Levels of greatness

The gap between the top handful and the level below is so stark.

Welcome to issue #65 of next big thing.

I’m constantly reminding myself that there are many levels to greatness.

Take the sport of tennis. You can be terrific at tennis in high school — #1 on the team, a state champion. You’re objectively great at tennis. But you’d lose in straight sets to the next level of greatness, someone who’s #1 on a D1 college team and vying for the national championship. That #1 college player is great. But they’d lose in the first round of a Grand Slam to a seeded player, someone in the top 32 in the world. And that top 32 player? They are exceptional. They’ve made it to almost the pinnacle of their sport. And yet, they have no chance of winning a Grand Slam at the moment, because they’d be beaten comfortably by Alcaraz or Sinner or Djokovic.

Since the start of 2017, only two men not named Federer, Nadal, Djokovic, Alcaraz, or Sinner have won a Grand Slam (Dominic Thiem and Daniil Medvedev, if you’re curious) out of 36 tournaments played. So there’s the level of greatness that those five players represent, the very highest level, and there are several levels of greatness (four? six?) below them, depending on your frame of reference. The gap between that highest level and even one level below is so stark that barely anyone has been able to touch their level for the past decade. That highest level on their best day can beat almost every other tennis player in the world, whatever level of relatively great, 6-0, 6-0.

What I’m describing is a version of the power law. It’s impossible to ignore as a venture capitalist, where a few investments and companies drive the majority of returns. It’s hard to internalize this sort of exponential function as a human (at least, it has been for me). But it’s out there, in all sorts of places. A few stories that command the most attention. A few cities in which most of GDP is concentrated. A few people that have the most wealth. Power, in the hands of a few.

What keeps me up is that I believe AI is already exacerbating this dynamic, and will only continue to, as any exponential function is wont to do. Since my post over the weekend about agentic engineering and how code generation has become much more autonomous since the start of the year, I’ve heard from a number of founders and companies, most of whom think that they are in the “bucket 2” I describe: “Those who are learning what’s possible by already generating a lot of code.” They may be, but the gap between them and the small set of companies in bucket 1 is huge.

AI builder and investor Matt Shumer, in a post that went viral this week, Something Big Is Happening, stated the following:

AI is now building the next AI

There’s one more thing happening that I think is the most important development and the least understood.

On February 5th, OpenAI released GPT-5.3 Codex. In the technical documentation, they included this:

“GPT-5.3-Codex is our first model that was instrumental in creating itself. The Codex team used early versions to debug its own training, manage its own deployment, and diagnose test results and evaluations.”

Read that again. The AI helped build itself.

This isn’t a prediction about what might happen someday. This is OpenAI telling you, right now, that the AI they just released was used to create itself. One of the main things that makes AI better is intelligence applied to AI development. And AI is now intelligent enough to meaningfully contribute to its own improvement.

Matt captured more eloquently than I did in my weekend post what is happening in code generation in these model companies: the models are now building themselves. And so, they are just going to get better and better, at an increasing rate, particularly when it comes to code generation.

It’s hard to understand exactly where this recursive self-improvement loop will take us in AI writ-large, beyond code generation, but one thing is for sure: a small handful of companies — Anthropic for sure, and likely OpenAI, Google, Cursor, Cognition too — are at the highest level of greatness when it comes to their ability to engineer software. And no one else is close.

---

I started next big thing to share unfiltered thoughts. I’d love your feedback, questions, and comments!

👇🏽 please hit the ♥️ button below if you enjoyed this post.