Y Combinator CEO says "vibe coding" is rewriting the rules of startup success

Big quote: The recently coined buzzword "vibe coding" is transforming the startup landscape, allowing companies to hit remarkable revenue targets with teams so small they're rewriting the rules of early-stage success. Garry Tan, CEO and president of Y Combinator, recently noted that this method allows companies to achieve significant revenue milestones with surprisingly small teams. For instance, some startups are reaching annual revenues of $1 million to $10 million with fewer than 10 employees, a feat Tan said is unprecedented in early-stage venture capital.

Vibe coding, a concept introduced by OpenAI cofounder Andrej Karpathy, uses large language models to perform the bulk of coding tasks, allowing developers to describe their intentions in natural language and letting AI generate most of the actual code.

Tan said that this method significantly accelerates software development. "You can just talk to the large language models and they will code entire apps. And if it doesn't do – if there's a bug, or if you want it to change, or you want it to look a different way, you don't have to, you know, go in there and write the code yourself." Relying on AI also means that startups no longer need expensive teams to develop software, he added.

There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper...

Currently, about 81 percent of Y Combinator's incubated startups are AI companies, with a notable 25 percent having 95 percent of their code written by LLMs. This shift is not without its challenges, however.

Tan noted that while LLMs excel at generating code, they struggle with debugging, leaving humans to fill this critical gap. "The humans have to do the debugging, still. They have to figure out, well, 'What is the code actually doing?'" he said in an episode of Y Combinator's Lightcone Podcast earlier this month.

Despite these limitations, Tan sees significant benefits in vibe coding. It makes investing in niche software more viable, as the development speed allows smaller markets to support substantial businesses. "This is really the good news," Tan said, noting that industries once considered too small can now sustain businesses with lean teams grossing hundreds of millions annually.

The trend is also particularly beneficial for young engineers struggling to break into a shrinking job market, as it allows them to create standalone businesses without relying on large tech companies, he said.

However, not all AI coding assistants are on board with this trend, as illustrated when Cursor recently showed unexpected resistance. After generating extensive code for a racing game, the AI refused to continue, advising the developer to complete the work personally to ensure he truly understands it. The AI's message underscored industry concerns that while vibe coding offers efficiency, it also poses challenges in fostering deep technical understanding.

Despite these challenges, the benefits of vibe coding are undeniable, allowing startups to stay lean and achieve rapid growth in previously unexplored markets. As Tan put it, "Maybe it's that engineer who couldn't get a job at Meta or Google, who actually can build a standalone business making 10 or 100 million dollars a year with 10 people. Like that's such a powerful moment in software."