Claude

The Untold Story of Higgsfield | Burning $4M a Month on AI Models | CEO, Alex Mashrabov

Sep 28, 2026 · 1:15:43

Alex Mashrabov, CEO of Higgsfield, tells Harry Stebbings his AI video company crossed $1BN in annualized revenue 18 months after hitting $1M, third-fastest behind OpenAI and Anthropic. He recounts burning $10M of a $16M seed before a camera-control pivot brought instant product-market fit, no paid marketing, and one customer growing from $99 a month to $6M a year. He argues Google and OpenAI will destroy $20 prosumer subscriptions, that proprietary models were a mistake — open weights earn 80%+ margins versus 20–30% closed — and that internal model spend tops $4M a month. He also covers engineers moving from Claude to Codex, moats as only outcomes and network effects, selling AI Factory to Snap for $166M, and his bet on $10BN revenue and $100BN+ scale.

Eight Sleep CEO: How AI Runs the Company | “Our Engineers Stopped Coding a Year Ago”

Sep 12, 2026 · 1:21:52

Eight Sleep Co-Founder and CEO Matteo Franceschetti tells Harry Stebbings that AI has transformed how his company operates: engineers stopped coding a year ago, AI agents run an email marketing business making close to $100M with zero employees, and paid media agents manage hundreds of millions in spend. He argues salaries are back to 2021 insanity, so Eight Sleep hires in Europe, and predicts AI spend will grow from 5% to 50% of engineering budgets while unit costs fall. He explains why Meta's reported CAC is 20% lower than reality, why word of mouth drives 40% of revenue with single-digit yearly churn, and why companies will become holding companies like Xiaomi. He also details the 300K mattress footballer story, Charles Leclerc's sponsorship and investment, selling in China through super apps like WeChat and RedNote, running with just 160 employees, and predicting kids born now will never work for money.

Town vs Instinct vs GrokBot | Why the AI Assistant Market Is Not a Bubble

Sep 7, 2026 · 1:16:01

Jean-Denis Grèze, Co-Founder and CEO of Town, joins 20VC's Harry Stebbings to argue the AI assistant market is not a bubble and that Town can survive Google, Apple and OpenAI. He explains Town's pivot from AI tax prep to an assistant living in email and calendar, arguing the winning product will have agent-level network effects. Grèze says startups are subsidizing frontier model costs today, pricing for 20–30% margins in 18 months as prices halve every 9–12 months. He predicts people will trust agents to decide what data to share within five years, calls token maxing the wrong metric, and says Apple could lose because it lacks cloud DNA and is tied to on-device privacy. He also refuses to inflate announced round valuations, calling the practice unethical toward employees.

How to Build a $100M Growth Engine: Lessons from Wispr Flow & Superhuman | Matt Swulinski

Aug 15, 2026 · 1:20:47

Matt Swulinski, Head of Growth at AI coworker Viktor and former growth lead at Wispr Flow and Superhuman, argues the ECOM paid-ads playbook is the right playbook for SaaS and that paid should start immediately. He details analytics setup, why 90% of SaaS companies get it wrong, and scaling creative to 400–500 new creatives monthly. He covers LTV-to-CAC, token-based economics, referral programs tied to usage limits, and paywalls right after the magic moment. He also discusses AEO as the new SEO, why X ads are polluted, and why AI-native 'systems thinkers' are the only hires that matter. He shares Viktor's affiliate program paying 10–15% revenue share as the highest-ROI channel.

Frontier Labs Threatened by Kimi? Should the US Ban Chinese Open-Source Models & Stripe Buys PayPal

Jul 23, 2026 · 1:27:39

Harry Stebbings, Jason Lemkin, and Rory O'Driscoll debate China's near-frontier open-weight models Kimi and Quan, arguing the US should not ban them but compete. They discuss OpenAI's 'AI Communism' tweet, the commoditized inference market leading OpenRouter to consider sale while Fireworks AI hits $1B ARR at $17.5B valuation. The panel notes the application layer has not arrived, with infrastructure spending far exceeding app revenues. They analyze the Stripe-Advent bid to take PayPal private at ~$60B, predicting a 35% premium deal. Finally, they touch on Databricks' $188B Series M, nuclear energy progress, and the strange inversion of private and public markets.