Paul GrahamCo-Founder

Whether you are building around microprocessors, the internal combustion engine, or modern artificial intelligence, the fundamental mechanics of launching a startup remain identical. You must understand customer problems, build something people want, and maintain relentless execution. While the underlying technological tooling evolves, human behavior and the core discipline of startup creation stay the same.

People have claimed YC peaked since 2008, but today's founders tackle problems that are vastly more ambitious and consequential than earlier consumer software. In early batches, projects often revolved around social forums, whereas modern cohorts work on intercontinental ballistic cargo transport, specialized space systems, and customized biotech interventions.

Startups are fighting complex diseases through targeted therapies and on-demand oncology research. By approaching severe medical challenges with the agile mindset of a technology startup, founders inflict a death of a thousand cuts on complex diseases, creating bespoke research platforms that benefit broader patient populations.

Disrupting entrenched tech monopolies like search engines cannot be achieved by attacking them head-on with identical products. You must wait for a paradigm shift that renders the legacy model obsolete. Conversational AI interfaces demonstrated that users sought direct synthesis of information rather than indexed lists of web links, transforming search behavior completely.

While financial upside exists in the background, day-to-day founder motivation is almost entirely driven by an acute fear of failure, operational disaster, and public embarrassment. Founders focus intensely on keeping their systems running and preventing catastrophic breakdowns, often realizing they have built substantial equity value only years later.

Starting a company is an extraordinarily inefficient and painful way to seek prestige. In university systems, you can select easier paths to acquire credentials, but entrepreneurship forces every founder into grueling, high-difficulty execution where failure confers zero social status.

A formidable founder is someone who relentlessly gets what they want regardless of obstacles. When investors back individuals who possess this stubborn determination, investor interests align directly with the founder's ability to navigate severe adversity and achieve their ultimate objectives.

You can still launch significant ventures with minimal capital. While compute infrastructure and token consumption can be costly initially, technological processing power consistently becomes cheaper and higher quality over time, allowing resourceful teams to achieve substantial traction before raising large capital rounds.

Founders in aerospace or hardware do not need to build complete physical vehicles immediately. By developing rigorous engineering simulations, publishing credible technical white papers, and booking future launch slots, founders establish sufficient credibility to unlock sequential tranches of investor funding.

Early AI researchers expected systems to evolve from simple biological organisms up to human capability while maintaining precision. Instead, large language models emerged immediately with broad human conversational fluency alongside factual hallucinations, requiring researchers to work backward from plausible language toward logical precision.

Instead of representing a sharp finish line, AGI is a broad, multidimensional spectrum. Certain machine capabilities already solve open mathematical problems at superhuman levels, while others struggle with basic real-world scheduling, demonstrating that intelligence expands unevenly across different domains.

Shipping velocity remains the single most reliable predictor of startup success. Even with access to automated programming assistants, significant variation persists in how quickly founders conceive original solutions, test them with users, and iterate based on feedback.

Starting a company is inherently isolating. A batch environment provides founders with an immediate community of peers facing similar technical hurdles, facilitates direct knowledge sharing, and creates an internal marketplace of early adopting customers who provide rapid feedback.

We originally set out to build an angel investment firm that invested small amounts of capital into early-stage founders using standardized legal paperwork. We organized the first batch over the summer as an alternative to corporate internships for college students, discovering that cohort-based investing accelerated both investor learning and founder progress.

While total cohort sizes have grown, YC maintains high-touch mentorship by partitioning each batch into smaller dedicated pods. Each pod operates like an independent small batch, preserving close peer relationships and direct partner interaction.

Generational companies will not emerge from generic trend categories but from exceptionally formidable founders. Founders with deep technical mastery, relentless drive, and high resilience will discover and build the defining platforms of the coming decades.