Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Applications, Coding AI

Key Concepts

• Agentic Infrastructure: The next generation of cloud infrastructure designed specifically to support AI agents rather than just human-written code.
• Vibe Coding: A colloquial term for using AI coding agents (like v0 or Claude Code) to rapidly generate, iterate, and deploy software without traditional manual coding.
• Local Reasoning: A design principle in code (exemplified by Tailwind CSS and React) that allows components to be self-contained and easily understood by AI models, facilitating "block-based" development.
• Token-Based Pricing: A shift from seat-based SaaS pricing to a model that measures and charges for "intelligence" (tokens) consumed.
• AI Gateway: A "CDN for tokens" that provides observability, failover, security, and load balancing for AI model interactions.
• Sandbox: The fundamental compute unit for agents, providing isolated environments (Docker containers) where models can execute code and "cut their teeth."

---

1. Main Topics and Key Points

• The Evolution of Vercel: Originally a platform for deploying front-end web applications, Vercel has evolved into a critical infrastructure layer for AI agents. It is currently valued at $9.3 billion.
• The "Peanut Butter and Jelly" Metaphor: Coding agents (the peanut butter) require a deployment platform (the jelly) to be useful. Vercel positions itself as the essential infrastructure for these agents.
• The Shift from Pages to Agents: The internet is moving from static, human-navigated web pages to dynamic, agent-driven experiences. This requires new infrastructure capable of handling long-running "thinking streams" that can last from seconds to entire days.
• The "Self-Driving" Cloud: Rauch envisions a future where cloud infrastructure is self-optimizing and self-healing, removing the need for human intervention (e.g., "pager duty" for engineers).

2. Real-World Applications

• Enterprise Adoption: Meta uses Vercel to power its Superintelligence Labs, allowing them to "vibe code" tools and train models faster.
• Business Transformation: Notion is integrating agentic capabilities using Vercel infrastructure.
• Operational Efficiency: Vercel’s own support agent now resolves 93% of user inquiries, demonstrating the shift from knowledge-based hyperlinks to interactive agents.
• Sales Engineering: Startups are using v0 to build custom, living software demos for prospects in real-time, replacing static slide decks.

3. Methodologies and Frameworks

• The Block Economy: Inspired by Mitchell Hashimoto, this framework treats software as modular "blocks" that agents can easily assemble.
• Dogfooding: Vercel builds its own tools (like the AI Gateway) by using its own platform, ensuring the infrastructure is battle-tested.
• Agentic Ergonomics: Designing APIs and code structures that are optimized for AI consumption rather than just human readability.

4. Key Arguments

• Deployment is the True Metric: Rauch argues that writing code is not what makes a developer special; deploying it and putting it in front of a customer is. Coding agents excel at this because they lack the "works on my machine" bias.
• The Death of "Static" SaaS: Traditional, rigid SaaS products are being challenged by highly tailored, AI-generated software that can be built in hours rather than months.
• The Return of the Whimsical Web: AI will enable a return to immersive, interactive, and "alive" web experiences (similar to the old Encarta encyclopedia) rather than the current "wall of text" internet.

5. Notable Quotes

• "Writing code doesn't make you special. Deploying code and putting it in front of a customer does make you special." — Guillermo Rauch
• "The world is made up of piles and piles and piles of software that doesn't run." — Guillermo Rauch
• "If Amazon started AWS today, I don't know if they would call it Amazon Web Services. They would call it Amazon Agent Services." — Guillermo Rauch

6. Data and Research Findings

• Market Dominance: Reports indicate that Vercel’s UI engine, shadcn, has a 90.1% market share, and Vercel’s deployment platform is used by nearly 100% of agents in specific testing scenarios.
• Performance Impact: Amazon data shows that for every 100 milliseconds of latency, conversion rates drop by 1%.
• Growth: Vercel has doubled its daily deployments since January, driven by the surge in agentic activity.

7. Synthesis and Conclusion

The transition to an "agentic internet" is the most significant shift in software history. Vercel is positioning itself as the "AWS of AI" by providing the necessary primitives—sandboxes, AI gateways, and deployment infrastructure—to support this new era. While traditional, rigid SaaS models face disruption, the future belongs to companies that embrace "speed of tokens," composability, and agent-friendly interfaces. The ultimate goal is a "self-driving" cloud where infrastructure manages itself, allowing humans to focus on higher-level innovation.