Key Moments
What Actually Makes A Startup Durable
Y CombinatorWant to know something specific about what's covered?
We've already dissected every moment. Ask and we will deliver (with timestamps).
Key Moments
AI is making software development trivial, pushing startups to focus on truly hard problems like regulation and physics for durability, not just code.
Key Insights
The cost of AI intelligence is decreasing by approximately 10x per year, making the argument that AI is too expensive obsolete within 6-12 months.
A year ago, an engineer might have been more cost-effective than an AI model for certain tasks, but today, the absolute best intelligence models are more effective for nearly all programming tasks.
Building a startup community from scratch is extremely difficult; YC partners advise joining an existing ecosystem in hubs like London or Paris, or even San Francisco, to surround yourself with better founders.
The 'hard bit' of a startup is shifting from pure software development to other areas like B2B sales, regulatory barriers (e.g., banking licenses), physics challenges, or hardware complexity.
The most successful founders launch early and often, iteratively testing hypotheses with the market rather than getting stuck in theoretical research or building.
YC's biggest bet is on people, not ideas; they back strong founding teams who can pivot and adapt, as demonstrated by the Brex founders who applied with a VR glasses idea but achieved massive success.
The rapidly falling cost of AI intelligence
The conversation opens with a discussion on the cost-effectiveness of AI versus human engineers. While AI can now perform tasks previously requiring human engineers, initial concerns about token costs are addressed by the rapid pace of AI development. The intelligence provided by AI is projected to become 10x cheaper per year, meaning that even if AI isn't cost-effective today, it will be within six to twelve months. This trend is so significant that YC partners advise their founders not to worry excessively about current AI costs as the economics will rapidly resolve themselves. Furthermore, a single engineer equipped with AI is now considered 1,000 times more effective than one without it, and for most programming tasks, AI models are now more effective than humans.
The importance of startup ecosystems and communities
For founders looking to build a startup, particularly in regions with underdeveloped startup communities, the advice is to prioritize joining an existing ecosystem. While founding a community is challenging, optimizing for startup success often means relocating to or engaging with established hubs like San Francisco, London, or Paris. These locations offer a critical mass of startups and founders, fostering an environment where ambitious individuals can learn from and push each other forward. The speaker draws on the example of GoCardless in London, which, along with a handful of other early companies, formed the kernel of the city's startup ecosystem, leading to a proliferation of subsequent successful companies. Early-stage startups, especially successful ones, often feel like cults due to their shared, unconventional beliefs and working methods, which lead to extreme effectiveness.
Identifying and focusing on the 'hard bit' for durable startups
In the age of AI, where software development has become significantly easier and cheaper, the definition of a 'hard bit' – the core differentiator for a startup's durability – is evolving. Historically, writing complex software was the hard bit, but this is no longer the case. Startups now need to possess other fundamental challenges to achieve durability. These could include brutally difficult B2B sales into specific industries, navigating stringent regulatory barriers like obtaining a banking license, solving complex physics problems for ventures like spaceship launches, or dealing with the inherent difficulties of hardware development. Founders are urged to shy away from the easier end of the spectrum and pick challenges that are more ambitious and inherently harder. Pure software plays, like Calendarly or DocuSign, are considered less durable because they are too easy to replicate. The key question for founders is to identify what makes their endeavor fundamentally difficult.
The evolving role of judgment and the human element in an AI-driven world
While AI offers immense capabilities, founders must consider where human judgment should not be outsourced. There's a correlation between the amount of thinking an individual does and the quality of their output; therefore, some partners at YC personally write most of their key communications to ensure deep thinking. YC itself is focusing not just on how AI can empower partners and founders but also on what AI cannot replace. These irreplaceable aspects include founder well-being, community building, and knowledge sharing derived from a collective of companies obsessing over AI. The 'power of witnessing'—having someone acknowledge and validate one's experiences—is seen as particularly crucial in the lonely journey of entrepreneurship, something a robot cannot replicate. YC partners also emphasize their deep, tailored understanding of individual businesses, which goes beyond aggregated data and is vital for providing relevant advice.
The empirical cycle of build, test, and learn
A common regret among YC founders is not launching soon enough. Founders often have a natural tendency to research or build more before selling, a bias that needs to be actively fought against. The core advice is to optimize for learning constantly by engaging with the market and users. If users aren't willing to use or don't want what's being built, it's time to go back to research or building. This forms a continuous cycle: build, talk to customers, build, talk to customers. While this cycle has always been fundamental, AI has accelerated the pace at which companies can achieve significant milestones. Founders are cautioned to inspect their biases, as comfort with building can lead to an avoidance of market feedback, which is terrifying because it might reveal foundational mistakes. However, confronting the market is essential for building something people actually want.
Co-founders: Essential for durability and emotional support
While theoretically a billion-dollar company could be built by one person, it's generally not economically logical. Adding a second person provides significant marginal benefits that outweigh the costs, especially in handling coordination problems that arise with larger teams. AI is expected to help compress company sizes, keeping them below Dunbar's number (around 150 people) where relationships can be maintained. However, the core benefits of a co-founder remain crucial. Co-founders provide emotional regulation—lifting each other up during crises of confidence or tempering hyperactive phases. The emphasis is less on complementary skill sets and more on finding someone you trust implicitly, who is smart, determined, has high integrity, and with whom you would be scared to compete. Technical founders are advised to find other deeply technical co-founders rather than business-focused ones, as business skills are generally easier to learn.
Distribution and deep technical wedges as startup differentiators
With the cost of implementing existing solutions via AI drastically reduced, founders are adding value in two primary ways: by advancing the state-of-the-art in a specific technical area to create a unique advantage, or by excelling at distribution and sales. Startups with deep technical expertise, such as building nuclear reactors or advanced space hardware, are explicitly funded by YC. However, the rise of AI is also prompting YC to re-evaluate the necessity of deep technical backgrounds for all founding teams, suggesting that smart, determined systems thinkers might suffice for some companies. Distribution is becoming increasingly critical as a hard-to-replicate advantage, especially in a crowded market. However, for very early-stage companies focused on achieving product-market fit, standing out involves providing exceptional, unscalable white-glove service to initial customers, directly addressing their precise needs.
The enduring importance of human connection and founder quality
Despite the advancements in AI, YC's fundamental investment thesis remains focused on people. The quality of the founding team is the biggest differentiator, often outweighing the initial idea. The Brex founders, for example, applied with a concept for VR sunglasses but were funded due to their brilliant team and prior success. YC partners continuously assess founder teams, recognizing that even with different perspectives, they often align on funding strong individuals. While AI can automate many processes, the human element—founder well-being, community, direct customer interaction, and co-founder relationships—remains paramount. The ability to adapt, learn, and overcome challenges, often facilitated by strong human connections, is what ultimately drives startup success.
Mentioned in This Episode
●Software & Apps
●Companies
●Organizations
●Concepts
●People Referenced
Startup Durability Checklist
Practical takeaways from this episode
Do This
Avoid This
Common Questions
Startups need to focus on the 'hard bit' of their business, which goes beyond simple software replication. This could involve complex B2B sales, navigating regulatory barriers, deep hardware challenges, or foundational scientific research. Durability comes from tackling ambitious problems that are difficult to replicate.
Topics
Mentioned in this video
Y Combinator (YC) is discussed as a platform providing resources, funding, and community for startups. It's highlighted for its speed in building AI tools for founders and its philosophy of investing in people.
The Massachusetts Institute of Technology (MIT) is mentioned in reference to a founder with a PhD in nuclear physics, representing deep technical expertise.
Mentioned as a company that emerged from the London startup ecosystem, demonstrating how a strong startup community can foster new businesses.
Mentioned as a company that emerged from the London startup ecosystem, showcasing the growth and interconnectedness of startups within a hub.
Mentioned as a company that might have a network effect, but is placed in the category of businesses that are too easy to replicate today, contrasting with more durable startups.
Used as a prime example of investing in brilliant founders, regardless of their initial idea, citing their past success and eventual sale. Also used as an example of early marketing strategies in SF.
Mentioned as one of the major AI labs and its models are discussed in the context of routing queries and potential competition with open-source models.
Mentioned as a company whose launches involved engineers that are now working on radio frequency test equipment startups, highlighting deep technical expertise.
Mentioned as the acquirer of Cursor in a hypothetical scenario to highlight the success that can follow persistent launching on platforms like Hacker News.
Mentioned as an early startup success by Elon Musk, illustrating the progression from smaller ventures to larger ones, akin to building rocket ships to Mars.
Mentioned as a subsequent success by Elon Musk, illustrating the growth trajectory from initial ventures to large-scale companies.
Mentioned as a major success by Elon Musk, representing the culmination of building progressively larger and more impactful ventures.
Mentioned for writing a post about the power of witnessing, highlighting the importance of human connection and acknowledgement in the lonely journey of a founder.
Mentioned in relation to his past prediction about one-person billion-dollar companies, sparking discussion on the viability and preference for co-founders in startups.
Cited as an example of a pure software company that is no longer durable in the current market due to ease of replication, suggesting founders avoid this end of the spectrum.
Mentioned as a frontier AI model subject to US export restrictions, highlighting the global impact of such policies and the potential for demand for competitors.
The AI model Claude is mentioned in a question about how to practically implement AI within a startup's structure.
Mentioned as a tool that can be used to make company information, like emails and customer support tickets, queryable by AI agents.
Mentioned as a tool, alongside Obsidian, for making company information queryable by AI agents, aiding in structured data access.
Mentioned as a platform where the founder of Cursor showcased their launch strategy, demonstrating persistence through varying levels of upvotes.
The founder of Cursor is used as an example of persistent launching on Hacker News, illustrating that success can come after many attempts.
The concept of Dunbar's number (approx. 150 people) is used to illustrate the ideal size for maintaining relationships within a company, suggesting AI can help compress companies to this scale.
Referenced in the context of coordination problems in larger teams, illustrating how the number of connections grows quadratically with the number of people, making management difficult.
More from Y Combinator
View all 610 summaries
21 minWhy Physical AI Is the Next Platform Shift
30 minWhat Big Tech Missed And How Startups Can Still Win
31 minHow Two French Engineers In New York Built The Company That Monitors The Entire Cloud
45 minOpencode CEO: Blocked, 20X Growth in 6 Months, Building the Coding Agent for the World
Ask anything from this episode.
Save it, chat with it, and connect it to Claude or ChatGPT. Get cited answers from the actual content — and build your own knowledge base of every podcast and video you care about.
Get Started Free