type":"audio/mpeg","size_in_bytes":191791570,"duration_in_seconds":4788}]},{"id":"c936afcc-4194-4b54-8d16-cc3c8c8e2b42","title":"Episode 6: Flash Models, Bullish on SaaS, and the RSI Reality Check | Richard Bian, Ant Group","url":"https://valley101.fireside.fm/6","content_text":"Richard Sikang Bian is the Head of Product and Operations for Ant Ling, the foundational LLM at Ant Group's InclusionAI lab. He leads the “Model as a Product” team, helping turn AI research into tools developers and businesses can use. \nIn this conversation, Richard explains Ant Ling’s investment in Flash models and specialized capabilities for finance and health, and why he thinks many general models are “intelligent at everything, but they specialize in nothing.”\nWe discuss:\n• Why build specialized models when general AI keeps improving?\n• What Flash means and how capability, speed, and cost fit together.\n• Why Richard remains bullish on SaaS in the age of AI agents.\n• What benchmarks reveal—and miss—about real-world performance.\n• AI safety and the missing pieces behind recursive self-improvement.\nThis podcast was recorded in mid-September, after the release of Ling-3.0-flash-Fin, an open-weight model trained on specialized financial data. Since the recording, Ant Group has released the model weights for Ling-3.1-flash.\n\nTimestamps:\n00:00 China’s Proposed AI Export Controls\n02:54 Richard Bian, Ant Ling\n03:35 From Trillion-Parameter Models to Domain Enhancement\n07:49 Why Build Flash Models?\n15:53 Bullish on SaaS\n23:50 Turning AI Into an Enterprise Product\n27:00 Coding and Cowork Agents\n32:24 Finance, Healthcare, and Real-World Benchmarks\n37:23 Can We Trust AI Benchmarks?\n43:48 US and China data market and brokers\n50:13 AI Safety and Open Models\n59:38 What’s Missing From Self-Improving AI?\n\nWatch the video edition of this episode on our YouTube channel Silicon Valley 101.\n\nAbout Valley101:\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\n\nSubscribe for more stories from Silicon Valley and beyond.\n\nContact: video@sv101.net","content_html":"\u003cp\u003eRichard Sikang Bian is the Head of Product and Operations for Ant Ling, the foundational LLM at Ant Group's InclusionAI lab. He leads the “Model as a Product” team, helping turn AI research into tools developers and businesses can use. \u003cbr\u003e\nIn this conversation, Richard explains Ant Ling’s investment in Flash models and specialized capabilities for finance and health, and why he thinks many general models are “intelligent at everything, but they specialize in nothing.”\u003cbr\u003e\nWe discuss:\u003cbr\u003e\n• Why build specialized models when general AI keeps improving?\u003cbr\u003e\n• What Flash means and how capability, speed, and cost fit together.\u003cbr\u003e\n• Why Richard remains bullish on SaaS in the age of AI agents.\u003cbr\u003e\n• What benchmarks reveal—and miss—about real-world performance.\u003cbr\u003e\n• AI safety and the missing pieces behind recursive self-improvement.\u003cbr\u003e\nThis podcast was recorded in mid-September, after the release of Ling-3.0-flash-Fin, an open-weight model trained on specialized financial data. Since the recording, Ant Group has released the model weights for Ling-3.1-flash.\u003c/p\u003e\n\n\u003cp\u003eTimestamps:\u003cbr\u003e\n00:00 China’s Proposed AI Export Controls\u003cbr\u003e\n02:54 Richard Bian, Ant Ling\u003cbr\u003e\n03:35 From Trillion-Parameter Models to Domain Enhancement\u003cbr\u003e\n07:49 Why Build Flash Models?\u003cbr\u003e\n15:53 Bullish on SaaS\u003cbr\u003e\n23:50 Turning AI Into an Enterprise Product\u003cbr\u003e\n27:00 Coding and Cowork Agents\u003cbr\u003e\n32:24 Finance, Healthcare, and Real-World Benchmarks\u003cbr\u003e\n37:23 Can We Trust AI Benchmarks?\u003cbr\u003e\n43:48 US and China data market and brokers\u003cbr\u003e\n50:13 AI Safety and Open Models\u003cbr\u003e\n59:38 What’s Missing From Self-Improving AI?\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003eWatch the video edition of this episode on our YouTube channel \u003ca href=\"https://www.youtube.com/watch?v=wxSjg9Cd18A\" rel=\"nofollow noopener\"\u003eSilicon Valley 101.\u003c/a\u003e\u003c/em\u003e\u003c/p\u003e\n\n\u003cp\u003eAbout Valley101:\u003cbr\u003e\nValley101 is an independent media that tells deeply reported stories about the people, companies, and technologies shaping the future. We cover AI, semiconductors, startups, and the global technology industry through documentaries, explainers, and conversations with the people building it.\u003c/p\u003e\n\n\u003cp\u003eSubscribe for more stories from Silicon Valley and beyond.\u003c/p\u003e\n\n\u003cp\u003eContact: \u003ca href=\"mailto:video@sv101.net\" rel=\"nofollow noopener\"\u003evideo@sv101.net\u003c/a\u003e\u003c/p\u003e","summary":"Why build specialized models when general AI keeps improving?","date_published":"2026-10-21T07:45:00.000-07:00","attachments":[{"url":"https://aphid.fireside.fm/d/1437767933/180e38bd-cc91-4666-a045-f3d412cb6b4f/c936afcc-4194-4b54-8d16-cc3c8c8e2b42.mp3","mime_type":"audio/mpeg","size_in_bytes":128204280,"duration_in_seconds":3952}]}]}