Anthropic

Fireworks CEO: AI Startups Are Scaling Into Bankruptcy (And How to Fix It)

Sep 27, 2026 · 57:46

Lin Qiao, co-founder and CEO of Fireworks AI, argues AI companies are scaling into bankruptcy because product-market fit no longer means a durable business, and that specialized intelligence built on private data cuts costs five to ten times. She explains product is no longer a moat now coding agents make it easy to copy, making accumulated customer data the real moat. Lin shares her framework for owning intelligence: invest 10-20% of AI spend in training to cut total cost of ownership four to eight times. She traces Fireworks' growth from Cursor at $2M ARR to 40 trillion tokens a day, while doubling from 150 to 300 people each quarter with seven co-founders. She also recounts lessons from Jensen Huang, joining Meta in 2015 to learn, and Fireworks' zero data retention policy.

Lambda Co-Founder: How to Pivot and Scale to $3 Million in 1 Year

Aug 26, 2026 · 50:50

Lambda co-founder and CTO Stephen Balaban explains how a $40,000-a-month AWS bill forced a pivot into GPU infrastructure that produced $3 million in revenue in 2017. He details Lambda's six pivots from facial recognition to workstations and says V100 GPUs from 2017 still earn money in its cloud, rebutting short sellers' 'three-year usable life' claim. He stepped down as CEO after 14 years because he disliked the fundraising treadmill and wanted to focus on technology, handing the role to a telecom-industry outsider. He explains why Lambda avoided traditional VCs, argues GPUs are becoming an asset class like toll roads, credits NVIDIA and Jensen Huang for the partnership, and predicts multiple AI models will survive because the market lacks network effects.

Saul Klein: Europe Is Quietly Building the New Silicon Valley

Aug 19, 2026 · 57:23

Saul Klein, Phoenix Court co-founder, argues that Europe—London especially—is quietly becoming the world's next Silicon Valley, calling the UK the third-biggest innovation economy after the US and China. He recounts building LoveFilm from Netflix-like DVD rental into a £100M+ revenue business via partnerships with Tesco, ITV, and Odeon, and joining Skype as it grew 400,000 users a day through network effects. Klein says founders should not blindly go to the US, capital is a commodity, and most of the world's 20,000 VCs add little beyond money; only 28 have backed 11+ companies to $100M revenue at seed. He details Phoenix Court's structure: a company, carry for every employee, and a foundation as 30% shareholder; he names London, Paris, Berlin, Stockholm, and Amsterdam as leading hubs.

a16z Partner: Hollywood Is Secretly Replacing Films With AI

Jul 23, 2026 · 50:27

Justine Moore, a16z partner and early backer of ElevenLabs, explains why AI microdramas — short, vertical soap operas — are becoming the fastest-growing entertainment format, with China's microdrama market now larger than its domestic box office. She argues model quality has reached 90-95% of filming, making it cheaper and faster, and professional storytellers are now using it to create compelling narratives. Amazon and Netflix have announced programs for fully AI-generated animations. ElevenLabs, which she backed at Series A, has crossed $500M in ARR. She highlights consumer agents like Town that automate tasks such as warm intros and expense forwarding, and sees the biggest opportunity in serving less technically sophisticated users and vertical industries like marketing and architecture.

Reid Hoffman, LinkedIn Founder: 99% of People Are Using AI Wrong

Jul 19, 2026 · 1:00:42

Reid Hoffman, LinkedIn co-founder, argues most people underutilize AI by treating it as a search engine rather than an agentic workforce, predicting a shift where humans become managers of AI agent teams. He explains that Blitzscaling today means using AI to rapidly build and discard prototypes, while distribution remains a critical moat for startups. Hoffman warns that negative AI perception in the US and Europe risks ceding leadership to Asia, which embraces the cognitive Industrial Revolution. He highlights the importance of taste and judgment as skills that AI amplifies rather than replaces, and contrasts Anthropic's enterprise focus with OpenAI's consumer strength. Drawing from his decade on Microsoft's board, he shares Satya Nadella's lessons on rebuilding culture, corporate partnerships, and treating AI agents as employees that work for the company. Hoffman also spotlights applications in drug discovery (Manus AI) and personalized AI to maintain human connections.

This Company Is Reinventing Venture Capital | Ben Miller CEO at Fundrise

Jun 28, 2026 · 50:44

Ben Miller, CEO of Fundrise, recounts how the 2008 crisis inspired him to democratize private markets, first through real estate and later venture capital (launched 2022), where they waited 12 months to deploy, investing in Vanta. He contrasts venture—regretting missed opportunities—with real estate—regretting bad deals—and argues that venture value-add is mostly overstated; team builds success. On AI, he says Anthropic is eating the application layer, making it hard for all but top labs, and predicts Mythos will unlock consumer apps; he notes upcoming IPOs of AI labs and SpaceX will reallocate $10 trillion, starving other sectors. He is bullish on longevity through his investment in Loyal, a dog longevity drug company, and sees human longevity within a decade. His admired AI leader is Dario Amodei for genuine concern about risks, and his personal trait is willingness to suffer.

Replit's President & Head of AI | Michele Catasta

Jun 16, 2026 · 49:58

Replit's President Michele Catasta explains how the company evolved from a 15-year-old open source side project to powering 85% of Fortune 500 by launching the first vibe coding agent in 2024, before the term existed. He describes the initial rough launch that attracted candid feedback, the scramble to build an enterprise sales team when IT departments knocked, and the belief that the coding problem is nearly solved—shifting focus to integrations. Catasta argues that technical moats are less important than execution moats built from a decade of battle scars and user trust. He predicts 2026 will be the year of 'agent managers,' where every role involves managing multiple AI agents rather than single-threaded work. The Visa partnership exemplifies Replit's strategy of providing reliable building blocks for monetization, allowing vibe coders to add payments with a single prompt.

Inside Mercury's Founding Story | Co-Founder & CEO Immad Akhund

Jun 9, 2026 · 50:11

Immad Akhund, co-founder and CEO of Mercury, explains how the startup bank captured one in three U.S. startups by launching with product-market fit from day one—a million-dollar deposit arrived within four days from a stranger—and why legacy banks never stood a chance due to their focus on loans, expensive branch structures, and IT rather than engineering teams. Mercury raised $200M at a $5.2B valuation without needing the money, driven by 2.5x growth in Q1 2024 from AI-fueled entrepreneurship. Akhund details building trust as a moat, using small teams and Amazon's working-backwards process to launch AI features like Mercury Insights and Mercury Command, which let users run banking tasks inside AI tools. He dropped all one-on-ones after learning from Jensen Huang's 60 direct reports, favors optimistic curiosity, and angel-invests 250K checks in contrarian areas like space and deep tech.

Inside Gusto's Founding Story | Co-Founder & CTO Eddie Kim

Jun 6, 2026 · 54:31

Eddie Kim, co-founder and CTO of Gusto, reveals how the company grew from a 2012 YC startup called ZenPayroll to $1B in annual revenue and 500,000 customers by cracking the small business market against investor skepticism. He credits the 15-year co-founder relationship with Josh Reeves and Tomer London to shared values and radical directness, arguing that packaging feedback undermines trust. Kim describes building Gusto Co-Founder, an AI agent that automates back-office tasks via chat, with a team of five in eight weeks using Claude Code, embodying how the distinction between engineer and designer disappears. He calls the blank canvas problem AI's main barrier to mainstream adoption and predicts solopreneurs will thrive as AI enables anyone to become a builder.

How Many $10 Billion Startups Will There Be? | Eric Hippeau

May 13, 2026 · 57:21

Eric Hippeau, co-founder of Lerer Hippeau, argues that while AI is transforming tech faster than ever, startups must avoid being mere wrappers around foundational models, as they can be instantly disrupted. He explains that physical AI, combining hardware and AI in areas like robotics and defense, offers more defensible opportunities. Hippeau warns that the startup market is disconnected, with inflated seed valuations assuming many will reach $10B, but historically there's no market for hundreds of such companies. He discusses healthcare, where AI-driven diagnostics could make primary care nearly free, but political will is needed to overcome entrenched systems. On media, he shares lessons from building Huffington Post—keeping stories live and separating opinion—and predicts that AI could restore trust by providing verifiable facts. Hippeau also reveals he has rebuilt his firm's workflow around Anthropic's Claude, cites AR glasses as underrated, and notes that solo GPs must build teams to succeed.

Will AI Models Ever Compensate Creators? Tony Stubblebine (Medium CEO)

May 6, 2026 · 53:35

Tony Stubblebine, CEO of Medium, argues the creator economy is ending as AI slop floods public channels, but a writing renaissance is emerging where authentic expertise and private communities hold value. He details Medium's turnaround from losing $2.5M per month and $37M in debt to profitability, achieved by restructuring investor relationships and refocusing on human-written content. Stubblebine warns that AI companies like OpenAI created a 'prisoner's dilemma' by scraping content without compensation, pushing high-quality writing into walled gardens. He introduces Medium's new AI-assisted writing app TK and private blogging groups as responses to the era. The episode explores how creators must diversify revenue streams and reinvent themselves, while per-inference pricing standards like RSL slowly develop.

We Ignored Silicon Valley Advice for 3 Years | Nicolas Sharp (Attio)

Apr 30, 2026 · 57:49

Nicolas Sharp, co-founder of Attio, explains why the CRM startup spent 1,000 days building in stealth—a deliberate counter to the 'ship fast' philosophy he learned from a previous venture—and how that architectural investment now powers Attio's AI-native platform. He argues that the half-life of product-market fit has collapsed, forcing companies to simultaneously iterate and scale, and that Attio's high-resolution data pipelines are critical for deploying AI agents. Sharp also details how Attio used manual waitlist onboarding to validate demand, how fundraising expectations have risen dramatically (Attio raised $52M at Series B with accelerating growth), and why Europe is having a breakout moment as an AI hub, citing talent from DeepMind and Palantir as catalysts.

Why Andreessen Horowitz Is Still Betting Big On Crypto

Apr 22, 2026 · 50:25

Ali Yahya, general partner at Andreessen Horowitz (a16z), explains why AI and crypto are converging faster than predicted, with privacy emerging as a key defensive moat for crypto startups. He defines AGI as the point when an agent can be fully economically independent and outlines five stages of autonomous commerce, from automated payments to agent-to-agent transactions. Yahya is skeptical about AI agents trading for consumers—since everyone using ChatGPT eliminates alpha—but sees potential for professionals with proprietary data. He notes that a16z's 70-person platform team provides legal, research, and go-to-market support, and that the cost of entry has collapsed, letting founders test ideas with smaller teams before raising capital. He also highlights prediction markets, now legal in the US, as a growing category and recommends robotics as the next frontier, calling physical intelligence a 10–15 year arc offering time for newcomers to enter.

Why Hand-Typing Code Will End Your Career in 2026

Mar 2, 2026 · 49:31

Dan Shipper, co-founder and CEO of Every, argues that AI agents like OpenClaw and vibe coding—prompting apps in English—are fundamentally reshaping how companies operate, from internal workflows to product development. Every, a 20-person company running four AI-native products (Quora, Spiral, Sparkle, Monologue), treats failed software experiments as content that builds distribution, and uses a 'two-slice team' approach where each product is managed by one person. Shipper details how he reclaimed the 'founder' title for legitimacy, the four-year 'flatline' period before the AI wave hit, and why over-hiring early is 'crazy—most of the time it doesn't end well.' He advises founders to automate every internal process with AI and build for both humans and agents.

He Built A $1.8B AI Company With No Experience | Max Junestrand

Dec 14, 2025 · 52:02

Max Junestrand, CEO of legal AI company Legora, explains how AI is rewiring the legal profession, arguing that not building in AI is the biggest career risk. Legora, which went through YC W25 and has raised $260M, doubles ARR every quarter by embedding with law firms to solve real problems. Junestrand details how AI adds a fourth step to the lawyer career ladder—junior associate, senior associate, partner, AI—pushing junior lawyers to focus on intelligence and client skills earlier. He describes Legora's shift from chat to large-scale document review and Word-native drafting, and predicts a move from billable hours to outcome-based pricing as AI automates more work. Junestrand also shares his fundraising approach: build a great business and investors come to you, with Legora's Series B closing in a week and Series C in a week and a half.

The AI Version of You Will Outperform You — Emad Mostaque

Nov 16, 2025 · 51:27

Emad Mostaque, founder of Stability AI and now Intelligent Internet, argues that within 1,000 days, AI will create digital twins that replace most cognitive jobs, but the bigger risk is cognitive colonialism—a handful of corporations controlling the AI layer closest to people. He explains why he left Stability after it reached 300 million downloads to build open, sovereign AI for healthcare, education, and government, citing a medical model that outperforms human doctors on benchmarks and runs on a laptop. Mostaque warns that graduate jobs are already disappearing, and by 2028 coding jobs as we know them will vanish, replaced by English-language AI. He advocates for universal basic AI alongside universal basic income, and announces Sage, an open-source policy engine for governments, and a ChatGPT alternative that routes through any model at cost. The episode details his vision for distributed civic AI funded via Foundation Coin, where compute goes toward helping people rather than maximizing profit.