What Actually Makes A Startup Durable

AI Economics and Efficiency

• YC advises founders to join the program to receive $1 million in free tokens and access to additional offers.
• The cost of intelligence is decreasing at a rate of roughly 10x per year for equivalent performance.
• YC asserts there is no current programming task where a human is more cost-effective than an AI model using the best available intelligence.
• Current AI capabilities allow an engineer with a model to be approximately 1,000x more productive than one without.

Founder Community and Geography

• Founders outside major hubs should join existing startup communities rather than attempting to build one from scratch.
• San Francisco remains the primary global hub, but London and Paris are viable alternatives for finding a critical mass of peers.
• Startup ecosystems grow through a "family tree" effect, where a few successful companies like GoCardless spawn dozens of others (e.g., TransferWise, Monzo).
• Early-stage founders benefit from the "cult-like" intensity of shared beliefs and working behaviors found in successful startups.

Human Judgment vs. AI Automation

• YC partners emphasize retaining human judgment on founder well-being and community building, areas where AI cannot replicate value.
• The concept of "witnessing"—having a human acknowledge the loneliness and normalcy of founder struggles—is a critical, non-automatable service.
• YC partners provide tailored advice based on deep, specific knowledge of a company, which a static database of 7,000 companies cannot replicate.
• A new experiment allows founders to book office hours with a virtual partner, though human interaction remains superior for emotional support.

B2B Viability and "The Hard Bit"

• Pure software startups are considered less durable as AI lowers the barrier to entry for replication.
• Startups must identify a "hard bit" to achieve durability, such as complex B2B sales, regulatory barriers, hardware physics, or deep industry knowledge.
• YC explicitly advises against building low-complexity software products (e.g., simple scheduling or document tools) that can be easily replicated.
• Founders are encouraged to view difficulty on a spectrum and intentionally select harder, more ambitious problems to build defensible moats.

Solo Founders and Team Dynamics

• While a one-person billion-dollar company is theoretically possible, YC data suggests solo founders statistically perform worse due to a lack of peer support.
• The primary utility of a co-founder is emotional modulation: providing support during lows and grounding during manias.
• AI is expected to compress effective company sizes from 2,000 down to under Dunbar's number (approx. 150 people) by handling coordination.
• YC advises finding a co-founder based on integrity, determination, and shared values rather than strictly complementary skills (e.g., technical vs. business).
• Complementary skills are less critical today because AI allows individuals to bridge skill gaps more easily than in the past.

Capital, Funding, and VC Strategy

• Founders building "hard" startups (nuclear energy, regulated finance, space) will still require massive capital, sustaining the role of venture capital.
• AI enables faster growth with less capital for simple software, but ambitious AI-native companies attacking harder problems will drive increased capital demand.
• YC funds founders because they believe in the team's ability to figure out the right path, noting that roughly one-third of batch companies pivot significantly.
• The "hard thing" has shifted from software implementation to other barriers, meaning funding needs now correlate with the ambition of the hard problem.

Student Access and Credits

• Students attending the event will receive over $25,000 in AI model credits and GPU access through YC's sponsorship agreements.
• This credit allocation is intended to neutralize the cost disadvantage between European students and US-based founders with API budgets.

Application Strategy and Timing

• Early Decision applicants must demonstrate a legitimate reason for not waiting; YC asks if they would fund the applicant immediately or wait a year.
• A lack of urgency in an Early Decision application (e.g., preferring to finish school) can signal insufficient commitment to YC standards.

YC Program Habits and Metrics

• Top-performing startups launch early and often, prioritizing real-world feedback loops over theoretical research.
• Successful founders focus on a single bottleneck every two weeks, overwhelming it before moving to the next constraint.
• YC emphasizes that "building something people want" requires constant confrontation with the market, not just internal development.
• The feedback loop for AI usage requires harnessing AI to memorize interactions and continuously update its own skills to improve output quality.

AI Implementation Guidelines

• To build an AI-native company, founders should pick one narrow process (e.g., post-call follow-ups or prototype generation during sales calls) rather than attempting total automation.
• Founders should make all internal data (emails, drives, tickets) queryable to their AI agents.
• Direct customer interaction is the last process to be automated; it is essential for maintaining context and closing the loop on product development.
• Talking to co-founders directly is another non-automatable function critical to resolving internal team friction.

Pivoting in the AI Era

• Pivoting is only justified when a founder has exhausted options and received real evidence that the core hypothesis is wrong.
• Pivoting due to sadness or sales friction is often a symptom of poor execution or lack of product-market fit rather than a strategic error.
• The best pivots occur when customers explicitly request a different direction or when working on a hard problem exposes a superior, adjacent opportunity.

High Agency Development

• High agency is defined as the belief that one's actions will produce tangible output.
• This trait can be built by repeatedly engaging in difficult but tractable projects, eventually scaling the complexity of these endeavors.
• YC suggests finding a co-founder to tackle side projects that are challenging enough to build resilience and momentum.

Distribution and Differentiation

• In an era where building software is easy, distribution becomes the primary differentiator and the "hard bit" of the business.
• Early-stage startups must compete through unscalable, white-glove service and deep personal engagement rather than broad advertising.
• Founders must identify their specific metagame; what works for a scaled company (billboards) is irrelevant for a startup seeking product-market fit.