Y Combinator CEO Garry Tan Reveals $200K Annual AI Token Spend, Envisions Personal AI Future
The High Cost of an AI Obsession
In a forthcoming episode of Walter Isaacson’s podcast The Builders, set to publish on Tuesday, Y Combinator chief Garry Tan disclosed that his passion for AI agents drives him to spend between $100,000 and $200,000 each year on tokens. He also joked that he has suffered from “cyber psychosis” and sometimes barely sleeps because of his excitement about working with AI.
“I spend about maybe $100,000 to $200,000 a year on tokens,” he told Isaacson.
Tan relies on a suite of 23 custom Claude Code skills, collectively called gstack, which he has open‑sourced on GitHub to help others understand what agents can do.
From Personal Assistant to Company Brain
Tan’s AI serves as both a research aide and a life manager. He relies on it to research Isaacson’s upcoming episode and prepare for potential questions, while the agent also reads his email, scans his Dropbox, and ingests two decades of personal therapy notes stored in Apple Notes.
“It’s read all of my email and all of my Dropbox. It’s read, like, all of my Apple Notes, even my therapy notes going back 20 years,” he said.
He likens the experience to having a Jiminy Cricket on his shoulder—constantly offering suggestions and anticipating his needs.
Y Combinator itself employs an internal AI agent named Stray Light. The agent has a distinct personality and a “company brain” that holds the complete context of every founder the accelerator has backed, every email exchange, and all past applications.
Lessons from an Early Adopter
As an early OpenClaw adopter who now counts himself among its power users, Tan emphasizes that AI agents should not be treated like traditional software. He warns that an excessive focus on repeatability and safety can diminish the human‑like qualities that make agents valuable.
“Some people are so worried about repeatability and safety that they actually trade off all of the beautiful parts of working with an AI agent, which basically can be so similar to a human,” he said.
He uses his agents for countless work projects and coding tasks, demonstrating how the technology can augment both professional and personal workflows.
The Vision for a User‑Controlled Future
While acknowledging that his token spending is a privilege unavailable to most, Tan argues that personal AI will become mainstream only if costs come down. He is skeptical of free services from giant tech firms that control critical algorithms.
“This is a huge company that controls the algorithm, and we all wish that they didn’t at some level,” he said. “So you could imagine a personal AI that you control, that you program.”
His disclosure of spending up to $200,000 annually underscores both the current expense of cutting‑edge AI and the transformative potential he sees in a more democratic, user‑controlled AI ecosystem.
Conclusion
Tan’s candid interview illustrates how a leading figure in Silicon Valley is living the promise—and the cost—of AI‑driven productivity. If the technology becomes more affordable and remains under user control, he believes personal agents will evolve from niche tools into everyday companions, reshaping how founders, executives, and everyday users operate.








