Inside Cursor: The Anatomy of a Generational Startup

Martin Casado, Sarah Wang, and Matt Bornstein from Andreessen Horowitz discuss their investment in and partnership with Cursor, the AI coding assistant that has become one of the most successful recent tech companies. The conversation begins by setting the context of early 2024, when GPT-4, Claude 3, and Llama 3 were the leading models, and Microsoft's Copilot dominated the coding assistance space despite being a plugin for VS Code.

The speakers emphasize that Cursor's core strategic decision was to build a standalone product rather than a plugin, based on the belief that the interface between human and model was the critical differentiator. Rather than pursuing a coding-specific foundation model like competitors, Cursor's founders recognized that models need to understand natural language broadly, making general frontier models more viable. This philosophy—that 'code in the future will look like pseudocode'—guided all their subsequent decisions.

The a16z team was initially attracted to Cursor because of the founding team's unusual focus and clarity. When asked about enterprise sales, founder Michael rejected conventional advice with conviction, arguing that self-serve growth wouldn't plateau. The investors noted that the team spent much of their pitch meetings saying 'no' to things, demonstrating exceptional discipline about what not to pursue.

The competitive landscape appeared daunting. Microsoft owned GitHub, VS Code, Copilot, and had partnerships with OpenAI, suggesting Cursor's path was nearly impossible. Yet the founders had thoughtful responses to every competitive concern, and the investors observed that even competitors like John Schulman from Anthropic tried Cursor and recognized its potential, despite initial technical limitations with large codebases.

Cursor's growth was extraordinary. The a16z Series A in May-June 2024 led to a Series B by October, with vertical momentum that defied typical growth patterns. However, competitors continually emerged—Windsurf gained traction through YC, Cognition focused on agents, and Claude Code launched with significant resources in May 2025. Throughout this, the Cursor founders remained 'paranoid but unfazed,' with Michael emphasizing that in large markets, formidable competitors are inevitable.

The company made dramatic strategic pivots. Initially focused on the IDE, they evolved toward an agent platform, then a model platform, fundamentally reinventing themselves as AI capabilities advanced. This required significant discipline—cannibalization of successful products when the market demanded it.

Regarding business model evolution, Cursor initially pursued pure self-serve but eventually recognized that enterprise represented superior margins and scale. The company's expansion into enterprise sales was remarkable for its speed and effectiveness. Rather than hiring traditional sales leaders immediately, they waited until product momentum supported it, then built a founding sales team with extraordinary intentionality. Sarah Wang describes how they researched top-performing account executives across leading companies, identifying individuals who drove specific transformations, then recruiting those specific people. This same rigor applied to engineering hiring—spending 40% of leadership time on recruiting was unprecedented.

The company culture reflected the founders' product obsession and attention to detail. Their San Francisco office featured Scandinavian design with no-shoes policies and slippers, creating an environment that felt like home and encouraged constant presence. Marketing decisions were similarly principled—rejecting spam-like tactics to preserve brand integrity. Importantly, the founders made hard personnel and strategic decisions quickly, a trait the investors describe as Elon Musk-like.

Cursor executed numerous acquisitions, initially focused on talent acquisition in a competitive hiring environment, later evolving toward strategic acquisitions like Graphite. The company proved particularly skilled at integrating acquired founders into its culture, treating this as a badge of honor rather than operational complexity.

The speakers identify strong parallels between Cursor and SpaceX: both pursued seemingly impossible competitive battles against entrenched incumbents, both received constant criticism and declarations of failure, both moved with exceptional speed and conviction, both were founded by technically exceptional teams with product focus, and both pursued enormous markets with long-term vision. They note that Cursor's eventual acquisition by Elon Musk represented perhaps the best cultural and strategic fit they've observed in M&A, with both companies sharing product engineering intensity, fast iteration, and commitment to hard decisions.