Lecture, Interview, Fireside Chat, Panel, Conference Presentation
Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Applications, Coding AI
- Company Valuation & Status: Vercel is a developer infrastructure company valued at $9.3 billion, founded by Guillermo Rauch, who also created Next.js.
- Market Scale: The company serves an estimated 20 million developers, a market that is expanding exponentially due to AI-driven coding agents.
- Business Model Shift: Vercel is transitioning from a pure "pages" infrastructure to "agentic infrastructure," optimizing for long-running AI agents and token streaming rather than just instant request-response web pages.
- Pricing Evolution: The industry is moving from seat-based SaaS pricing to token-based pricing models that measure intelligence consumption, potentially rendering traditional "death of SaaS" narratives obsolete in specific contexts.
- Deployment Dominance: Vercel holds a near-monopoly status among coding agents, with data showing Claude selected Vercel for deployment in 86 out of 86 observed instances when using ChatCN for UI components.
- Customer Adoption: Major enterprises like Meta and Notion are utilizing Vercel's infrastructure to accelerate model training and embed agentic capabilities into their core products.
- Support Automation: Vercel deployed its own support agent which now resolves 93% of user inquiries, significantly improving efficiency and customer experience.
- Product Strategy: The company employs a "full-stack" philosophy (inspired by Steve Jobs), building the entire stack from domain names (DNS) and compute sandboxes to UI components, rather than relying on third-party integrations.
- Technical Architecture: Vercel's infrastructure, including AI Gateway and Sandbox, reuses 95% of the underlying "rocket engine" from its original CDN and compute platforms, allowing for massive scale with minimal new development.
- Open Source Bet: The platform's success is anchored in a heavy bet on open source, leveraging frameworks like Next.js, React, and Tailwind CSS which possess "local reasoning" properties ideal for AI consumption.
- Future Vision: Rauch predicts a return to "whimsical" and immersive web experiences using 3D generation and video, contrasting with the current "wall of text" on Wikipedia, while maintaining the necessity of human-centric interfaces.
- Security & Governance: New product categories are emerging specifically for "agent cybersecurity," including sandboxes that protect against data exfiltration and tools for governing agent behavior and guardrails.
- Self-Driving Cloud: The long-term goal is a "self-driving cloud" where agents automatically configure, optimize, and patch software without human intervention, eliminating traditional pager-duty incidents.
- Developer Experience: Vercel aims to be the "front door" for emerging ideas, emphasizing easy domain acquisition and instant deployment to match the velocity of agent-generated code.
- Market Short: Rauch identifies a "short" position on static content aggregators (like Stack Overflow) and low-code drag-and-drop builders that treat code as scarce or difficult to produce.
- Market Long: He is "long" on companies that embrace instant, consumption-based token access and high-velocity infrastructure capable of handling massive, unpredictable demand without rigid rate limits.
- Personal Ambition: If not building for Vercel, Rauch expresses a strong interest in space tech (specifically Mars colonization) and energy breakthroughs like fusion and geothermal power.
- Growth Metrics: Since October 2024, driven by models like Opus 4.5, Vercel saw a 3X growth rate, doubling the number of daily deployments and expanding the total addressable market for software creation.
- Economic Theory: The company operates on a "block economy" theory, providing pre-built, agentic-friendly blocks (infrastructure primitives) that allow AI to compose new software solutions efficiently.
- Infrastructure Evolution: The "sandbox" product mimics the personal computing revolution by providing agents with pre-provisioned computers to extract higher IQ performance, similar to how humans are given laptops.