Paul Graham on YC's Evolution, Founder Ambition, and AI's Impact

Paul Graham (PG) recently participated in a discussion at the original Y Combinator (YC) offices, marking the 47th YC batch and the 21st year of the program. Despite having given this talk numerous times, PG revealed that he prepares it from scratch each time, starting with a blank text file a few hours beforehand. He noted that while the specific content changes, the core principles of starting a startup remain largely consistent, regardless of technological advancements like microprocessors, AI, or the internal combustion engine.

The Evolution of YC Startups: From "Good Old Days" to Serious Ambition

PG addressed the common sentiment that YC has "jumped the shark," a criticism he's heard since around 2008. He refers to these as the "supposedly good old days" because critics often claim YC was once great but has since declined. However, PG argues that the startups YC funds now are far more serious and ambitious than those in earlier batches.

He contrasted early YC companies like Reddit, which, while valuable, isn't on the same scale as current ventures. He cited an example of a recent YC startup working on "intercontinental ballistic cargo," an idea he described as "ambitious" and "serious."

Other examples of serious startups in recent batches include those focused on curing cancer. PG highlighted a startup that provides on-demand research for cancer patients, essentially "inflicting on cancer the death of a thousand cuts." This approach acknowledges the complex nature of cancer, which may not have a single cure but can be combated through multiple strategies. This particular startup also notably assisted Sid Sijbrandij, CEO of GitLab, in his successful battle with cancer, embodying the YC community's collaborative spirit. PG noted that this startup was a realization of an idea he had: to create a service that helps everyone approach cancer with the "founder mode" mentality that Sid adopted.

Frighteningly Ambitious Ideas and the New Google

PG revisited his 2012 essay where he coined the term "frighteningly ambitious ideas." He observed that many of these ideas have since come to fruition. One such idea was the creation of a "new Google." PG had predicted that a new Google wouldn't emerge by directly competing with the existing one but by waiting for technological shifts that render the old model obsolete. He believes OpenAI embodies this, as it has changed how people seek information, moving from web searches to direct AI queries. OpenAI, a side project of Sam Altman, was incubated at YC.

The Driving Force Behind Founders: Ambition and Fear of Failure

PG emphasized the critical role of ambition in founders. He knew this even before YC started, having experienced the grueling nature of building a startup himself. He explained that mere intelligence or talent isn't enough; founders need a powerful drive to overcome immense obstacles.

Interestingly, PG noted that while the prospect of becoming a billionaire is a long-term motivator, the day-to-day drive for founders is often the "fear of failure." This includes the fear of disaster, looking foolish, or critical systems crashing. He likened it to someone trying to prevent their model train set from breaking, focusing intensely on the immediate problem rather than the distant reward. Many founders are surprised when they realize they've become billionaires, often needing someone else to do the math for them.

He also discussed how ambition is largely an inborn quality. While some founders might initially appear less ambitious, it's often because they've been conditioned not to show it, particularly in environments where obedience is valued over independent action.

Startups as Credentials: A Misguided Approach

PG strongly cautioned against viewing startups as mere credentials for a resume. He argued that a failed startup is not a valuable badge, and a successful one becomes a life's work, not just another entry on a list. He compared applying to YC for prestige to applying to Harvard. While a Harvard degree can be obtained through "easy majors," there are no "easy majors" in the startup world. It's akin to being forced to study theoretical physics at Harvard – incredibly demanding and unsuitable for those seeking an easy credential. Starting a startup is "brutally hard" and not an efficient way to "seem cool."

The Meaning of "Formidable"

PG explained that the term "formidable," frequently used within YC, describes a person who "gets what they want." This quality is highly valued by investors because if a formidable founder achieves their goals, the investors, whose interests are aligned, also succeed.

The "Lean Startup" in the Age of AI

Addressing Patrick Collison's recent assertion that "lean startups may be dead," PG admitted he hasn't read "The Lean Startup" book and isn't entirely clear on its definition. However, he challenged the idea that starting with little money in a focused area is obsolete.

He argued that technology consistently becomes cheaper. While AI tokens are currently expensive due to GPU shortages, their prices (especially for inference) are expected to drop dramatically. He believes that even in the age of AI, it's still possible to start a startup with limited funds.

For capital-intensive ventures like rocket startups, PG explained that founders can still start lean by focusing on design, simulations, and expert validation to reach milestones that attract further investment. He cited the example of Philip from StarCloud, who secured funding by writing a white paper and booking a launch, leveraging his existing credibility.

Surprises in AI Development

PG, who studied AI in the 1980s, shared his surprise at the actual trajectory of AI development. He and others had anticipated a gradual progression, starting with simple, perfect AI (like a "perfect fly") and slowly scaling up to human-level intelligence. Instead, what emerged were "full-on human" AIs that were initially "terrible" and prone to "bullshitting," akin to an undergraduate trying to bluff their way through a paper. The development path was inverted: starting with human-like but flawed AI and then working towards perfection, a phenomenon he calls the "jagged frontier."

Regarding the definition of Artificial General Intelligence (AGI), PG suggested that the Turing test remains a strong benchmark. He noted that the concept of AGI, once seen as a clear finish line, now appears as a "smear" with varying degrees of achievement across different AI capabilities.

The Enduring Importance of Shipping Speed

Despite the advent of powerful AI tools, PG asserted that the best predictor of startup success remains the pace at which they ship new products and features. He observed that even with AI, many startups still struggle with shipping speed.

The primary change AI has brought to startups is the emergence of "giant AI bills." Historically, salaries were the dominant cost for startups. Now, the expense of GPUs and tokens can amount to tens of thousands of dollars daily, introducing a new significant cost factor.

The Advantages of a YC Batch

PG outlined several benefits of participating in a YC batch:

• Community and Colleagues: Starting a startup can be a lonely endeavor. YC provides a community of fellow founders, offering a sense of camaraderie and shared experience.
• Learning from Peers: Founders can learn from the technical challenges and solutions of other startups in the batch.
• YC GDP: The "YC GDP" refers to the opportunity for startups to sell their products or services to other companies within the batch. These early adopters are ideal users who make quick decisions and are receptive to pitches.

The Genesis of Y Combinator

YC originated gradually. Initially, the idea was to create an "angel firm" – a firm that would invest small amounts of money early on, using standardized paperwork, bridging the gap between individual angel investors and large VC firms.

To learn how to be investors, PG and his co-founders decided to fund many startups simultaneously. They envisioned it as an alternative to summer jobs for college students, creating a program during the summer. They believed that if they weren't "real investors," it wouldn't matter because the participants were just "college students" and not "real founders." However, both sides quickly became "real," and the "batch" model proved so effective that they decided to continue investing in that manner.

YC's Unchanging Core

Despite its growth, YC has changed very little. PG stated that it's essentially "the same thing but more of it." The problems startups face remain consistent. While the number of startups has increased, they are organized into smaller "pods" of about 70, maintaining a similar dynamic to earlier, smaller batches.

The Future of Trillion-Dollar Companies

PG believes the next trillion-dollar company will emerge from "formidable" founders. He emphasized that it's not about a specific idea (though some ideas are more promising than others) but about the founders themselves. He doesn't anticipate a significant change in the characteristics of founders in the future, noting that 20 years of data show they still look much the same.