Vercel CEO Guillermo Rauch in Conversation with Supabase: From Pixels to Tokens, Progressive Disclosure of Complexity and the Evolution of AI Cloud Architecture

"Fireside Chat with Guillermo Rauch | Supabase Select 2025" (Supabase Select Summit Fireside Chat, featuring Vercel founder and CEO Guillermo Rauch / @rauchg), here are the key points summarized: 1. Open Source & Monetization • Developer Experience (DX) is consuming infrastructure: From early development of Mongoose, Socket.io to founding Next.js, modern databases and infrastructure cannot just provide bare instances to users; they must offer excellent full-stack, out-of-the-box DX (driving layers, framework integration, and workflow monitoring). • Non-zero-sum game capitalism logic: Open source and commercialization are not opposed. By growing the Next.js pie and making it the default standard for building web applications, Vercel can share in the commercialization; the ample funds obtained can then be reinvested into open source projects to ensure 24/7 response to CVE security vulnerabilities and technological iterations. 2. Product Philosophy: Progressive Disclosure of Complexity • Day 0 / Day 1 ultimate experience: • In the era of AI programming (like v0, Cursor, etc.), developers (and even AI Agents) have extremely high expectations for tools. The setup of databases or computing power must be completed within seconds, with calls requiring only a few lines of code or very few tokens. • Day 0 concept: Become the default base before developers conceive projects (like the deep integration between Supabase and Vercel), avoiding the "Duckling Syndrome" that leads to fixed choices. • Day 1,000 scalability: Many startup products are only good for "weekend prototypes," but are abandoned by users when it comes to handling larger data, enterprise-level compliance, and complex workflows. Excellent products must maintain Day 1 simplicity while gradually releasing underlying advanced capabilities as needed. 3. Vercel's New Chapter: From Pixels to Tokens • Transition from Frontend Cloud to AI Cloud: • Over the past decade, Vercel has focused on creating the fastest front-end pixels and ultimate UI experiences (like Next.js, shadcn/ui, etc.). • With the penetration of AI, many future software will have "Invisible Interfaces." Users only need to provide high-level goals, while the backend consumes a large number of tokens to collaboratively complete complex orchestration. • Core infrastructure layout for AI: • AI Gateway (Token CDN): Similar to how CDNs solved static resource distribution, AI Gateway is responsible for handling intelligent routing, disaster recovery fallback, retries, and caching for multiple model vendors. • Sandbox (Agent's EC2): A secure isolated execution environment and virtual machine for Agents (long-running CLI tools will be launched soon). • Event-driven and durable workflows: For Agent scenarios like Deep Research that last for hours, providing support for interruption recovery and checkpoint continuation. 4. Founder Spirit, Team Management, and Execution • Talent standards IGI (Intelligence, Grindset, Integrity): • As an immigrant from Argentina, he understands that breaking barriers requires strong resilience and a grindset mentality. • Ideal team members possess three elements: Intelligence (high intelligence), Grindset (extreme effort), Integrity (high integrity). • There are only two roles in the company: At Vercel, everyone must either be "Close to the code" or "Close to the customer." • Iteration speed and skepticism (Stay Sus): • Pursue "directional high-speed iteration." • Founders maintain daily personal QA and a "sus" attitude towards products, always assuming that products are not usable in specific extreme scenarios, thus forcing the team to continuously refine edge details. 5. Language and Communication: Core Skills of Technical Leaders • Communication is a hard skill: Influenced by Argentine literature and Borges, he has long focused on the study of language and precise expression. • The return of language value in the prompt era: With the rise of AI and prompt engineering, precise high-quality language communication, abstract expression ability, and solid mathematical foundation are equally important.