Chris Paik: code is becoming free — creativity and agency are the new scarce resource

Summary

Chris Paik, partner at Pace Capital, argues that code is becoming a commodity. LLMs now handle the translation from natural language to working software for free. Creativity and agency are the scarce resources that will define the next era of software.

Paik frames the debate between concept thinkers and systems thinkers. He lands on the side of concept thinkers. With natural language instantly converted into working software, the ability to articulate a clear idea matters more than the ability to implement it.

Utility versus leisure

Paik divides contexts into utilitarian and leisure settings, and AI disrupts each differently.

In utilitarian settings, people want friction removed and humans out of the loop. Waymo over a chatty Uber driver. Silent coffee checkout over the iPad tip prompt. Any role that exists primarily to help someone reach a goal they already know they want is vulnerable. If what you do gets in the way, that job disappears.

Leisure contexts work the opposite way. People want to watch other humans, not machines. Chess engines have dominated human players for decades, yet Magnus Carlsen and Hikaru Nakamura have massive audiences and Chess.com has built a real business. Nobody watches Stockfish. The human story—the craft, the mythos, the constraints the artist worked within—is the product. Daniel Arsham chips a piece of rock off a wall with an axe, and the fragment commands a significant premium because of who swung it. AI can replicate the output but not the story behind it.

Jobs and the leisure expansion thesis

Paik is genuinely bullish on full employment. Productivity booms historically generate leisure time, and leisure time generates new categories of work. The first board games were invented concurrent with irrigation. Modern professional sports were born from factory workers competing on weekends after the industrial revolution. By that pattern, AI-driven productivity should expand leisure hours and, with them, demand for athletes, storytellers, chefs, and entertainers at scale.

Ken Burns illustrates the point. A ChatGPT deep-research report on the American Revolution is available now and may be more factually exhaustive than anything Burns produces. But people will pay to watch Ken Burns tell the story through his own lens, in his own voice, with his own interpretive choices. The same logic applies to David Senra's Founders Podcast—retelling biographies of great entrepreneurs in a way that is compelling precisely because it is filtered through a specific human perspective.

Apple and on-device inference

Paik calls Apple's positioning on on-device inference through its MLX library the most underappreciated announcement from WWDC. The library enables inference on Apple silicon without cloud calls. The current constraint on AI app development is that inference is metered. A viral free app can blow through a developer's credit limit overnight. Apple moving inference to the device is roughly equivalent to game developers no longer paying per rendered pixel. Paik expects that within one to two years, 80 to 90 percent of queries developers would currently route to a cloud-hosted model will be runnable locally, for free, with offline capability and privacy intact. He also notes Apple's Anthropic partnership on Xcode as an underappreciated move that reinforces the same direction. The outcome he is waiting for is an LLM-native "Flappy Bird moment" — a single-developer app that goes massively viral once inference cost hits zero at the edge.

Top-down versus bottom-up companies

Paik distinguishes companies by how they were founded. Top-down companies—Apple, Tesla, SpaceX, Amazon, Epic Games—are started by visionary founders who are largely indifferent to how execution happens. Those organizations hire innovators and build the internal machinery to ship multiple flagship products over their lifetime. The market sees the products. It misses that the real asset is the innovation engine itself.

Bottom-up companies—Google, Facebook, Instagram, Reddit, Airbnb, Twitter—are started by hackers who capture lightning in a bottle with perfect timing. The violent product-market fit that makes them successful also makes them structurally incapable of shipping a second flagship product. Every hire is an optimizer because there is no incentive to innovate when you can improve the core product by 1 percent. These companies expand through M&A instead, and often do not recognize their own architecture. Meta buying Instagram and ByteDance buying Musical.ly are among the best investments of the last two decades, but they are acquisitions, not inventions.

On taste, Paik believes taste is alive at Apple because the culture was built to hire innovative DNA from the start. At bottom-up companies, what looks like taste is often ornamental—a well-timed design agency engagement that gets misread as something deeper. Craigslist is a perfect bottom-up company. No taste required.