AI is making it cheaper to build games. It's doing nothing for the cost of building the right game. That distinction is about to crush more studios than any technology shift in the past decade.
Ran Mo is the CEO of Proxima and creator of Suck Up — arguably the first commercially successful game to use AI at runtime. The game generated over 100 million YouTube views with zero marketing spend. Before founding Proxima, Ran led product teams at EA on The Sims franchise and spent time at YouTube and BCG.
In this episode, Ran live demos AI-assisted coding in Unity, breaks down his "trunk and leaves" framework for where AI helps (and where it destroys your codebase), and makes the contrarian case that AI in actual gameplay is overhyped — even though he built the first game to prove it works.
We go deep on: why vibe-coded games are architecturally unusable at scale, the power law that's about to eliminate mid-tier studios, why Ran hired a marketer with 1M TikTok followers, and the Taoist philosophy that keeps him sane while Silicon Valley burns out around him
Whether you're a game dev, studio head, or just trying to understand where AI actually moves the needle in creative industries, this conversation will sharpen your thinking.
🔗 Ran Mo's newsletter: newsletter.ranmo.me 🔗 Proxima: proxima.gg
Outline:
0:00 Intro
1:15 Ran Mo's background and founding Proxima
4:38 How Suck Up actually uses AI
8:49 Why Proxima won't use AI for art or design
12:32 Live demo: AI coding in Unity
22:33 Vibe coding gone wrong — the Stardew Valley test
25:14 The trunk and leaves framework
29:24 Real productivity numbers (not Twitter hype)
34:57 The technically capable designer
38:58 Power law distribution is coming for games
43:48 Suck Up's $0 marketing playbook
46:04 The Courage to Do Nothing
52:35 The velocity trap
58:13 Hot take: AI in gameplay is overhyped
61:01 Ran's personal story
66:15 Go your own way
69:15 Final advice: slow down to speed up