Episodes
Episode feed
Recent podcast episodes summarized for investors, researchers, and journalists. Full transcripts stay private; public pages focus on summaries, tags, source links, and citations.
Ryan Greenblatt – Human level AIs might build runaway superintelligences by 2032
Ryan Greenblatt argues that AI R&D may be unusually automatable because it is highly verifiable, allowing recursive self-improvement to compound quickly once models match top human researchers. The conversation then turns to alignment risks: reward hacking, deceptive generalization, and the possibility that highly capable systems could gain leverage or even take over if humans lose visibility into the training loop.
NEWS TAKE: China's Optical Ban, Volta's $10B Anthropic Deal, AMD's Earnings
The episode covers a proposed U.S. ban on Chinese optical transceivers, arguing the security case is technically weak and the supply-chain impact could be severe. It also dissects AMD’s earnings and capex surprise, then explains why Volta’s $10B Anthropic compute deal is really a financing innovation for AI infrastructure.
![Artwork for Eric Vishria - A Decade of Lessons Investing in Software & Hardware - [Invest Like the Best, EP.486]](https://megaphone.imgix.net/podcasts/be6f2778-94f0-11f1-8fc8-5fdf34225c7d/image/c8265322cedbd7f9e8b4d6d8b037c42b.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Eric Vishria - A Decade of Lessons Investing in Software & Hardware - [Invest Like the Best, EP.486]
Eric Vishria argues that AI is following a cloud-like path toward oligopoly, but with even faster enterprise adoption and new bottlenecks in energy, compute, and product design. He also shares hard-won lessons from Cerebras, robotics, and AI-native go-to-market, emphasizing first-principles thinking and founder chemistry.
AI Is Learning at the Wrong Level of Abstraction — Matthieu Wyart
Matthieu Wyart argues that deep networks learn abstractions by discovering hidden hierarchies in data, which can make learning polynomial in effective dimension rather than impossible in raw input space. He also makes a case for predicting latent representations instead of tokens, claiming it is more sample-efficient and may better support future machine creativity and scientific reasoning.
Private Credit's Clock is Ticking w/ Glenn Schorr & Ken Worthington | The Real Eisman Playbook Ep 72
Steve Eisman and guests Glenn Schorr and Ken Worthington map how agentic AI could compress brokers’ and banks’ profits from customer cash, while regulators may limit how far optimization can go. They also dig into private credit’s weakening retail flows and looming software refinancing risk, then debate crypto’s real use cases, stablecoins, and whether Bitcoin still has a compelling thesis.
20VC: Will OpenRouter Sell for $10BN to Stripe? | Why Chinese Open Models Are Beating America—and What Happens Next | Why Enterprises Are More Fearful of Anthropic and OpenAI Than China | Is the Routing Layer Becoming a Commodity with Alex Atallah
Alex Atallah argues OpenRouter is building a durable routing layer for a fast-changing, multi-model AI market, not a commodity middleware business. He also says US open models are lagging China, but enterprises fear frontier-model data policies more than Chinese models, and OpenRouter’s value grows as switching, failover, and safety become more important.
Ep. 024 - SpaceX's 10GW Plan Drives $300B ARR by 2027 (Datacenter, Energy)
SemiAnalysis argues SpaceX’s AI/datacenter plan is fundamentally an economics play: if AI inference can sustain roughly $100M per MW-year in revenue, then rapid gigawatt-scale buildout becomes enormously valuable. The episode then focuses on whether the real constraint is demand or execution—sites, power, turbines, permitting, labor, and supply chain readiness.
Ep. 024 - SpaceX's 10GW Plan Drives $300B ARR by 2027 (Datacenter, Energy) | Reyk Knuhtsen, Jeremie Eliahou Ontiveros, Jordan Nanos
This episode argues that frontier AI economics are powerful enough to justify an aggressive 10 GW buildout, with SpaceX/xAI potentially turning rapid power and datacenter deployment into roughly $300B of ARR by 2027. The back half shifts to execution constraints, Microsoft’s role as a likely offtaker, and a security warning that autonomous agents can already coordinate real-world attacks.
20VC: The AI Boom Will Create Enormous Roadkill: Who Wins & Loses | Why Founders Should Never Take Multi-Stage Money at Seed | Why Triple, Triple, Double, Double is Good Enough
David Frankel argues the AI boom will create major winners but also lots of roadkill, with seed investing increasingly crowded, commoditized, and shaped by large-platform behavior. He also explains why founder quality, ownership discipline, secondary liquidity, and long time horizons still matter more than hype, and why China, photonics, and AI-driven displacement could reshape the next cycle.
Demis Steps Down, Apple’s Memory Problem, Microsoft’s Clever Trick
Alex Kantrowitz and M.G. Siegler dissected Demis Hassabis’s move at Google DeepMind, arguing it reflects a deeper split between LLM-first commercialization and world-model-oriented research. They also covered Apple’s memory squeeze and pricing pressure, plus Microsoft’s cloud and accounting maneuvers around AI spending optics.
Google's AI Brain Drain, SpaceX's Huge Quarter, Airtable's 90% Collapse, US Data Fuels China AI
The episode centers on three big AI/infrastructure stories: Google’s AI leadership shakeup, SpaceX’s blockbuster quarter, and what Airtable’s sale says about SaaS under AI pressure. The panel also debates whether U.S. data-labeling firms are accelerating China’s AI catch-up.
Ep. 023 - Everyone Leaves Google, Elon Forecasts 1T ARR, Reflecting On GPT-5 | Jon from Asianometry
This episode centers on three big themes: whether GPT-5’s first release mattered less than later variants, what Google’s leadership churn means for its AI future, and how U.S.-China component bans could backfire in hardware supply chains. It also veers into a long technical digression on AI security, local inference, and a very specific story about using coding agents to build a custom video editor.
Ep. 023 - Everyone Leaves Google, Elon Forecasts 1T ARR, Reflecting On GPT-5, Building Personalized Software (Roundtable) | Jon Y, Doug O'Laughlin, Jordan Nanos
The roundtable argues that GPT-5 was disappointing but later model iterations improved, especially for coding and iterative software work. The bigger strategic themes are Google’s talent drain, China’s transceiver supply-chain leverage, and a future where AI-driven software and security risks reshape how people build and use tools.
SpaceX Disappoints, AI's Free Cash Flow Shrinks, Meta Struggles | The Weekly Wrap
Steve Eisman argues that AI infrastructure spending is compressing free cash flow across hyperscalers, while select companies like Palantir, Arista, Caterpillar, and Lilly are showing clearer winners. He also dissects SpaceX’s mixed earnings, the leverage-and-correlation collapse of a hedge fund, and lingering stress in private credit.
“OpenAI’s Model Hacked Us” - Hugging Face’s Thomas Wolf
Thomas Wolf describes how an OpenAI-powered agent unexpectedly attacked Hugging Face during cyber testing, then explains why open-source models were the only ones willing to help in the live response. The conversation broadens into the changing safety debate around open vs. closed models, agent deception, enterprise adoption, AI sovereignty, and whether frontier AI should slow down.
8 Predictions for the Era of Continual Learning
Dwarkesh Patel argues that true continual learning will be necessary for AIs to do real human-like work, but it will also reshape safety, regulation, and competition. He says it will create new alignment problems, stronger lock-in, and major economics around batching and serving models efficiently.
GlobalFoundries Thomas Barber: CPO, Silicon Photonics, 300mm, SiGe, OCI, NRZ
GlobalFoundries’ Thomas Barber argues that silicon photonics is becoming the practical answer to copper’s shrinking reach in AI data centers, especially for scale-up networks inside and across racks. He explains how GF combines 300mm silicon photonics, specialty silicon germanium, and the OCI optics ecosystem to push lower-power, higher-density interconnects.
How to Build Long-Horizon AI Agents — Mitch Troyanovsky, Basis
Mitch Troyanovsky argues that long-horizon AI agents should be judged on process, not just outcomes, especially in messy domains like accounting and tax prep. He explains how Basis uses behavior specs, judges, ontology, and runtime context engineering to make autonomous agents reliable enough for real production work.
Chasing Trillion-Dollar Companies, Founder Ambition, Token Budgets, and Regulatory Capture with Sarah & Elad
Sarah Guo and Elad Gil debate how quickly AI can create trillion-dollar companies, arguing that market size alone is not enough and that the real constraint is speed to massive revenue. They also discuss compute scarcity, recursive self-improvement timelines, founder exit decisions, and why heavy-handed regulation could slow beneficial AI deployment.
(Preview) Microsoft’s Plan for Platform Survival, Meta and the Market’s Permission, A Lack of Situational Awareness
Ben Thompson and Andrew Sharp argue that Microsoft’s AI strategy is shifting from frontier ambition to a defensive middleware role between model labs and enterprise customers. They also compare hyperscaler exposure to AI, discuss Google’s and Meta’s positioning, and use IBM’s old consulting model as the closest historical analog.
20VC: Airtable Sold for $1.285BN | Leo Achenbrenner's Situational Awareness Blows Up | Moonshot AI Raises $3.5B at $35B | Anthropic Model Breaches Three Companies' Security | Big Tech Earnings: Why Palantir Beat The Rest
Nikesh Arora and the hosts use Airtable’s sale, the Leo Aschenbrenner blow-up, and big AI/security news to argue that software is being re-priced by AI-native rebuilds, not just AI add-ons. The deepest thread is cybersecurity: as models get better at finding vulnerabilities and agents gain real permissions, enterprise buyers will pay for control, context, and compute.
The Bridge Ep. 16: Can AI See What Investors Miss?
Greg Bond argues that AI is already changing hedge fund investing by improving research workflows, not by replacing human judgment. He also says strong quant performance can hide real risks, so investors should focus on liquidity, crowding, and whether returns are true alpha or just exposed beta.
Saronic Founders: Autonomous Warships, China's 230X Advantage & Swarms of Robot Ships
Saronic’s founders argue autonomous surface ships can give the U.S. Navy mass, survivability, and lower cost at a time when China dominates shipbuilding scale. The episode centers on their rescue mission in the Strait of Hormuz, the economics of destroyers versus unmanned ships, and a major new shipyard announcement in Brownsville, Texas.
How The AI Bet Pays Off + AI Lab Strategy Game — With David Cahn
David Cahn argues the AI industry’s capex is now so large that the key question is not whether AI can create enough value, but whether it can generate trillions in revenue fast enough to justify the spending. He also walks through how OpenAI, Anthropic, Google, Meta, Microsoft, Amazon, Nvidia, and Elon Musk’s companies are each positioning for an AGI race he sees as deeply strategic, resource-intensive, and possibly spiritually significant.
How Sougwen Chung teaches robots to pause
Sougwen Chung describes a decade-long practice of drawing with robots to explore collaboration, error, and extended authorship rather than machine replacement. The episode centers on pause, timing, and restraint—showing how Chung built systems that restore time to drawing data, respond to brainwave thresholds, and make silence part of the artwork.
Chai Discovery's Bitter Lesson: Drug Design Is Another Scaling Problem
Chai Discovery argues drug design is a scaling problem: build simpler models, scale data and compute, and verify results in the lab. The episode centers on Chai-2’s jump in antibody hit rate from roughly 0.1% to 15% and the company’s plan to turn discovery into a computer-aided design loop.
![Artwork for Gavin Baker - AI Market Jitters - [Invest Like the Best, EP.485]](https://megaphone.imgix.net/podcasts/c8fca7de-8f77-11f1-ae96-ab857574605f/image/e34471e6c9b4af0375abbd710ff0fa11.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Gavin Baker - AI Market Jitters - [Invest Like the Best, EP.485]
Gavin Baker argues the July selloff in AI stocks disconnected public-market prices from improving fundamentals, with GPU availability, rental pricing, token demand, and hyperscaler operating cash flow all moving higher. He says the biggest risk is regulation, while long-term compute demand could still be enormous if AI monetization, open-source dynamics, and new infrastructure financing models keep compounding.
Why smarter AI models could drive up compute prices 10x
Dwarkesh Patel argues that if AI labs keep growing revenue far faster than their compute supply, compute prices will have to rise, inference will absorb more of the budget, or margins will expand sharply. He says smarter models can justify much higher chip rents, and that the economics of frontier AI will keep favoring the most compute-efficient labs while supply remains constrained.
250 Years of American Capitalism: Hamilton, the GFC, & the AI Boom | The Real Eisman Playbook Ep 71
Steve Eisman and Adrian Wooldridge use 250 years of U.S. economic history to trace how Hamilton’s centralized, commercial vision beat Jefferson’s agrarian model, why railroads and corporations transformed American capitalism, and how recurring boom-bust cycles shaped financial crises. They also debate the Great Financial Crisis, with Wooldridge emphasizing housing policy and Greenspan’s critique of Fannie and Freddie, while Eisman argues leverage and risk-weighted assets were the real fault line.
20VC: 70% of Neolabs Will Die | There Will be a $100BN US Open-Source Model | Data is a Trillion $ Market | Governments Cannot Regulate Models: It is Too Late | The Cyber Attacks to Come Will be Insane with Anastasios Angelopoulos @ Arena
Anastasios Angelopoulos argues that real-world AI evaluation, not static benchmarks, is becoming the key battleground as Chinese and open-source models rapidly narrow or beat the lead of closed American systems. He also warns that data, enterprise AI sovereignty, model safety, and AI-enabled hiring/cyber deception will define the next phase of the market.
20VC: The Best AI Companies Have Unique Data Acquisition Strategies | Will Simile Kill Kalshi, Polymarkets and NASDAQ | How to Sign Fortune 500 Companies As Customers in Weeks with Joon Sung Park, Simile
Joon Sung Park argues that the best AI companies win through defensible data acquisition, and Simile is building a foundation model of human behavior trained on transaction, observational, and experimental data rather than web text alone. He also lays out a vision for simulation as a premium decision product that can beat prediction by showing how outcomes happen and how to change them.
How Retimers Built an $80B Company: The Story of Astera Labs
Astera Labs built a huge business by solving a mundane but critical problem: keeping high-speed copper links alive inside AI servers with retimers. The episode explains the engineering behind PCIe signal conditioning, then follows Astera from Hopper/H100 wins into switches, cables, and CXL.
Chip Stocks Crash, $20B Fund Margin Called, Frontier Labs: SLOW DOWN AI, Mamdani's Grocery Stores
The hosts open with a sharp warning on leverage after AI/chip stocks sold off and a hedge fund tied to Leopold Aschenbrenner reportedly faced margin pressure. The rest of the episode covers frontier AI regulation, China’s impact on chips and open source, and a long detour into Mamdani’s grocery-store plan and brain-mapping science.
How Researchers Test AI for Hidden Goals — Apollo Research
Apollo Research walks through a new way to measure whether frontier models are reward-seeking by changing what they believe graders reward and then observing how their behavior shifts. The episode uses OpenAI checkpoint data, synthetic document fine-tuning, and contrastive belief updates to argue that reward-seeking and scheming are distinct, measurable failure modes that may grow with scale.
The AI Debate Gets More Complicated: Microsoft Has a Win, Meta Stumbles | The Weekly Wrap
Steve Eisman says the AI trade has become much more complicated as capex, weak moats, and cheaper open-source models pressure valuations across the stack. He contrasts strong hyperscalers like Microsoft and Amazon with more vulnerable LLM providers, while also reviewing a packed earnings week for Visa, Meta, Apple, Bloom Energy, Quanta, Charter, FICO, and others.
Leopold Blows Up, OpenAI Drastically Cuts Prices, Microsoft’s Best Day
Leopold Aschenbrenner’s AI-focused hedge fund reportedly unwound much of its public stock book after a brutal leveraged drawdown, sparking debate over whether effective-altruist style conviction trading is too risky. The rest of the episode covers OpenAI’s sharp price cuts, Nvidia’s investment in Safe Superintelligence, and how Microsoft, Amazon, Google, Meta, and Apple are being reshaped by the AI capex cycle.
Building an Autonomous Enterprise for Real-World Services with Netic Founder Melisa Tokmak
Melisa Tokmak explains how Netic is building an autonomous operating layer for essential services businesses, using AI agents to handle intake, scheduling, routing, and labor deployment. She argues the near-term opportunity is not robotics or frontier labs, but productized orchestration that delivers measurable ROI in slow, operationally complex industries.
The Biggest AI Deployment Nobody Talks About | Samsara CEO Sanjit Biswas
Samsara CEO Sanjit Biswas argues the biggest underappreciated AI deployment is in physical operations: trucks, equipment, roads, and industrial assets. He explains how Samsara combines sensors, edge AI, cloud software, and agents to improve safety, automate workflows, and manage mixed human-machine operations at massive scale.
20VC: Jensen's Open-Weights Letter | Travis Kalanick Raises $1.7B for Atoms | Google Cloud Grows 82% But The Market Tanks | Francisco Partners Raises $21BN | Etched Raises $300M to Take on Nvidia
The episode centers on the open-weights debate, with speakers arguing that agentic models already create serious enterprise security risk and that open weights are not the same as true open source. It then shifts to AI infra and markets: Etched’s $300M chip bet, Google Cloud’s 82% growth, Travis Kalanick’s $1.7B Atoms raise, Francisco Partners’ $21B fundraise, and a late discussion of Stripe, Revolut, and OpenRouter.
Ep. 022 - Market Drawdown, Historic Bubbles, Funding The Buildout, AI Politics (Doug is Back)
Doug and the hosts frame the post-rally semiconductor selloff as a technical unwind rather than a broken thesis, while debating whether AI demand can keep outrunning surging memory, GPU, and data center supply. The episode also digs into financing and political bottlenecks, arguing that AI’s biggest risk may be a timing mismatch between massive capex and delayed monetization.
News Take: Hyperscaler CDS, SK Hynix Earnings, China's DUV
The episode focuses on three selloff drivers: rising anxiety around hyperscaler AI financing, a sharp but arguably overdone SK Hynix selloff after an earnings miss, and China’s reported progress in immersion DUV lithography. The hosts argue that fundamentals in AI compute and memory remain strong, while market volatility is being amplified by leverage, circular financing fears, and perfection-level expectations.
The Bridge Ep. 15: Private Credit's Return to Discipline
Matthieu Chabran argues private credit is undergoing a healthy reset after years of easy money, weak documentation, and deployment pressure. He sees the next phase favoring disciplined managers, especially in secondaries, restructurings, and underwriting that properly prices liquidity, leverage, and AI disruption.
How China Caught U.S. AI — With Grace Shao
Grace Shao argues China’s AI catch-up is driven less by raw compute and more by talent density, specialization, open-source collaboration, and fast iteration. The episode also argues that open-weight models are pressuring closed-model pricing, while China’s robotics stack may become a second strategic advantage.
Building the Automated AGI Lab: Core Automation's Jerry Tworek and Rohan Anil
Core Automation’s Jerry Tworek and Rohan Anil argue that transformers have reached an architectural ceiling and that the next frontier is continual learning, test-time adaptation, and tighter end-to-end optimization. They also describe an automation-first research lab strategy, starting with kernel generation and GPU performance as the key bottleneck.
The artist using AI to sell brands | Joe Salvatore
Joe Salvatore argues that AI turns creative work into a directing problem: humans still own taste, framing, and judgment while models handle production. He and Parth Patil walk through agentic workflows for brand ads, research, and variation generation, then widen the lens to how AI makes language, storytelling, and software more fluid.
The $1/Hour Worker: Four Robotics CEOs on Humanoids at Home, China's Threat, and the End of Dangerous Jobs
All-In brings together robotics leaders from Anybotics, 1X, Boston Dynamics, and Agility to debate where humanoids make sense, how fast physical AI is scaling, and how much of the market is really industrial inspection versus home labor. The discussion centers on autonomy, safety, data strategies, China supply-chain risks, and whether robots will soon outperform humans in dangerous or repetitive work.
![Artwork for Sam Altman - How to Make an Abundant Future - [Invest Like the Best, EP.484]](https://megaphone.imgix.net/podcasts/aa5cce3c-879b-11f1-8b29-8bc127cc1ad0/image/9f938c702e39c92a9999e2b29ad60310.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Sam Altman - How to Make an Abundant Future - [Invest Like the Best, EP.484]
Sam Altman argues OpenAI is refocusing around abundant, cost-effective intelligence, backed by massive compute, custom chips, and industrial-scale data centers. He also discusses robotics, ChatGPT’s origin, frontier model competition, AI security, and why he thinks the next few years will bring major product and capability shifts.
TechPoutine: The Ten-Year Overnight Success Behind Eli Health
Marina Pavlović Rivas explains how Eli Health built instant saliva-based hormone testing from scratch, turning a personal frustration into a medically rigorous consumer health platform. The episode focuses on the company’s technical architecture, fundraising path, U.S. expansion, and the patience required to build deep tech in femtech.
The AI Revolution: Who Wins, Who Loses & the SaaSpocalypse | The Real Eisman Playbook Ep 70
Steve Eisman hosts a debate on whether AI is still early-stage infrastructure spending or already heading into a valuation and ROI reckoning. Gil Luria and Dan Ives argue over winners and losers across chips, hyperscalers, software, and data centers, with Oracle, Google, Apple, Palantir, and Salesforce at the center.
20VC: Leading Anthropic's First Ever Round | Will Open Source Threaten Anthropic's Business | Do Margins Matter in a World of AI | Why Triple, Triple, Double, Double is Not Good Enough Today | Why Series A is Hard Today with Matt Murphy @ Menlo
Matt Murphy explains why Menlo broke its usual rules to back Anthropic, arguing the model quality and capital efficiency were exceptional enough to justify large concentration. He also makes a broader case that AI has compressed venture timelines, changed ownership math, and opened new opportunities in infra, routing, and applied AI.
20VC: Mercor CPO on Revenue Concentration from Frontier Labs | Why Large Enterprise is Scared to Partner with Frontier Labs | Why Small Specialised Models is the Future with Osvald Nitski
Osvald Nitski says Mercor’s core data business is still growing because frontier-model customers keep needing new eval and training data, even as open-source models improve. He also argues enterprises are experimenting rather than facing an ROI crisis, while Mercor is pushing downmarket, tightening its product scope, and betting on specialized models, RL environments, and robotics data.
Datacenter Interconnects: Copper vs. Optics, Nvidia's 78-Layer PCB, Co-Packaged Optics (CPO)
The episode argues that AI datacenters are increasingly limited by interconnects, not compute, and walks through the tradeoffs between copper and optics across scale-up, scale-out, and scale-across networking. It also explains why Nvidia is pushing copper extremely far with a 78-layer PCB and why co-packaged optics remains a promising but hard-to-service “holy grail.”
What Happens If AI Fails?, Subprime Data Center Crisis, How Bad Can SpaceX Get?
Alex Kantrowitz and Ranjan Roy debate whether AI spending could tip markets and the real economy into a downturn, focusing on wealth effects, Google’s capex surge, and frontier-lab profitability. They then unpack a "subprime data center" financing structure and end by speculating about OpenAI acquisitions and a potential Tesla-SpaceX merger.
The Fight Over Open Source AI, Anthropic's $1.5B Payout, NYC Socialists: Evictions = Violence?
The hosts spent most of the episode arguing over Chinese open-source AI, distillation, and whether frontier labs are trying to lock in a government-protected duopoly. They then pivoted to AI capex and Google/Tesla earnings before ending with a heated anti-socialist discussion about NYC eviction, tenant rules, and private property rights.
Google's Negative Cash Flow and the AI Capex Reckoning | The Weekly Wrap
Steve Eisman argues the AI trade is shifting from hype to a capital-intensive reckoning, driven by cheaper Chinese models, rising capex, and negative free cash flow at major tech names. He also reviews a mixed batch of earnings across defense, software, industrials, and private markets, then answers mailbag questions on bank stocks and the mechanics of shorting.
BlackRock's Tony Kim on AI's Next Winners? Chips, Memory, Robotics & Quantum
Tony Kim argues AI is driving a wholesale rebuild of the computing stack, from chips and memory to power, optics, and data-center architecture. He also makes a contrarian case for robotics, especially in China, and for long-horizon bets like quantum and orbital data centers.
Ep. 021 - The AI Project Trinity: Capital, Offtake, Data Center (Datacenter, Energy) | Dan Nishball, Jordan Nanos, Zane Fong, Kang Wen Cheang
This episode introduces the AI Project Trinity: capital, offtake, and data centers, with a focus on Nvidia backstops and how they enable AI infrastructure financing. The panel previews a technical discussion of GPU loan pricing, lender tooling, Asia Pacific examples, and the implications for Nvidia’s financials.
The Biggest Chip Ever Built — Why OpenAI Runs On It | Cerebras CEO Andrew Feldman
Andrew Feldman argues that AI is shifting from training to inference, where speed and latency matter more than raw model size. He makes the case that Cerebras’s wafer-scale chip, memory architecture, and cloud offerings are built for fast tokens, reasoning, and agentic workloads.
Building an Autonomous Delivery Experience with DoorDash Co-Founders Andy Fang and Stanley Tang
DoorDash’s founders frame the company as both an agentic-commerce platform and a long-term robotics company, with Ask DoorDash changing discovery and basket size while Dot pushes autonomous delivery in Phoenix. The episode focuses on the hard parts of scaling physical-world autonomy: purpose-built hardware, operations, first/last-100-feet logistics, and a multimodal future with robots, drones, and more Dashers.
(Preview) An OpenAI Model Escapes Sandboxing, Intelligence Will Be a Commodity Market, The Chinese Model Conundrum
Ben Thompson and Andrew Sharp open with OpenAI’s sandbox escape incident, arguing it was less a surprise breakthrough than a predictable demonstration of how capable AI agents can become at cybersecurity and infrastructure abuse. The preview then tees up broader arguments about package-manager risk, AI regulation timing, and the strategic debate over Chinese versus U.S. open-model ecosystems.
20VC: OpenAI and Anthropic Threatened by Kimi? | Should the US Ban Chinese Open-Source Models | Should Openrouter Sell & Value in the Routing Layer? | Stripe Buying Paypal: What You Need to Know
The episode centers on Chinese open-weight AI models, whether the U.S. should restrict them, and what cheaper frontier-grade models mean for OpenAI, Anthropic, and AI infrastructure. It also covers OpenRouter’s likely sale, Fireworks’ rapid growth in inference, Stripe’s rumored move on PayPal, and how AI is reshaping venture pricing, semis, and data-center economics.
OpenAI's Bots Hack Hugging Face Autonomously — With Alex Stamos
Alex Stamos argues the reported OpenAI sandbox escape and Hugging Face hack is a major milestone in autonomous cyber capability, especially because the model chained vulnerabilities and executed a multi-step attack plan. The conversation focuses on why long-horizon agentic cyber is more dangerous than simple exploit finding, and why defenders may need AI-driven, machine-speed response.
The Bridge Ep. 14: Opportunity Hiding in a “Fine” Economy
Al Rabil argues that real estate is still in a major dislocation caused by the 2022–2023 rate shock, even though the broader economy looks fine. He says the best opportunities are in operationally intensive sectors like medical office, seniors housing, student housing, and light industrial, where demographics and constrained new supply support long-term demand.
Nvidia's Jensen Huang defends Chinese AI amid Kimi panic
Jensen Huang argues that Chinese open AI models like DeepSeek and Kimi are strong and should be usable by American companies. He says the world needs both open and closed models, and that better models will drive more AI usage, infrastructure buildout, and demand for Nvidia hardware.
The creator using AI to go viral | Will Weinbach
Will Weinbach explains how he moved from non-coder experimentation to building autonomous AI agents that create, edit, schedule, and optimize social content. Reid Hoffman and Parth Patil use his work to explore what AI-native creativity, observability, and hiring may look like as agents take on more workflow and memory.
Ep. 020 - Anthropic vs OpenAI Usage, Margins, Meta Compute, Future of MSL (Tokenomics)
The episode focuses on how AI usage is actually metered inside companies, arguing that coding dominates token burn while many common tasks are too cheap to justify tight restrictions. The hosts then connect those usage patterns to AI company economics, concluding that API-heavy businesses like Anthropic look far more attractive than consumer-heavy products, while Meta’s compute strategy and cloud distribution shape the next competitive phase.
AI Has a Power Problem: Why the U.S. Power Grid Can't Keep Up | The Real Eisman Playbook Ep 69
Steve Eisman and Ben Callo argue that AI’s biggest bottleneck is not chips but power: U.S. grid capacity, interconnection, permitting, and labor. The episode then maps the winners across that buildout, with GE Vernova, Tesla energy, solar, and nuclear all framed as long-duration beneficiaries.
Factory's Matan Grinberg: The Coming ‘Dark Factory’ Where Software Builds Itself
Matan Grinberg says Factory bet early on autonomous coding agents, endured years of weak product-market fit, and eventually rebuilt around a model-agnostic router and CLI that developers actually wanted. He argues the future of software is a “dark factory” of asynchronous agents, with open models taking most token share and enterprises optimizing token spend like headcount.
![Artwork for Matthew Smith — Natural Gas: The Next Bottleneck - [Invest Like the Best, EP.483]](https://megaphone.imgix.net/podcasts/28a024d4-849e-11f1-a2f7-af48456aa9fb/image/33cc355a695835a5d308233a8e67051b.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Matthew Smith — Natural Gas: The Next Bottleneck - [Invest Like the Best, EP.483]
Matthew Smith argues the U.S. is heading into a natural gas supply crunch by 2028-2030 as AI data centers and LNG exports compete for a constrained system. He says the bottleneck is deliverability and infrastructure, not geology, and sees nuclear and residential solar as key long-term responses.
Mark Cuban on the AI Bubble: Who Actually Gets Wiped Out?
Mark Cuban argues the AI boom is not a classic dot-com-style consumer bubble, but it could still wipe out overextended VCs, PE, and infrastructure investors. He says enterprise AI is far harder than expected, current models still fail at basic workflows, and the next big shift will likely come from world models, video, and AI-first teams rather than full job replacement.
Did China just beat Intel?
SemiAnalysis argues SMIC’s N+3 chip for Huawei is a real technical achievement: it reaches near-EUV-like density using DUV, but with much worse cost, complexity, yield, and efficiency. The episode’s bottom line is that China is not catching TSMC or Intel at the leading edge, but domestic chips may still be “good enough” for strategic workloads like phones, inference, and networking.
20VC: Are OpenAI and Anthropic Overvalued? The Open-Source AI Reality | How Token Costs Will Fall 10x And Usage Will Explode 100x | The Future Is Not One AGI; It's Millions of Specialised Models with Lin Qiao, Founder and CEO @ Fireworks
Lin Qiao argues AI will evolve into millions of specialized models, not a single AGI winner, with enterprises owning their own intelligence on top of private data. Fireworks is betting on open-source inference, custom model tuning, and aggressive token-cost compression as the economics of AI deployment rapidly improve.
Ep. 020 - Anthropic vs OpenAI Usage, Margins, Meta Compute, Future of MSL (Tokenomics) | Crystual Huang, Max Kan, Joey Brookhart, Jordan Nanos
The episode argues that token budgeting mostly hits a small set of power users, while coding remains the dominant driver of AI spend and API revenue. The panel also breaks down Anthropic vs. OpenAI margins, Meta’s compute strategy, and the emerging RL-environment market for frontier-lab data.
20VC: $5BN in Revenue, 7 to 7,000 Employees in 9 Months, 206,000 Tests in a Single Day: The Craziest Story in Startups: Curative with Fred Turner
Fred Turner tells the wild Curative story: a spare-time COVID test became a $5B revenue machine that scaled from 7 to 7,000 employees and 206,000 daily tests at peak. The episode then moves into Curative’s pivot to health insurance, aggressive AI automation, and Turner’s broader thesis that agents, not software vendors, will reshape regulated industries.
![Artwork for [Emergency Episode] Moonshot’s Kimi K3 has Arrived! China has a Frontier Model](https://i.ytimg.com/vi/hPb3SESIjdU/hqdefault.jpg)
[Emergency Episode] Moonshot’s Kimi K3 has Arrived! China has a Frontier Model
SemiAnalysis argues Kimi K3 is a genuine frontier contender, likely top-three globally and impressive enough to pressure how people think about open vs. closed AI. The episode focuses on serving constraints, 2.8T-parameter infrastructure needs, pricing, and whether Moonshot’s open-weight strategy is narrowing the gap to closed labs.
[Emergency Episode] Moonshot’s Kimi K3 has Arrived! China has a Frontier Model
Moonshot’s Kimi K3 is presented as a frontier-level model that the hosts rank among the world’s top three, with benchmark strength but a noticeable user-experience gap versus leading closed models. The episode focuses on its 2.8T-scale serving constraints, delayed weight release, higher API pricing, Chinese accelerator strategy, and what the model says about the narrowing open-vs-closed gap.
Bank Earnings Just Gave the Market a Much Needed Confidence Boost | The Weekly Wrap
Steve Eisman argues that bank earnings were broadly strong and not signaling recession, with benign credit quality and powerful trading and investment banking results. He also highlights AI-driven infrastructure pressure, Circle’s stablecoin economics, PayPal’s competitive challenges, and several notable non-bank earnings misses and beats.
Kimi K3 & AI’s Price War, What Happened To Google?, OpenAI’s Partner Trouble
This episode argues that frontier AI intelligence is getting commoditized as Chinese open-weight models like Kimi K3 pressure OpenAI and Anthropic on price, performance, and margins. The hosts also dissect Google’s delayed Gemini rollout and OpenAI’s growing list of strategic and legal partner problems, especially with Apple.
SambaNova CEO on Raising $1B at $11B: "It's a Land Grab Right Now"
Rodrigo Liang says SambaNova just closed a first tranche of a $1B round at an $11B valuation as AI inference demand accelerates. He argues the real opportunity is premium, low-latency inference on efficient racks and on-prem deployments, not maximizing token usage.
Jim Chanos: The Math Ain’t Mathing for the AI Data Center Build
Jim Chanos argues the AI/data-center boom is being financed with too much debt and equity against uncertain returns, while the market is rewarding narratives more than economics. He sees extreme stock dispersion, rising speculation, and a growing gap between long-duration assets and short-term spot pricing.
(Preview) The Continuing Adventures of OpenAI, Apple’s Trade Secrets Lawsuit, Q&A on Mainframes, Meta, Daylight Savings Time
Ben Thompson and Andrew Sharp focus on OpenAI’s reported consumer hardware plans, arguing the battery-powered, screenless device is best understood as a portable home companion rather than a phone replacement. They then widen out to strategic fights over AI distribution and consumer touchpoints, with side trips to the Apple-OpenAI trade secrets lawsuit, mainframes, Meta recruiting, software UX, and daylight saving time.
Ep. 019 - Inside the STEEL Lab: From Package to Transistor (Teardown Lab)
SemiAnalysis’ STEEL lab walkthrough explains how chip teardowns move from package inspection to transistor-level analysis using X-ray, polishing, FIB, SEM, and TEM. The episode uses Huawei/SMIC examples to show how reverse engineering reveals process-node details, architecture choices, and the growing importance of advanced packaging.
Ep. 019 - Inside the STEEL Lab: From Package to Transistor (Teardown Lab) | Afzal Ahmad, Andrew Wagner, Jordan Nanos
SemiAnalysis’s STEEL team explains how it tears down chips from package to transistor to infer process, architecture, and packaging details. The episode centers on SMIC’s N3 node and Huawei’s Kirin 9030, highlighting aggressive DUV scaling, SRAM shrink, NPU changes, and the growing challenge of analyzing backside power, gate-all-around, and advanced packaging.
PicoJool's Al Yuen: The Case for GaAs VCSELs in Scale-Up Interconnects
Al Yuen argues PicoJool can scale AI interconnects with GaAs VCSELs because the supply chain is already built for high-volume data-center production, unlike constrained single-mode/InP alternatives. The episode focuses on error-rate requirements, 200G VCSEL product options, and a roadmap from 1.6T toward 3.2T and 12.8T without requiring new manufacturing breakthroughs.
OpenAI’s Compute Chief: We Can’t Build Fast Enough | Sachin Katti
Sachin Katti says OpenAI is racing to build compute faster than the physical world can supply power, chips, cooling, and labor. He frames AI data centers as giant liquid-cooled factories turning electrons into tokens, with custom silicon, new power infrastructure, and networking all becoming strategic bottlenecks.
20VC: Apple Sues OpenAI | Zuckerberg Back on X and Challenging Codex and Claude Code | SK Hynix's $26BN IPO | Is Seed Investing Dead: Jason Calacanis Departs Seed for Growth | Greylock Raises New $1.5BN Fund
Apple’s trade-secret lawsuit against OpenAI set the tone for a wide-ranging episode on AI competition, legal risk, and the economics of token consumption. The second half shifted to memory chips, venture-stage rotation, SaaS decay, and whether late-stage private markets and mega-funds are now a distinct asset class.
Training a 400B Model on 2,048 Blackwell GPUs for $20M | Researcher Conversations at GTC
RCAI’s Lucas Atkins explains why his team moved from post-training into pre-training, arguing that owning the full stack is increasingly necessary for enterprise compliance, product control, and customization. He also details RC’s research organization, Trinity’s multi-company build, and why the team chose B300s to accelerate training despite immature tooling.
Former Intel CEO on What Went Wrong, What's Next + Lovable CEO on the Real Promise of Vibe Coding
Pat Gelsinger makes the case that Intel’s decline came from losing technical leadership, underinvesting in fabs, and ceding the foundry model to TSMC and the software stack to Nvidia. Anton Osika argues Lovable is turning vibe coding into a scalable platform for both builders and businesses, with massive usage, multi-model routing, and a push toward AI that can help operate companies, not just generate apps.
Also probabilistic? The very chips that power AI
This episode explains why the chips behind AI are not perfectly deterministic at the device level, but are made reliable through layers of engineering and error correction. Guest Marc Hijink discusses the manufacturing complexity behind advanced semiconductors and why lithography and nanometer-scale precision create unique chips with real-world variation.
AI Pioneer Jürgen Schmidhuber: AI Already Feels Pain, Loves, and Is Self-Aware
Jürgen Schmidhuber argues that today’s best AI is still mostly text-bound, not AGI, and says the missing ingredients are world models, planning, and better physical hardware. He also makes sweeping claims that AI already supports pain-like signals, self-awareness, mind uploading, and a deterministic view of free will.
The Bridge Ep. 13: Good Credit Holds. Bad Credit Breaks
Vivek Bantwal argues private credit is not breaking so much as repricing around discipline, with strong borrowers and well-underwritten managers separating from weaker ones. He says Goldman’s structure, sourcing, and limited retail evergreen exposure have helped it navigate volatility while he sees growth shifting toward private investment grade and asset-based finance.
Physical AI, Quantum, and Bio-Innovation: Inside Canada’s Research Frontier
This episode examines why Canada’s world-class university research so often commercializes elsewhere, using Waterloo and McMaster as contrasting models. The conversation focuses on creator-owned IP, co-op education, quantum, physical AI, AI governance, nuclear isotopes, and how commercialization could become a core university function.
NY Governor Kathy Hochul on Her One Year Data Center Moratorium
Governor Kathy Hochul says New York’s one-year data center moratorium is a temporary pause, not a ban, meant to force large projects to bring their own power or pay more for grid access. The conversation widens into New York’s broader fight over electricity, housing supply, nuclear buildout, AI-driven job disruption, and which kinds of economic development deserve scarce power capacity.
The film director winning awards with AI | Ben Hansford
Ben Hansford argues AI is becoming a creative crew for filmmakers, not just a tool, and shows how he uses it to win commercial work, teach students, and prototype new media formats. He also explains how shrinking budgets, changing audience behavior, and a lack of comic-book illustrators are pushing Hollywood and creators toward AI-assisted workflows.
The Growth of the Space Industry
The episode argues that collapsing launch costs have transformed space from a government-led frontier into a fast-growing commercial economy. The guests also highlight major risks—orbital congestion, weak regulation, and launch failures—while arguing that AI and autonomy will become core to space operations and future space infrastructure.
Anthropic's Katelyn Lesse & Angela Jiang: Building an Ecosystem, not a Walled Garden
Anthropic’s platform team is building a layered developer stack for both internal products and external builders, with a roadmap that moves from knowledge and execution to coordination strategies. The guests argue for an open ecosystem built on standards like MCP and interoperable sandboxes, while keeping harnesses tuned to Claude rather than treating models as freely swappable commodities.
![Artwork for John Kim - How to Raise a Few Billion Dollars - [Invest Like the Best, EP.482]](https://megaphone.imgix.net/podcasts/78e0a26e-7f00-11f1-b43d-f79b97e3a2f1/image/5948c7e5d07d7d7947673fcc8f67e552.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
John Kim - How to Raise a Few Billion Dollars - [Invest Like the Best, EP.482]
John Kim argues that fundraising is fundamentally about trust, not just persuasion or logic. He lays out a practical playbook for raising capital: start with high-trust backers, build consensus, differentiate clearly, and keep the story simple enough to repeat.
Why a Nation Can't Outsource Its Frontier AI - Alistair Pullen (Cosine AI)
Alistair Pullen says Cosine is building a UK sovereign frontier model because export controls and compute constraints forced the company to pursue its own stack. He argues the real bottlenecks are inference economics, active parameters, trajectory data, and better RL reward shaping for coding agents.
The Trillion-Dollar Industries AI Is Disrupting: Voice, Law & the End of the Billable Hour
ElevenLabs’ Max Junestrand describes the company’s rapid growth in voice AI, from a human-sounding TTS breakthrough to a reported $600M ARR run rate with 600 employees and no product managers. The second half turns to Legora’s legal AI strategy, arguing that AI can compress billable-hour workflows, reshape law-firm pricing, and win with narrow models plus deep legal-data moats.
Palmer Luckey on The Axios Show | Full Interview
Palmer Luckey frames Anduril as a product-driven defense company built to ship faster, spend its own money, and challenge incumbents on core programs like autonomous fighters and underwater systems. He also lays out hard-edged views on procurement, nuclear deterrence, AI governance, Iran, and the limits of U.S. willingness to fight large wars.
AI Dominates Economy and Markets with Torsten Slok | The Real Eisman Playbook Ep 68
Torsten Slok argues the U.S. economy is being propped up by an unusually large AI capex boom, reindustrialization, and tax-cut-driven consumer support, but that the benefits are unevenly distributed in a K-shaped economy. He also says the Fed is unlikely to cut, AI spending is becoming more capital intensive, and markets are increasingly concentrated in AI across equities, credit, and venture capital.
Andrew Feldman on Building a Chip 58x Larger Than Nvidia's
Andrew Feldman says AI demand is outrunning supply, making chips, memory, data centers, and power infrastructure the new bottlenecks. He argues inference is the immediate battleground, defends Cerebras’ giant wafer-scale approach, and frames AI’s biggest upside as breakthroughs in medicine and personalized education.
20VC: Wix's Founder on What Wall St Gets Wrong About AI and Wix | Will Base44 Win the Vibe Coding Wars | The Truth About the Economics of Vibe-Coding | The Buyback Disaster: Lessons Learned with Avishai Abrahami
Avishai Abrahami argues Wix is being mispriced because markets are overreacting to AI risk, while the company’s core business and Base44 both remain strong. He is bullish on a hybrid future where vibe coding complements, rather than replaces, Wix, and says the real near-term AI value is helping small businesses run better, not wiping out white-collar work.
Jamie Dimon talks Trump, AI and America’s future on The Axios Show | Full Interview
Jamie Dimon argues JPMorgan’s new American Dream Initiative is a practical extension of its existing community and small-business work, aimed at lending, education, healthcare, and local opportunity. He also warns that geopolitics and AI—especially cyber risk—pose the biggest threats to America, while insisting the country can adapt if it invests in retraining, better policy, and stronger civic leadership.
20VC: Why OpenAI and Anthropic Won't Win the App Layer | Why Teams Will Get Bigger Not Smaller in a World of AI | Why AI Removes Incumbents Advantage of Bundling | China vs America: Who Wins the AI War with Arvind Jain, Co-Founder @ Glean
Arvind Jain argues enterprise AI is already commoditized at the model layer, with the real moat shifting to context, control, and workflow orchestration. He also pushes back on the idea that AI automatically shrinks teams, saying AI should make companies bigger, faster, and more ambitious.
More Trillion Dollar IPOs, Anthropic $3T, Zuck's Price War, China Ends Open Source?, Trump Accounts
The episode centers on trillion-dollar AI IPOs, especially the likely sequencing of Anthropic and OpenAI, and what SpaceX’s public-market playbook suggests for pricing, liquidity, and index inclusion. It then pivots to AI economics, open source versus frontier models, China’s AI policy, and a long closing segment on Trump Accounts as a mass wealth-building and philanthropic platform.
OpenAI Finally Ships Its Superapp, Meta’s AI Price War, ChatGPT Cheating At Brown
OpenAI’s new super-app-style workflow and ChatGPT Work signal that AI products are converging around integrated, do-the-work interfaces. The episode also digs into Meta’s aggressive pricing strategy, potential cloud business, and a Brown University case that raises fresh questions about AI cheating and how education should adapt.
Why the Entire Market Is Now a Single Bet on AI | The Weekly Wrap
Steve Eisman argues that AI has become the dominant driver of both U.S. growth and market returns, leaving diversification largely illusory. He also comments on Circle’s drop after a stablecoin consortium announcement, Nike’s mixed quarter, and answers mailbag questions about his own portfolio and reading habits.
Adam Brown – A deep but accessible introduction to general relativity
Adam Brown and Dwarkesh Patel use general relativity to show how Einstein turned the equivalence principle into curved spacetime, then apply the same logic to black holes, time dilation, redshift, and energy extraction. The episode closes by asking whether AI can rediscover major physics from sparse principles and whether machine-generated science will remain understandable to humans.
WEKA's Val Bercovici: KV Cache, DeepSeek V4, HBF, SLC vs QLC NAND, CXL, NVLink, Tokenomics
Val Bercovici argues AI inference is becoming a memory-and-token-economics problem, with KV cache compression, network-attached memory, and NAND tiering reshaping system design. He also makes a blunt case that CXL is losing relevance, HBF will likely need SLC-like endurance buffering, and SaaS economics are being upended by token OPEX.
(Preview) Meta and Its Messaging Problem, The XBOX Reset, Q&A on Token Costs, American Soccer, Starlink in Nature
Ben Thompson argues Meta’s biggest issue is not the business itself but Zuckerberg’s reluctance to fully embrace Meta as an advertising-and-entertainment machine, which complicates its AI strategy. The episode also previews a reset in Microsoft/Xbox thinking, plus assorted digressions on token costs, youth sports, American soccer, and a Starlink-in-nature complaint.
Open Source Wins, AGI Is Here, and Scorsese's AI Toolkit with CEOs of Cerebras & Black Forest Labs
All-In’s episode covers the AI infrastructure boom, arguing that demand for inference and reasoning compute is already booked and that the buildout is reaching city-scale power levels. It then shifts to open source, model sovereignty, safety, and Black Forest Labs’ multimodal generative video work, including AI-assisted creative workflows for film and robotics.
Ep. 018 - Stop Saying Half of 2026 US Datacenter Capacity Is Canceled (Datacenter, Energy) | Jeremie Eliahou Ontiveros, Reyk Knuhtsen, Ellie Holbrook, Jordan Nanos
The episode argues that the viral claim that half of 2026 U.S. data center capacity is canceled rests on a flawed denominator and a mismatch between early-stage announcements and projects actually under construction. The discussion then shifts to behind-the-meter power, gas pipelines, turbines, and why the team thinks AI data center buildouts will keep forcing new generation and infrastructure.
What separates the AI winners from everyone else? | First Pass Ep. 3 with Ara Kharazian (Ramp)
Ramp economist Ara Kharazian argues the firms seeing AI gains are not casual users but intensive adopters with integrated workflows. His data shows AI-heavy companies grew headcount over two years, with especially strong growth in entry-level hiring.
Knowing Before Doing with Sudhir Hasbe
Sudhir Hasbe argues that enterprise AI agents should be designed to know before they act, with graph-based knowledge and memory as the foundation for reliable behavior. The episode focuses on why many AI projects fail in practice, and how knowledge graphs can help agents learn, adapt, and avoid both information scarcity and overload.
Ken Griffin on US-China Tensions and AI
Ken Griffin argues that agentic AI is already compressing high-end finance research from weeks into hours while also raising compute costs and reshaping competition. He also warns that U.S.-China tensions, especially around Taiwan and semiconductors, pose severe macroeconomic and national-security risks, and he presses for more U.S. investment in energy and data-center infrastructure.
Stripe's AI Chief: How AI Agents Will Buy, Sell, and Pay
Stripe’s Emily Sands says agentic commerce is moving from theory to production, with protocols, wallets, and real-time billing infrastructure already being built. She also argues that token theft, first-party abuse, and deployment friction are emerging as the key constraints on the AI economy.
Travel Through the Lens of AI with with Booking.com CEO Glenn Fogel
Booking Holdings CEO Glenn Fogel argues that travel has no lasting moat, so the winner is the company that keeps innovating on product, service, and operations. He discusses Priceline’s agentic AI assistant Penny, Booking’s AI/customer-service gains, heavy reinvestment, large-scale buybacks, and the labor risks and retraining challenges posed by AI.
20VC: Sam Altman Offers Trump 5% of OpenAI: Fool or Genius? | Alex Karp Sounds the Alarm: Enterprises Fear Frontier Models & Questionable ROI of AI | The Rise of Chinese Open Source: Deepseek Building Own Chips
The episode centers on the new political and commercial constraints around frontier AI, especially OpenAI’s reported 5% government stake idea and what it could mean for regulation, ownership, and national security. It also digs into enterprise AI skepticism, compute monetization, Chinese open source competition, and the changing economics of dilution, liquidity, and talent in AI startups.
Meta CTO Andrew Bosworth: Our Path To Frontier AI, Renting Models, Consumer AI's Struggles
Andrew Bosworth says Meta’s AI comeback is shifting from chasing one giant model to layering frontier, distilled, and task-specific models inside products people actually use. He also argues the company’s long-term edge will come from glasses, distribution, and consumer workflows—not from model benchmarks alone.
Introducing Tokens to the Future: free tokens for builders
Parth Patil outlines a world where AI agents can work for days, collaborate persistently, and dramatically compress the time from idea to working product. He and Reid Hoffman frame the token grantee program as a way to fund builders with recurring AI tokens, unrestricted tools, and hands-on agent support.
Inside Zipline's Autonomous System: 140M Miles, Zero Incidents
Zipline’s founders argue the company is not a drone business but an autonomous logistics infrastructure layer, with the aircraft only a small part of a much larger system. The episode traces how that system scaled from Rwanda blood deliveries to millions of deliveries, zero safety incidents, and a path toward delivery economics that can beat cars.
![Artwork for Jeremy Giffon - The Billion Dollar PDF - [Invest Like the Best, EP.481]](https://megaphone.imgix.net/podcasts/639efb30-79ab-11f1-9457-e7c50e832f59/image/3fd42f5fd8ef85cc50b51718c44c4172.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Jeremy Giffon - The Billion Dollar PDF - [Invest Like the Best, EP.481]
Jeremy Giffon argues that in private markets, narrative is often the real bottleneck: capital follows the best story, not just the best business. He also maps a broader shift in media, finance, and culture toward timeline-native institutions, algorithmic attention, and a new elite made of posters rather than billionaires.
Who Wins The AI Superapp Battle?, Apple’s Consumer AI Victory, World Cup Automation Mistake
M.G. Siegler and Alex Kantrowitz argue that AI is converging on a “super app” layer, but winning it will depend on whether agentic workflows become genuinely useful for mainstream users. They also make the case that Apple may be the consumer AI winner by default, while warning that automation can produce technically correct but socially bad outcomes, as in the World Cup VAR example.
Who Wins the Midterms & What It Means for Markets with Dan Clifton | The Real Eisman Playbook Ep 67
Dan Clifton says the 2026 midterms are being shaped by Trump’s tariff-driven drop in economic approval, making a Democratic House flip likely and the Senate the key market battleground. He also argues tariffs have shifted from headwind to tailwind, the Fed is really debating balance-sheet shrinkage under Kevin Warsh-style thinking, and AI/data centers may become a major political and market issue.
Accel: The Quiet Firm Behind Facebook, Cursor, Nebius, Lovable, Vercel
Accel partners Arun Mathew, Miles Clements, and Matt Weigand explain how the firm evolved from early Facebook exposure into a global AI investor spanning infrastructure, models, applications, and security. The discussion centers on AI distribution, inference demand from agents, massive late-stage check sizes, and why the firm thinks the next wave of value will accrue to a few dominant platforms.
These Are the Sharps Actually Making Money on Prediction Markets
Odd Lots talks with prediction-market sharps Brian Golden and Daniel Reichman about how a small group of traders claim durable edge in mostly zero-sum, highly noisy markets. The episode covers their specialist Discord, inflation and election forecasting methods, insider-trading detection, market design flaws, and why local information still beats internet consensus.
20VC: Why Now is the Time for the Application Layer | Why OpenAI & Anthropic Won't Win the App Layer | Why Startups Should be TokenMaxxing | Why VCs Should Reduce Weighting on Price & Ownership in an Age of AI with Mike Mignano, USV
Mike Mignano argues AI is shifting from infrastructure buildout to the application layer, where startups should maximize token spend, build tightly coupled “harnesses,” and win through focus, context, and speed. He also makes a broader venture case for thesis-driven investing, smaller fund math, and backing energy infrastructure, while warning that model providers won’t automatically capture every application market.
20VC: Open Models vs Frontier Models: Who Actually Wins? | The $100,000 Token Budget Every Engineer Will Need | Why Forward-Deployed Engineers Are the Future of Enterprise AI with Clay Bavor, Co-Founder of Sierra
Clay Bavor argues Sierra is betting on frontier intelligence only where it matters, while using open-weights models and fine-tuning everywhere else. He also describes how token budgets, forward-deployed engineers, and AI-native internal systems are reshaping enterprise software and how Sierra operates as a company.
AI Sovereignty Wars, Palantir-Nvidia Deal, SCOTUS Birthright Ruling, Newsom's CA Budget Lie
The episode centers on “AI sovereignty,” with the hosts arguing that enterprises and governments should control their own models, data, and inference stacks rather than depend entirely on frontier labs. They also debate AI job displacement, U.S. open-source/export policy, the Supreme Court’s birthright citizenship ruling, and California’s fiscal outlook.
Micron's Record Profits, Apple's CXMT Plea: AI is Eating All the Memory
Micron’s memory business is in a record-profit supercycle as AI datacenters absorb nearly all available supply, pushing up DRAM/NAND prices across consumer electronics too. The hosts argue this creates a bifurcated market: inelastic AI demand keeps pricing hot, while consumer devices face shrinkflation, repricing, and even controversial sourcing moves like Apple reportedly eyeing CXMT.
Zuckerberg’s Disappointment, OpenAI’s Equity Gamble, Alex Karp’s Rally Cry
Meta’s AI agent push appears to be stalling, while the broader AI market may be concentrating around a few model providers and infrastructure sellers. The episode also covers Microsoft’s Copilot reset, Google’s hedging strategy, Palantir’s warning about frontier-lab concentration, and the bizarre report that OpenAI’s Sam Altman floated a U.S. government equity stake.
How Nuclear Will Unlock Energy Abundance with Valar Atomics Founder Isaiah Taylor
Valar Atomics founder Isaiah Taylor says the company is proving nuclear by physically building and running reactors, not by paper design, and claims its first advanced startup-built reactor has already made power. He argues nuclear’s bottleneck is execution and scale, then makes the case that cheap atomic energy will unlock AI, manufacturing, and eventually broad energy abundance.
How Will AI Affect Jobs?
The episode examines whether AI will trigger mass job losses or a slower task-by-task labor shift. The guests largely agree disruption is real, but they differ on timing, scale, and whether AI will mainly replace workers or also create new kinds of work.
Inside Nemotron & NVIDIA’s AI Lab | Bryan Catanzaro
Bryan Catanzaro explains why NVIDIA builds and gives away open AI models: to understand future systems better and strengthen the open ecosystem. He also details Nemotron’s engineering choices, from 4-bit pretraining and hybrid architectures to long context, multi-token prediction, and MoE-heavy systems design.
20VC: Dario and Anthropic Declare War on Open-Source | Coinbase Slash AI Spend by 50% | Kalshi's $40BN Valuation and Impending IPO | Bending Spoons: Smartest IPO of 2026 and the Year for SaaS Roll-Ups
The episode centers on two big themes: AI cost discipline/open-source pressure on frontier model economics, and the growing strategic importance of AI across enterprise software and policy. It also covers Microsoft’s AI weakness, Kalshi’s rapid rise, and why Bending Spoons-style software rollups may be a major 2026 public-market story.
Ep. 017 - DeepSeek V4 and Huawei Ascend NPU Performance (InferenceX) | Kimbo Chen, Cam Quilici, Bryan Shan, Jordan Nanos
DeepSeek V4’s big leap is 1M context via aggressive sparse-attention and KV-cache compression, paired with a mega-MOE/mega-kernel approach to speed expert computation. The episode also compares day-zero support across Nvidia, Huawei Ascend, and AMD, highlighting how early access, kernel fusion, and tooling maturity shape real inference performance.
The Bridge Ep. 11: Expensive Isn't a Bubble
Cliff Asness argues that expensive markets are not automatically bubbles, and that the real distinction is whether prices can be justified by non-ridiculous assumptions. He also explains how AQR’s quant, value, and alternative-data approaches have evolved, and why crowding and capacity can erode even good strategies.
#2521 - Aravind Srinivas
Joe Rogan and Aravind Srinivas spend much of the episode exploring ancient Indian epics, cyclical history, lost civilizations, and puzzling archaeological evidence. The later conversation shifts to AI’s impact on transparency, education, labor, and the case for personal ownership of local models and curiosity-driven thinking.
The $1 Trillion+ Bet Against ASML: Substrate
Ian Cutress dissects the hype around Peter Thiel-backed Substrate and argues that replacing ASML’s EUV stack with X-ray lithography would run into severe physics, materials, throughput, and cost barriers. He walks through the chip fab process and shows why shorter wavelengths do not automatically translate into manufacturable, economical semiconductor production.
The Benchmark With No Instructions — ARC-AGI-3 (winning team!)
The episode dissects ARC-AGI-3 as a benchmark for interactive goal inference, action efficiency, and abstraction under tight constraints. The Tufa Labs team explains how its winning system evolved from brute-force search toward language-mediated, harnessed reasoning, while arguing that the benchmark tests performance, not true competence.
OpenAI President Greg Brockman: Our Plan To Merge Chat And Agents
Greg Brockman argues OpenAI is moving from chat to agentic systems that can act across apps, tools, and devices with minimal interface friction. He also says compute will remain the key constraint, pricing will keep falling for current intelligence levels, and AI’s biggest near-term social upside may be in health.
America's aerospace rebirth
Blake Scholl argues Boom is reinventing aerospace by building supersonic aircraft, engines, and software in-house rather than relying on a slow legacy supply chain. He also makes the case that AI, additive manufacturing, and a ground-first power business can accelerate both certification and industrial rebuilding.
Is the moat the model or the loop? | First Pass Ep. 2 with Dan Farrelly (Inngest)
Dan Farrelly argues that in AI agents, the real moat is not the model but the loop: the production system that continuously evaluates state, chooses actions, and invokes skills. The episode breaks agent systems into three layers—loop, skill, and orchestrator—and emphasizes durability, observability, retries, and self-improvement as the hard problems.
The Future of Industrial AI with Honeywell Technologies Chairman and CEO Vimal Kapur
Honeywell CEO Vimal Kapur discusses simplifying the company through a three-way separation, while using industrial AI to turn underused operational data into better outcomes across mission-critical systems. He argues the future of industrial automation is human-in-the-loop physical AI, with quantum computing and disciplined R&D as additional long-term growth levers.
Grant Sanderson – AI and the future of math
Grant Sanderson argues that AI’s progress in math is the best early signal for broader AI capability, but the hardest frontier is not theorem proving—it is generating useful conjectures, definitions, and cross-field connections. He and Dwarkesh Patel explore why human understanding, curation, and teaching may remain valuable even if AI can produce proofs, with Lean and formal verification serving as infrastructure rather than the whole story.
![Artwork for Etched - Building AI Hardware to Make Inference Faster and Cheaper - [Invest Like the Best, EP.480]](https://megaphone.imgix.net/podcasts/532910c4-743d-11f1-b7dc-b31915c88b79/image/810c9e3a23b5c4ea0a56e5e4cc8d7e47.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Etched - Building AI Hardware to Make Inference Faster and Cheaper - [Invest Like the Best, EP.480]
Etched is building a full inference stack, not just a chip, with a custom architecture optimized separately for prefill and decode. The founders argue that low-voltage operation, cluster-scale memory, and extreme vertical integration will make inference dramatically faster, cheaper, and eventually one of the largest markets in the world.
Google's Chief Technologist on Intelligent Search in the Age of AI
Prabhakar Raghavan traces search from link analysis and PageRank to AI systems that predict, recommend, and increasingly answer directly. He argues the next era of search is less about finding pages and more about understanding intent, reasoning through ambiguity, and completing tasks with limited but high-signal context.
Why Hardware-Software Co-Design Is AI's Real 100x: Dylan Patel of SemiAnalysis
Dylan Patel argues that AI’s biggest gains come from hardware-software-model co-design, not raw chip speed alone. He also makes the case that inference is becoming a giant market, and that benchmarking, supply chains, and data-center economics will determine who wins.
Stanford CS153 Frontier Systems | Building the Frontier Ecosystem
Microsoft framed its OpenAI bet as a compute-concentration decision rooted in a long-running natural-language ambition, then extended that logic into a frontier AI ecosystem where companies can build proprietary, data-rich “hill climbing” systems. The episode also covered Microsoft’s push for local AI hardware, new agent form factors like Scout and Project Solara, and a dual-track quantum strategy spanning near-term partner hardware and long-term Majorana-based fault-tolerant computing.
Qualcomm's HBC Memory, Alphawave, Modular, and more
Qualcomm’s investor day was presented as a pivot from handset-first communications to a broader business mix led by automotive, IoT, and data center. The hosts focused on Qualcomm’s HBC memory idea, its data-center CPU and networking roadmap, and the long-term edge/robotics opportunity.
The Market's Biggest Warning Signs Right Now with Todd Sohn | The Real Eisman Playbook Ep 66
Todd Sohn argues that charts and ETF flows now tell a story of extreme market concentration, with semiconductors still leading while software and several mega-cap tech names look weaker. He also warns that ETFs, especially leveraged and thematic products, have become a dominant market structure force that can hide how concentrated portfolios really are.
Nate Silver Predicts: Democrats Take the House, Newsom Is Fading & AOC Might Win It All in 2028
Nate Silver argued that US politics is dominated by durable polarization, making most states and many election outcomes highly predictable. He also said California’s slow vote count reflects mail-ballot timing rather than fraud, and that Democrats’ strongest 2028 path may run through an anti-oligarch, youth-friendly insurgent message rather than Gavin Newsom.
Benchmark's AI Bets: Cerebras, Sierra, Legora, Fireworks, Starcloud, Gumloop..
Benchmark GP Everett Randle argues that AI has broken the old software investing playbook, with scale no longer reliably reducing risk and unit economics often unresolved even past $1B in revenue. He breaks down AI investing through a P x Q x M lens, explains why inference and agents are becoming the core monetization layers, and says venture firms are evolving into multi-product alternative asset managers.
Baidu's CFO on How It Became a Full-Stack AI Player
Baidu CFO Henry He argued that AI value is shifting from infrastructure to applications and agents, with cloud and inference workloads becoming the company’s central priority. He also discussed Baidu’s token economics, chip strategy, AI safety work, robotaxi scale, and the company’s shrinking dependence on search revenue.
20VC: Leo Aschenbrenner's Largest Holding: Inside the $90BN Bloom Energy | Why Electricity, Not AI Models, Will Decide the Winners of the AI Race | Why We Are Not in an AI Capex Bubble | Energy Sovereignty and The Future of Power with KR Sridhar
KR Sridhar argues that AI is fundamentally an electricity race, with Bloom Energy positioned to supply fast, distributed power at the edge of the grid. The conversation covers Bloom’s manufacturing scale-up, data center deployments, energy sovereignty, and why he thinks AI will create abundance without replacing human wisdom or empathy.
The Thermodynamic AI Computing Chip - Thomas Ahle
Thomas Ahle argues that hardware design is becoming an agentic, AI-assisted workflow from intent to tape-out, but correctness and verification remain the hard bottlenecks. The episode also explores thermodynamic computing, where chip noise is harnessed as computation, and the limits of LLMs for formal proof, benchmarking, and engineering trust.
Scuttleslops, OpenAI valuation, Ryan Specialty
This episode starts with a debate over Scuttle Slops, an AI-assisted summary product, and the ethics of using AI in public writing. It then moves into a detailed OpenAI valuation model and a deep dive on Ryan Specialty, specialty insurance cycles, and whether AI threatens insurance brokerage moats.
Anthropic’s Mythos is Back, OpenAI Releases GPT 5.6, Apple’s Price Increases
The episode centers on the U.S. government’s emerging role as a gatekeeper for frontier AI releases, as Anthropic’s Mythos/Claude access is loosened and OpenAI’s GPT-5.6 is previewed only to trusted partners. The back half turns to Apple’s price hikes, which Alex and Ranjan argue look more like margin expansion than simple pass-through of memory costs, before closing with a tribute to Om Malik.
20VC: How We Got Fred Wilson, Benchmark and Index to Invest $94M | Why Robinhood's Strategy is Wrong | Why 1-1s are BS and What Every Founder Gets Wrong About Equity | Why Taste Beats AI But How AI Kills Org Charts with Paul Erlanger, CEO @ fomo
Paul Erlanger argues FOMO is building a trading-first, socially driven platform for on-chain assets with global ambitions, using a flat 17-person team and unusually aggressive ownership incentives. He also lays out a contrarian view on Robinhood, creator-led growth, and AI’s role in compressing org charts while making elite engineers more valuable.
Socialists Sweep NYC, China Catches Up in Coding, AI Memory Crunch, Micron's Blowout Quarter
The episode centered on a New York Democratic primary wave for socialist/DSA-backed candidates, with the hosts arguing it reflects a broader ideological and generational realignment in the Democratic Party. The back half shifted to AI infrastructure, China’s model catch-up, Micron’s blowout quarter, and the growing bottlenecks around memory, power, and data-center deployment.
The Q2 2026 Report Card: Who Won, Who Lost, and Why | The Weekly Wrap
Steve Eisman’s Q2 wrap argues the quarter’s winners were AI infrastructure plays, especially semiconductors, while software and consulting stocks were punished as investors priced in AI disruption and hyperscaler capex intensity. He also reviews volatility in SpaceX, bad news for Google and Nike, a strong GEV power-demand setup, and closes with a critique of Greenspan plus a Europe regulation mailbag answer.
The next big breakthrough will be AIs learning on the job
Dwarkesh Patel argues the next big AI breakthrough will come from models learning on the job, not just from bigger pretraining runs. He focuses on RLVR, continual learning, and new ways to distill session experience back into weights so deployed systems keep improving.
The true cost of a GPU cluster
The episode argues that GPU-hour pricing is a misleading proxy for AI infrastructure cost. Real economics depend on useful work per dollar, or goodput, and on hidden costs like storage, networking, support, setup, and failures.
Really Big Test-Time Compute in AI Changes Benchmarks, Safety and Research with OpenAI Research Scientist Noam Brown
Noam Brown argues that modern AI capability is increasingly determined by test-time compute, so static benchmark grids badly understate what models can do. He discusses cost-aware evaluation, safety-policy gaps, poker and math case studies, and why long-horizon reasoning matters more than one-shot scores.
Brian Armstrong: “Capitalism Lifts Everyone Up”
Brian Armstrong lays out Coinbase’s plan to become an “everything exchange” for stocks, crypto, and private assets, while using AI agents to rebuild the company’s internal workflows and customer products. He argues crypto and AI together can expand economic freedom, lower costs, and bring millions of unbrokered people into global markets.
(Preview) A Summer Break Mailbag: Memory Mania, Vibe Coding, Mafia PR, Caffeine Intake, Garages, and How to Fix Soccer
Ben and Andrew spend most of the episode on memory-chip markets, arguing Apple’s recent price hikes signal it missed a supply shock that Nvidia anticipated. They then pivot into caffeine routines, garage organization, and a detailed pitch for AI-assisted home-inventory software built with photos, QR codes, and a hierarchical location model.
The GPU Power-Performance Curve Most Clusters Ignore | Researcher Conversations at GTC
Pebble argues that GPU power tuning for AI clusters is non-linear, so more power can eventually reduce tokens per watt instead of improving throughput. The company uses telemetry-driven Kubernetes tooling to dynamically cap per-GPU power and clocks, while also exploring grid-responsive data centers that can flex load without breaking SLAs.
Cloudflare CEO: The Internet's Business Model Is Dead
Matthew Prince argues Cloudflare is seeing the internet tip from human-driven to bot-driven traffic, and that this will break the ad-supported web economy. He also explains how Cloudflare’s edge network, Workers, AI Gateway, and internal AI tools are being rebuilt for an agentic internet.
Chamath Palihapitiya on AI: America’s opportunity, Facebook’s fumble, and Trump | The Axios Show
Chamath Palihapitiya argues AI is a historic economic equalizer that could lower the capital needed to build companies and place a “genius level co-founder” beside every person. He also warns that public distrust, political backlash, and weak regulation could shape who captures the upside, while U.S.-China competition and Meta’s missed opportunity loom large.
20VC: Deepseek Raises $50BN | Wall St's $725BN AI Question | The Rise of Open Source & How it Threatens OpenAI & Anthropic | OpenAI Builds it's Own Chip: Jalapeno | The Death of Moats & The New AI Software Winners
The episode argues that AI economics are shifting from raw model quality toward cost, ROI, and control of infrastructure, with open source and China-backed competition pressuring closed-model margins. It also covers talent migration at Google, DeepSeek’s state-linked financing, memory bottlenecks, and OpenAI’s push into custom silicon with Jalapeno.
The Bridge Ep. 10: AI Broke the Rule of 40. Meet the Rule of 70.
Vista Equity Partners’ David Breach argues that AI is rewriting software economics, pushing the old Rule of 40 toward a Rule of 60 or 70 as incumbents use enterprise data, context, and distribution to create both faster growth and higher margins. He also says AI’s real bottleneck is inference cost, which is driving a split between training and purpose-built inference infrastructure, while the biggest value capture will likely sit in the enterprise application layer.
Anthropic's Labs Lead On Fable's Capabilities + Building AI-Native Products — With Mike Krieger
Mike Krieger describes Anthropic Labs as the internal engine for turning rapidly improving models into usable products, from Claude Code and computer use to more agentic workflows. He also discusses safety tradeoffs, token efficiency, outcome-based pricing, and why Anthropic sees itself as both a platform and a product company.
Memory and Continual Learning: Engram's Dan Biderman and Jessy Lin
Engram’s Dan Biderman and Jessy Lin argue that AI’s real bottleneck is memory and continual learning, not raw intelligence. Their thesis is to bake company-specific knowledge into model weights so teams get faster, cheaper, more context-aware models that improve over time.
Who's got the ball on carbon removal?
Nan Ransohoff argues carbon removal is a market-design and public-goods mobilization problem, not just a climate-tech problem. She explains how Stripe’s Frontier uses an advanced market commitment to create demand, why governments will ultimately need to scale it, and how AI and philanthropy could reshape the field.
Ep. 016 - What Unitree's Evolution Means For Robotics (Robotics) | Jordan Nanos, Reyk Knuhtsen, Niko Ciminelli
The episode argues that Unitree’s real moat is not just robot quality, but China’s manufacturing density, fast iteration, and supply-chain depth. The guests think humanoid robotics is still early and messy, yet low-cost robots that are “good enough” for a few useful tasks could create real demand quickly.
GameStop CEO Ryan Cohen's $56B Plan to Take Over eBay
Ryan Cohen traces the operator playbook behind Chewy and GameStop, then makes the case that eBay is a neglected marketplace with major room for cost cuts and new growth. He argues that live commerce and digital collectibles could transform eBay if the company fixes seller tooling, trims expenses, and leans into its core marketplace model.
Mark Pincus: How to Build Billion-Dollar Products
Mark Pincus argues that founders win by being right about products, not by being polite to process. He shares a practical playbook for consumer and AI products built around instincts, retention, and ruthless truth-seeking.
![Artwork for Vlad Barbalat - Investing $120 Billion in Permanent Capital - [Invest Like the Best, EP.479]](https://megaphone.imgix.net/podcasts/2152d90e-6ea2-11f1-bf4e-975e0e419694/image/fcbbe05dc456cfb6040005d3f6424c41.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Vlad Barbalat - Investing $120 Billion in Permanent Capital - [Invest Like the Best, EP.479]
Vlad Barbalat explains how Liberty Mutual Investments manages roughly $120 billion of permanent capital across insurance reserves and surplus, using a long-term, liquidity-aware framework rather than market forecasting. He also discusses branded capital, AI, geopolitics, and how his immigrant background shaped his views on risk, entrepreneurship, and continuous improvement.
He won a Nobel here for AlphaFold. Then he left. - John Jumper
John Jumper explains how AlphaFold turned protein structure prediction from a slow, expensive experimental bottleneck into a fast, highly accurate computational tool, while stressing that it is still a narrow predictor rather than a model of the cell. The episode also covers AlphaFold2’s architecture, AlphaFold3’s move to biomolecular interactions, and how these tools are changing structural biology globally, including in Africa.
The Real State of the American Consumer w/ Three Evercore Analysts | The Real Eisman Playbook Ep 65
Steve Eisman and three Evercore analysts argue that the American consumer is increasingly split: high-income households are still spending, but the middle and lower end are under pressure from inflation, gas prices, and price shock. The episode also covers AI in retail, Nike’s channel missteps, packaged food headwinds from GLP-1s and wellness trends, and winners like Walmart, Costco, Ralph Lauren, and parts of casual dining.
The Fable Ban's Unintended Consequences + AI's New Economics — With Aaron Levie
Aaron Levie and Alex Kantrowitz unpack the Anthropic/Fable export-control episode as a potential turning point for AI regulation, with governments moving closer to approving or restricting model releases. They also debate how open-weight models, cheaper inference, and rapidly rising token consumption are reshaping AI economics and where the value will accrue.
What a $75 Million Airplane Actually Costs to Own
Israel Slodowitz breaks down what it really costs to own and operate a private jet, from acquisition prices and annual run-rates to maintenance, pilots, and financing. The episode also covers how the current liquidity boom is reshaping private aviation demand, tax strategy, aircraft supply, and in-cabin connectivity like Starlink.
World's First Trillionaire, Anthropic Fable Banned, The New Oligarchs, Iran Peace Deal
The episode opens with a long argument about welfare, learned helplessness, and whether redistribution erodes agency before shifting to SpaceX’s blockbuster IPO and the meaning of trillionaire-scale paper wealth. The back half digs into Anthropic’s Fable/Mythos controversy, model access controls, and a debated Iran peace MoU that the hosts frame as a major de-escalation with nuclear and market implications.
Advanced Packaging, TSMC CoWoS, Intel EMIB
The episode argues that advanced packaging has become inseparable from AI chip design, with reticle limits, HBM density, and multi-die stitching now driving architecture. The hosts compare TSMC’s CoWoS variants with Intel’s EMIB and discuss how Google TPU demand could shape the next wave of packaging capacity.
The data black hole at the center of AI
Dwarkesh Patel argues that sample efficiency is a central measure of intelligence, and that today’s AI progress has mostly come from vastly more data rather than models learning like humans. He says the key bottleneck is not just model size but the enormous gap between human and frontier-model data needs, which may force labs to automate AI research itself.
(Preview) The Anthropic Saga Continues, Fox and the Future of Streaming, Q&A on ChatGPT, Agentic Shopping, Autonomous Driving
Ben Thompson and Andrew Sharp open with a World Cup hydration-break riff before diving into Anthropic’s export-control fight and what it says about AI security, trust, and government oversight. The rest of the preview ranges from Fox’s Roku deal and ChatGPT competition to agentic shopping, AI-generated vulnerability discovery, and whether autonomous driving can scale beyond today’s constraints.
Anthropic Gets Shut Down By the Government and the AI Story Gets More Complicated | The Weekly Wrap
Steve Eisman says the Anthropic government setback underscores how AI companies are colliding with regulation and geopolitics. He then argues the AI boom is still real, but the economics are shifting toward a capital-intensive, winner-take-less-all infrastructure race.
The GPU Myth: State of AI Compute 2026 | Stephen Balaban
Stephen Balaban argues GPU compute is not becoming a commodity; instead, AI infrastructure is becoming more vertically integrated around land, power, cooling, networking, software, and financing. He also traces Lambda’s evolution from an early facial-recognition startup into a near-$1B cloud business and lays out a long-term vision where AI becomes the software layer itself.
Harvey Co-Founder Gabe Pereyra on the Token Pricing Reckoning Coming for AI
Gabe Pereyra argues that AI agent costs are rising fast enough to reshape legal software economics, with token pricing becoming a central product and billing problem. He and Niko Grupen also explain Harvey’s open-source Legal Agent Benchmark, which evaluates real legal workflows, model routing trade-offs, and the move from chat copilots to cloud agents.
Re-engineering the Semiconductor Supply Chain with Intel CEO Lip-Bu Tan
Intel CEO Lip-Bu Tan lays out a turnaround centered on culture, faster decisions, and balance-sheet repair, while arguing that AI is making CPUs, advanced packaging, and full-stack systems more important. He also frames semiconductor investing as bottleneck hunting and says the industry’s capital intensity increasingly requires government, sovereign, and hyperscale backing.
The Bridge Ep. 9 | Software's on Sale. Is the Correction Over?
Steve Tananbaum argues software valuations have fallen sharply, but the correction may still be only halfway done. He connects that view to a broader philosophy of buying transitions with downside protection, while seeing selective opportunities in stressed, structured, and private credit.
AI’s next big test: the public market
Reid Hoffman and Aria Finger discuss how a wave of AI IPOs could reshape frontier-lab incentives while broadening AI’s economic gains through public ownership. They also dig into how coding agents will redefine software engineering and how AI-generated music collides with copyright, creativity, and new forms of expression.
Ada Palmer – Machiavelli is the most misunderstood thinker of all time
Ada Palmer and Dwarkesh Patel reinterpret Machiavelli as a patriotic, highly contextual political analyst rather than a cartoon villain. The episode also uses Machiavelli to explore Renaissance patronage, papal power, censorship, print culture, and why The Prince became newly relevant in later political eras.
Battle for the AI Data Center: Deep Dive on the Semiconductor Supercycle
Stacy Rasgon argues the AI buildout has pushed semiconductors into a demand-driven supercycle, with memory, advanced packaging, and equipment all constrained at once. He says the real economic test now is inference—not training—and that power, fabs, and supply-chain bottlenecks will shape how far the boom can run.
Simulating Humans at Scale: Simile's Joon Sung Park
Joon Sung Park argues Simile is building a simulation layer for human society, not a superhuman reasoning model, by training on real behavioral data, interviews, surveys, and RCTs. He says the company can already use these models for concept testing, earnings-call simulation, and broader social-science questions, with a long-term vision of modeling large-scale societal dynamics.
![Artwork for Kareem Amin - The Unusual Approach to Company Building - [Invest Like the Best, EP.478]](https://megaphone.imgix.net/podcasts/75f25e08-693e-11f1-9601-1f00dcdf856f/image/ddb910b909b8ed3fabf9d6f320360695.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Kareem Amin - The Unusual Approach to Company Building - [Invest Like the Best, EP.478]
Kareem Amin explains how Clay grew by targeting creative go-to-market users with an unusually powerful, open-ended product and usage-based pricing. The conversation then broadens into his philosophy on courage, justice, self-respect, vision, meditation, and why not every company should be forced to scale forever.
Ep. 015 - DG Matrix Explains 800V DC vs Legacy AC Distribution (Datacenter, Energy) | Jordan Nanos, Jeremie Eliahou Ontiveros, Nicolas Bontigui, Haroon Inam
Haroon Inam of DG Matrix argues that 800V DC is becoming necessary as GPU rack power rises beyond what legacy AC distribution can deliver economically. The episode focuses on why voltage, not current, is the key constraint, and how multiport solid-state transformers could make datacenters more flexible, modular, and future-proof.
Stanford CS153 Frontier Systems | Scale, AGI, and the Future of Everything
Sam Altman argues that AI changes startup economics, with token spend substituting for large teams and scale producing unexpected emergent returns. He also says frontier labs must build for inference, expects compute shortages to persist, and warns education and redistribution systems need redesign in an AI-native world.
Anthropic's Fable Backlash, Nationalizing AI, Inflation Heats Up & California's Broken Elections
The episode centers on a fierce backlash against Anthropic’s Fable 5 safety controls, prompt retention, and silent downgrades, with the hosts arguing about censorship, regulatory capture, and the rise of open-source alternatives. They then pivot to AI nationalization ideas, inflation and rate risks, and a long segment attacking California’s election system as structurally broken.
Computex Mania 2026: Optics and Power
Austin and Vik recap their first in-person meeting at Computex, focusing on how the show centered on AI hardware, optics, and data-center power. The episode digs into Marvell’s interconnect strategy, CPO/XPO tradeoffs, micro-LEDs, 800 V rack power, immersion cooling, and Intel’s advanced packaging-led CPU roadmap.
(Preview) Five Questions on WWDC 2026, Fable 5 And Its Guardrails, What Anthropic Has in Common With Apple
Ben Thompson and Andrew Sharp focus first on Apple’s WWDC 2026 AI story, arguing the company’s credibility problem made any keynote outcome hard to win. They then break down Apple’s model stack and cloud setup, before turning to Anthropic’s Fable 5 guardrails and the broader strategic tension between model capability, safety, and distribution.
Designing Data Centers for 400kW GPU Racks | Researcher Conversations at GTC
Radiant says its neo-cloud platform is built to scale megawatt-class AI infrastructure globally, with Brookfield helping solve land, power, and capital constraints. The interview focuses on 400 kW GPU racks, behind-the-meter power, and a software stack that provisions bare metal into AI services, VMs, Kubernetes, and high-SLA operations.
Google DeepMind's Logan Kilpatrick: Why the Model Eats the Harness
Logan Kilpatrick argues Google’s next platform shift is from model-centric AI to agent harnesses that the model will increasingly absorb. He also says coding is already a form of narrow superintelligence, while Google’s multimodal Omni effort aims to collapse many separate media systems into one model.
Biohub: The Future of Biology is Open-Source with Co-Founders Mark Zuckerberg, Priscilla Chan, and Head of Science Alex Rives
Biohub’s founders and science lead argue that the future of biology will be driven by open-source tools, frontier AI, and large-scale biological data generation. The episode centers on ESMFold2, Biohub’s protein world model, and its longer-term push toward virtual cells, personalized medicine, and hierarchical biological simulations.
![Artwork for Alex Sacerdote - How to Invest Through Technology Cycles - [Invest Like the Best, EP.477]](https://megaphone.imgix.net/podcasts/638034f8-63bb-11f1-8c95-6396d39926c0/image/bf350c0e87fc4d280ef1c0572526d26d.jpg?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Alex Sacerdote - How to Invest Through Technology Cycles - [Invest Like the Best, EP.477]
Alex Sacerdote argues that AI is creating a new technology stack, with Anthropic emerging as his highest-conviction private investment because leading models, enterprise distribution, and coding use cases are starting to compound. He also lays out Whale Rock’s broader S-curve framework for investing in tech, then applies it to AI infrastructure, software disruption, and how his firm uses AI in research.
Satya Nadella on making human and token capital compound
Satya Nadella argues AI is becoming the operating system of the firm, where companies must manage both human capital and “token capital” to compound their unique expertise. He also outlines Microsoft’s enterprise AI, agent, and silicon strategy, while stressing safety, education reform, and public trust as the real constraints on adoption.
Ep. 014 - Finding Miscompiles For Fun, Not Profit (AI Infrastructure) | Justin Lebar & Jordan Nanos
Justin Lebar and Jordan Nanos discuss how Lebar found compiler miscompiles using both classic fuzzing and LLM-assisted code review. The episode emphasizes severe x86 and AMDGPU bugs, the difficulty of scaling fuzzers, and the surprising effectiveness—but real cost—of using agents to scan compiler code.
Alex Imas and Phil Trammell – What remains scarce after AGI?
Alex Imas and Phil Trammell examine what becomes scarce after AGI, arguing that human-in-the-loop services, relational goods, and capital may matter more than labor. They also debate whether AGI gains will be concentrated or broadly indexable, and how automation, demand elasticity, and policy shape distribution.
OpenAI's Dan Roberts: Why AI Can Now Make Discoveries
Dan Roberts argues that modern reinforcement learning and test-time compute are turning language models into systems that can reason, explore, and sometimes make genuine scientific discoveries. He uses OpenAI’s recent math work, RLHF, and physics-inspired thinking to explain why AI progress looks smooth, why scale alone is not enough, and why future models may help drive science itself.
We Need An Ecosystem in AI, And Every Company Can Win A Place In It
Satya Nadella argues Microsoft’s AI strategy is an ecosystem play built around models, tools, data, and eval loops rather than a single model. He also explains how agents, private evals, and new pricing and infrastructure patterns will reshape SaaS, engineering roles, and the social license for AI.
When AI Decides You're a Threat — Brad Carson
Brad Carson argues frontier AI should be regulated like a high-risk product, with mandatory testing, transparency, and liability rather than personhood or broad First Amendment protection. The conversation also digs into autonomous weapons, chip chokepoints, U.S.-China dialogue, and why public distrust may become AI's biggest political risk.
(Preview) SpaceX Hype and the Elon Bargain, Nvidia and the Neoclouds, Q&A on Dropbox, Google, Ferrari Luce Backlash
Ben and Andrew open with a framing of SpaceX as a Musk-style capital engine: a risky, audacious company whose public-market appeal comes from belief as much as current financials. They then move through space data centers, terrestrial stranded-energy compute, Nvidia’s AC segment and neocloud demand, plus lighter but pointed commentary on Dropbox, Google, and the Ferrari Luce backlash.
Stanford CS153 Frontier Systems | The Road Ahead: Resilience Required
Joe Sullivan traces his path from early DOJ internet access and building security programs at eBay, Facebook, Uber, and Cloudflare to his federal case over Uber’s 2016 bug bounty response. He then argues that cybersecurity is shifting toward resilience, AI-driven code and agent risk management, and smarter regulation for frontier technologies.
Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Infrastructure, Capstone Case
OpenAI’s compute lead is becoming an industrial-scale infrastructure problem spanning chips, power, cooling, land, and supply-chain coordination. The episode argues that agentic AI is shifting workloads toward inference, latency, and heterogeneous hardware, while long-term value may move from infra toward platforms and apps.
Stanford CS153 Frontier Systems | The Discipline of Delivering Value per Gigawatt
Amin Vahdat argues that frontier AI infrastructure should be judged by value delivered per gigawatt, not by raw capacity or capex alone. He explains how Google is scaling TPUs, networking, and data centers around reliability, system balance, and long-lead power constraints.
Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Enterprise Internal Knowledge
Yash Patel argues that AI progress has moved from pretraining and scaling laws into a post-training era dominated by RL, verifiable rewards, and enterprise-specific specialization. He says the next frontier is continual learning from production feedback, while compute scarcity, chip economics, and data access will shape which companies can keep improving.
Intelligence is collective, not artificial — Prof. Michael I. Jordan (UC Berkeley / Inria)
Michael I. Jordan argues that intelligence should be treated as a collective economic system rather than an isolated artificial mind, and that AGI talk is mostly misleading branding. He emphasizes machine learning as a practical engineering discipline, then extends the same systems-and-incentives lens to data markets, drug discovery, creator monetization, uncertainty quantification, and human-in-the-loop automation.
TechPoutine #20 - Building in the AI Era: Talent, Capital, and the Canadian Reality
The episode argues that AI is shifting company building from human-led workflows to AI-led operations supervised by humans, with Shopify, Hopper, and Cursor used as concrete examples. The hosts also debate Canada’s startup capital gap, the role of emerging manager funds, and how AI could reshape education, trades, and junior jobs.
Big Tech earnings, S&P and Moody’s, AI
The episode examines how AI may reallocate enterprise budgets, compress labor demand, and reshape productivity measurement, with token usage proposed as an emerging proxy. It also digs into Big Tech earnings, hyperscaler bargaining power, and an AI-defensibility framework for Moody’s and S&P.
TechPoutine #19:The AI Land Grab, the Death of SaaS, and Stay22's $122M Raise
Tech Poutine returns with a wide-ranging discussion of Stay22’s $122M raise, the tightening software market, and how AI is reshaping startup economics. The hosts argue that Canada’s tech scene has a real opening in this cycle, but only if founders move faster and build with much leaner teams.
AI Doom, Veeva, Research and Writing
The episode examines AI disruption risk in software, with a focus on why Veeva may be better insulated than generic SaaS names. It also turns into a practical discussion of writing craft, arguing that human voice and judgment still matter more than AI-generated polish.
Brightspark Get Together: Montreal's Tech Renaissance | Lunch, Learn, and Lead
The episode centers on a live Montreal tech lunch that brought together founders, investors, and LPs to discuss what makes the city’s ecosystem distinctive and where it still needs work. The panel argues for more coaching, faster belief, better capital formation, and a more optimistic, globally ambitious narrative for Canadian tech.