Garry Tan on Spotting Extreme Winners, Advice for Founders in AI + Lessons from Paul Graham

Winning the game (0s)

• There is a sense of anticipation about significant upcoming changes that are perceived as positive for business owners. (0s)
• A meme called "learn to cook, loser" has gained popularity, originating from a discussion about the value of learning to cook as a life skill. It is seen as a metaphor for taking initiative and creating value. (22s)
• Paul Graham is highlighted as a successful figure, having built Y Combinator (YC), and is now perceived to be living an ideal life in Europe, focusing on writing and family. (1m24s)
• Y Combinator is described as a unique and successful concept, providing funding and a network to promising startups, which often leads to significant success for a portion of them. (1m55s)
• YC operates by investing in startups and bringing together top talent, resulting in a high success rate where 5 to 10 percent of companies become highly valuable. (2m39s)
• The program is likened to a college, but instead of charging tuition, YC provides financial support to the participants. (3m9s)

Being the Harvard of startups (3m16s)

• The discussion highlights the importance of becoming an institution, using examples like Harvard, BlackRock, and Goldman Sachs, which are seen as both formidable and welcoming. This is achieved through high standards and clear communication, as exemplified by Paul Graham and Jessica Livingston. (3m17s)
• Paul Graham is noted for his clear communication and logical thinking, which are essential for creating impactful pitches. His approach, along with Jessica Livingston's supportive nature, helped create a unique and welcoming environment for startups. (3m44s)
• Y Combinator (YC) is described as having a simple yet effective process for selecting startups, often investing in people before they even have a concrete idea. Their straightforward approach includes a 10-minute interview and a standard 7% equity deal for all companies. (5m43s)
• Despite the simplicity of its process, YC has a high success rate in picking winners, attracting successful startups at a higher rate than other investors. This success is reflected in the significant market cap generated by the companies they support. (6m14s)
• The text mentions the impressive financial outcomes of YC's investments, such as the Airbnb round, where a small investment turned into a substantial return during the IPO. This demonstrates the effectiveness of YC's strategy and the potential for significant returns with a good portfolio management strategy. (7m1s)

How YC outperforms most Silicon Valley investors (7m31s)

• Over the past 18 years, Y Combinator (YC) has become one of the top 10 Silicon Valley companies in terms of value creation. (7m38s)
• YC has generated approximately $20 million per year per employee in market cap value, with a total value creation close to $100 million per year per employee. (8m7s)
• YC's limited partners (LPs) include entities similar to those that invest in traditional venture capital funds, such as pension funds. (8m31s)
• An analysis of investments in YC companies from 2018 to 2020 showed exceptional returns: the top decile achieved a 16x return, the top quartile an 8x return, the median a 5x return, and the bottom quartile a 3.3x return. (9m8s)
• These returns are significantly higher than typical venture capital returns, where only the top 25% of investments are usually profitable, and the top quartile typically achieves a 3x return. (9m32s)

Spotting extreme winners (9m36s)

• Early-stage extreme winners often resemble average companies, with highly technical, smart, and earnest individuals being a common profile. Brian Armstrong of Coinbase is an example, having started as a solo software engineer without prior entrepreneurial experience. (9m48s)
• The rise of large language models in 2024 has significantly impacted entrepreneurship, enabling software to perform tasks typically done by knowledge workers, potentially without management challenges. (10m51s)
• AI tools are being utilized in various business applications, such as creating prototype websites, writing copy, and data analysis, though finding inspiration for their use can be challenging. HubSpot has surveyed global marketing leaders to identify strategies that differentiate high-growth businesses using AI. (11m27s)
• The ability to communicate effectively with virtual developers is becoming increasingly important, as advancements in AI allow for more intuitive interactions where users can instruct AI to perform tasks without needing to code extensively themselves. (12m38s)
• There is speculation about the potential to use data and spreadsheets to predict the success of startups, though the challenge remains in attracting talented individuals to execute these ideas. (13m39s)

Capital-as-a-service (14m14s)

• The concept of "Capital-as-a-service" involves automating the investment process by connecting a company's financial metrics, such as their Stripe account, bank account, and QuickBooks, to an AI system that can evaluate the company's potential and decide on funding. This approach contrasts with traditional methods, such as the 10-minute interviews conducted by Y Combinator, where partners personally assess and work closely with startups. (14m14s)
• The idea is that an AI system could perform due diligence by analyzing raw data and benchmarking a company against industry standards to determine if it is a worthwhile investment. This method was previously experimented with for growth-stage companies under the name "eightball," which involved connecting data to assess a company's performance. (15m29s)
• An example of AI's potential in business is highlighted by a story where an individual used ChatGPT to create a content website that ranked well on search engines and generated significant ad revenue within a few months. This approach involved the AI providing ideas and guidance, with the human acting as an agent to execute the tasks. (15m55s)

How Garry hustled at 14 to get his into financial security (16m34s)

• At the age of 14, Garry Tan was involved in cold calling companies listed in the Yellow Pages to secure work, which helped his family move from an apartment to a house. (16m35s)
• Growing up in a financially unstable environment, Garry's father, an engineer, struggled with alcoholism and job retention, while his mother worked two jobs as a home health aide and certified nursing assistant. Despite financial challenges, Garry had access to computers, which played a significant role in his development. (16m54s)
• Garry learned web design at a young age, winning awards and starting an underground newspaper with friends, which they published online. This experience taught him that age was not a barrier on the internet, as people could not discern their young age from their writing. (17m53s)
• Motivated by his interest in the internet, Garry used the Yellow Pages to find businesses and offered his web design services, eventually securing a job at a local web design firm. This job involved creating websites for cities like Pleasanton, Fremont, and San Jose, where he earned $7 to $10 an hour and learned skills such as graphic design and HTML. (19m4s)
• The text describes an early experience in business where a young web designer, despite being only 14, presented himself as an award-winning designer. This highlights the challenges and creativity involved in starting a business at a young age. (20m1s)
• A company called Infol Lane is mentioned, which had a small setup with a conference room, an engineer's den, and a reception desk. The CEO of this company was also a regional bank manager at Fremont Bank, illustrating the diverse backgrounds of early tech entrepreneurs. (20m37s)
• The narrative includes a story about a tech guy who walked into a local bank, leading to the bank manager investing his own money and becoming a co-founder of a web design firm in the early 1990s. This reflects the informal and direct nature of business dealings at the time. (21m2s)
• The text touches on personal aspirations, with a desire to achieve the American dream, such as helping family with a down payment on a house, which the family still resides in. This demonstrates the impact of early business success on personal life. (21m43s)
• A prediction made in 2001 about the future of communication is highlighted, where it was suggested that blogs would replace traditional social interactions. This prediction is seen as a precursor to modern social media platforms like Instagram and Facebook. (22m22s)

Turning down Peter Thiel's offer to start Palantir (23m4s)

• Garry Tan was 27 years old when he first got involved with Y Combinator (YC), which is considered relatively late, as he looks young in photos. (23m21s)
• Before joining YC, Tan worked at Microsoft, and Peter Thiel offered him his entire salary to quit his job and co-found a new company, Palantir, which is now a $95 billion market cap company, but Tan turned down the offer. (23m34s)
• Tan attended Stanford University for computer engineering, which changed his life, and he had fraternity brothers Joe Lonsdale and Stephen Cohen, with whom he tried to start several companies. (23m56s)
• In 2004, Tan thought he had a real job with health insurance, but his fraternity brothers, who were a couple of years behind him at Stanford, were going to start a company, and Peter Thiel's approach was to surround himself with the smartest people. (24m16s)
• Peter Thiel's strategy for starting a company is to write down the smartest people he knows on a sheet of paper, take them out to dinner and lunch, and persuade them to quit their jobs and join him. (24m56s)
• The PayPal Mafia, including Peter Thiel, fought a "land war in Asia" against eBay, and the exit for PayPal was not big enough to make people wealthy, but it allowed them to take risks and build businesses on other people's platforms. (25m15s)
• Peter Thiel was likely worth at least $100 million when he offered Tan $80,000 to quit his job and co-found Palantir, which is a significant amount of money, but Thiel had enough wealth to make such an offer. (26m14s)
• A trade involving a gold Rolex for a co-founding engineer of a potentially billion-dollar company is considered fair, highlighting the value placed on key talent in startups. (26m24s)
• The speaker reflects on a past experience at a French restaurant, where a business pitch was made, and notes the importance of understanding the world through direct experience rather than relying on external narratives like media coverage. (26m45s)
• The speaker regrets not joining a company due to a backward way of thinking, realizing later that being proactive and gathering firsthand information is crucial for success. (27m53s)
• The realization that large companies and governments lacked access to software engineers and data expertise was a key insight that led to the success of ventures like PayPal, which used graph databases to detect fraud. (28m40s)
• The speaker discusses how Peter Thiel and others discovered the lack of advanced technology in government agencies, which informed their business strategies and innovations. (29m10s)
• The use of graph-based methods can reveal insights that traditional SQL or row-based methods might miss, as demonstrated by the discovery that three-letter agencies lacked access to certain technologies. (30m2s)
• It is important to learn from individuals with direct experience in unusual or unique fields rather than relying solely on media or social media for information. (30m29s)
• The startup Palantir, associated with Peter Thiel, initially appeared to follow a typical startup trajectory with periods of uncertainty and lack of revenue before eventually becoming successful. (31m1s)
• Founders are advised to be patient and willing to invest significant resources, as exemplified by Peter Thiel's approach with Palantir, which required years of persistence before achieving substantial success. (31m44s)
• Palantir's journey is often seen as an exceptional case, highlighting the challenges and perseverance needed in the early stages of a startup, especially for first-time entrepreneurs aiming to reach their first million dollars in revenue. (32m4s)

Garry's first million (32m20s)

• The journey to financial success can be long and challenging, as illustrated by the experience of working in tech for 18 years before achieving significant financial gains. This included dealing with substantial credit card and student loan debt. (33m37s)
• The first major financial success came from selling Twitter stock at its IPO, despite earlier involvement with Palantir, which took many years to become successful. (33m17s)
• A company called Posterous was founded in 2008, offering a simple blogging platform via email, which grew significantly until the emergence of Instagram. (34m22s)
• Posterous was initially successful due to the novelty of the iPhone and the lack of easy photo-sharing apps, but it eventually lost traction to Instagram, which offered a free app with a network and utility features like filters. (35m40s)
• An investor, Chris Sacca, expressed disappointment in the founders' initial negative reaction to Instagram, highlighting the importance of understanding market trends and competition. (35m11s)
• Founders should be cautious about allowing media or others to define their competition, as this can lead to playing the wrong game. For example, Posterous was framed as a competitor to Tumblr, but its real competition was how people shared photos from their phones. (36m1s)
• It's important for startups to focus on their own path and user needs rather than being influenced by external pressures such as investors or media. Posterous could have been profitable by charging its dedicated users instead of chasing the idea of being a free social network. (36m24s)
• Successful companies like Weebly focused on profitability and user satisfaction rather than continuously raising venture capital. This approach led to significant financial success without relying heavily on external funding. (37m26s)
• Founders should have a clear understanding of their business model and identity, as demonstrated by the experience with Peter Fenton at Benchmark, who questioned whether Posterous was a platform or a network. A lack of clarity can hinder investment opportunities. (38m33s)
• The ultimate goal for startups should be to create something people want, which inherently leads to financial success and enterprise value, rather than just aiming to appear successful by raising funds. (39m3s)
• The discussion highlights the perception of working on projects that are considered less serious, such as blogging platforms, compared to more serious companies like Palantir, which is involved in preventing terrorist attacks. It emphasizes the importance of not accepting external narratives about the significance of one's work. (39m32s)
• The narrative around Facebook is used as an example, where early members, including Mark Zuckerberg, believed in the transformative impact of social media, despite external skepticism. This belief was rooted in the idea that media would undergo significant changes every few decades. (40m41s)
• The importance of being a creator in today's business landscape is discussed, particularly for consumer-focused companies. Building an audience online is crucial because major tech companies like Google and Facebook control access to large audiences, and businesses must pay to reach them. (41m5s)
• For consumer businesses, distribution is critical, and the dominance of big tech companies in controlling user access is a significant challenge. Even with advancements in AI, the distribution advantage of these companies remains strong, requiring businesses to pay for user acquisition. (42m10s)
• The conversation touches on the challenges faced by consumer businesses in breaking out over the past decade, with AI potentially changing the landscape but not yet disrupting the distribution power held by major tech companies. (42m23s)
• In the current digital marketplace, platforms like Google and Facebook operate as perfect marketplaces where every additional dollar of gross margin is extracted through a second price auction system. (42m58s)
• The most genuine form of competitive advantage, or "alpha," is derived from sharing interesting and valuable content, with platforms like YouTube and TikTok offering significant distribution opportunities, especially for new creators. (43m20s)
• TikTok is noted for its generous distribution to new creators, while Meta (Facebook and Instagram) is criticized for focusing primarily on established creators, which may hinder innovation and growth for newcomers. (43m35s)
• In the B2B sector, a significant majority (70-80%) of companies are business-to-business, and the most popular event at Y Combinator (YC) is the sales mini-conference, highlighting the importance of sales skills and mindset shifts for founders. (44m0s)
• Founders are encouraged to embrace rejection, as a high rate of rejection is common in sales. Achieving a 5% conversion rate, whether online or through direct sales, is considered highly successful, as it can lead to substantial enterprise contracts. (44m21s)

Early days at YC (44m57s)

• Garry Tan discusses the daily routine of people working on projects or businesses, emphasizing the search for inspiration and entertainment from sources like Twitter and podcasts. (44m57s)
• He highlights the value of learning from the experiences and traits of admired individuals, such as Elon Musk, Paul Graham, and Sam Altman, noting that these stories, while not perfect, are inspiring. (45m37s)
• Garry Tan shares a humorous story about working with Paul Graham, who gave him his first opportunity as an investor and advisor to startups after experiencing burnout from a previous venture. (46m10s)
• He describes the early days of Y Combinator (YC) in 2011, when it was a small operation with a simple setup, and how people underestimated its potential, likening it to a volunteer camp. (46m40s)
• During this period, Dropbox had already become a billion-dollar company, and Airbnb was on the verge of achieving the same milestone, marking a turning point for YC's recognition as a significant force in Silicon Valley. (47m41s)
• Paul Graham's unconventional approach to business was evident in his behavior, such as wearing shorts and Birkenstocks, and scraping off the sticker from a corporate-looking copy machine, indicating his dislike for formality and corporateness (48m39s).
• There is an insidious nature to formality and corporateness, and having to wear a suit or follow conventions can be detrimental, as it may be associated with "weird soul-sucking things" that no one would choose by default (49m55s).
• Whenever there is prestige and convention, one should be wary of it, and try to fight against formality by being matter-of-fact and unconventional (50m20s).
• Companies can become bad when they prioritize formality over what makes sense, and leaders should reflect on whether they would want to be in a meeting if they weren't in charge (50m39s).
• Starting a company can involve "playing business" by getting a business card before getting a customer, but it's essential to focus on making the product and getting customers before getting caught up in formalities (50m54s).
• The Silicon Valley culture can perpetuate lameness, such as using OKRs and measuring things quarterly, but some of these practices can be useful for large companies like Airbnb, which has 6,000 employees (51m7s).
• It's challenging to distinguish between what is true and what is necessary for a company's growth, and leaders must be aware of the potential traps of formality and convention (51m26s).

The edge of startups with a 2-pizza team (51m31s)

• It is now possible to create highly successful businesses with small teams, often referred to as "two-pizza teams," which can generate hundreds of millions to billions of dollars annually, empowered by large language models and requiring fewer than 20 people. (51m31s)
• Large corporations are unprepared for the rise of agile startups that can quickly adapt and focus on customer needs, potentially leading to significant changes in market dynamics and cost structures. (52m6s)
• There is uncertainty about whether these changes will be inflationary or deflationary, but the expectation is that products and services will become better, cheaper, and faster. (52m32s)
• The current generation of young professionals, who have grown up with the internet and large language models, are well-positioned to drive these changes, while older generations may not be fully aware of the impending shifts. (52m51s)
• Businesses that are agile and adaptable can avoid traditional business process automation and outsourcing, instead leveraging large language models to improve efficiency and reduce costs. (53m31s)
• Startups are advised to target companies that spend heavily on call centers and offer solutions that replace these with large language models, potentially leading to the emergence of many new companies generating significant revenue in the near future. (53m44s)

Advice for founders in AI (54m36s)

• The development of tools to assist in building AI applications is crucial, with a focus on test-driven approaches to ensure reliability and effectiveness. (54m45s)
• Founders face the risk of creating impressive demos to attract investment without implementing proper testing, which can lead to unreliable products. (55m0s)
• It is important for companies to build their understanding of the world based on direct customer interactions and data, which involves observing and analyzing how businesses operate and identifying areas for automation. (55m22s)
• Current large language models (LLMs) have limitations, and founders should break down tasks into smaller steps to improve performance, as LLMs can handle tasks at approximately a 120 IQ level. (56m12s)
• There is concern about the rapid advancement of LLMs from companies like Anthropic, OpenAI, and Meta, which may lead to significant improvements in AI capabilities, potentially automating tasks currently performed by founders. (56m34s)
• The hope is that companies establishing strong brands and competitive advantages now will benefit from reduced costs and increased market capture as AI models improve, allowing them to outperform existing software incumbents. (57m20s)
• Companies with established brands and effective evaluation processes may see their cost structures decrease significantly with the next generation of AI models, enabling them to maintain a competitive edge. (57m32s)
• The strategy involves leveraging data access, smarter systems, and a more effective sales force to create a strong market position, making it costly for competitors to switch away from these systems. (58m0s)
• The discussion highlights the importance of seizing opportunities in the tech industry to build significant companies, such as Microsoft or Salesforce, with the potential to become multi-billion dollar enterprises. (58m32s)
• There is a reflection on Sam Altman's transition from Y Combinator (YC) to focusing on AI, with initial skepticism about his potential to create a generational company. However, Altman's foresight in AI, which was being discussed by influential figures like Elon Musk and Larry Page, eventually proved to be visionary. (58m46s)
• Initially, there was doubt about the potential of AI leading to a doomsday scenario, but over time, there has been a shift towards believing in the capabilities of large language models. (1h0m3s)
• Sam Altman was recognized for his ability to identify and invest in AI early on, bringing together top AI researchers and resources, which was a significant achievement given the competition from major tech companies like Meta, Google, and Microsoft. (1h0m44s)
• The text criticizes large tech organizations for their inability to evolve and make smart decisions, likening them to "big daycares" that fail to utilize the potential of their talented employees effectively. This inefficiency is contrasted with the success of OpenAI, which managed to make significant advancements in AI. (1h1m21s)
• Large organizations like Google face significant bureaucratic challenges that can impede progress, even when they achieve groundbreaking work such as the development of Transformers and large language models. The complexity of navigating internal politics and resource allocation can be more difficult than the technological breakthroughs themselves. (1h2m9s)
• OpenAI, despite its impressive advancements with models like GPT-3, GPT-3.5, and GPT-4, initially focused on being a research lab rather than a product company. It required substantial internal efforts to release consumer-facing products like ChatGPT, which became one of the most successful consumer launches in history. (1h3m36s)
• The current conversation among leading thinkers is heavily focused on AI, with excitement about the potential for small teams of smart individuals to create systems that automate tasks traditionally performed by humans. This includes automating routine knowledge work, which is seen as a poor use of human potential. (1h4m32s)
• There is an expectation that small, software-focused startups will emerge, capable of transforming large payroll expenses into significant software revenue. This shift could free up human resources for more intellectually stimulating and valuable work. (1h5m29s)

The spoon-bending story (1h5m57s)

• A story is shared about attending a spoon-bending party at Burning Man, where participants were encouraged to bend spoons with their minds, but were subtly guided to use their hands to achieve the effect. (1h5m57s)
• The event was likened to a magic trick, with a magician-like figure leading the group and suggesting that warming up the spoon with hands might help in bending it. (1h6m31s)
• By the end of the session, about half of the participants had bent spoons, while the others were amazed and curious about how it was done. (1h7m30s)
• The story serves as an allegory for taking action in real life, emphasizing that while ideas are important, actual effort and work are necessary to achieve results. (1h7m51s)
• The narrative highlights the importance of combining belief in one's ideas with practical action, such as engaging with users and building tangible things. (1h8m2s)