Home / Founders / Jake Cannell
Cannell argued years before the deep learning boom that intelligence comes down to compute, and he co-founded Vast.ai to rent out the world's idle GPUs so that control of AI computing power stays spread across many hands.
Jake Cannell is co-founder and chief executive of Vast.ai, a Los Angeles, California-based marketplace where independent hosts rent out GPUs to AI researchers, startups and companies at prices set by supply and demand. The company describes him as a machine learning engineer, GPU programmer and AI theorist, and he was writing about where AI was headed years before he started it.
Writing About Compute on LessWrong
Starting in 2010, Cannell published essays on LessWrong, an online forum about rationality and AI, arguing that intelligence depends mostly on computing power, according to Vast.ai's company history. His 2015 essay, The Brain as a Universal Learning Machine, argued that the human brain is mainly one general-purpose learning system, and Vast.ai says it forecast human-level machine vision and language from scaled-up deep learning.
Co-founding Vast.ai With Christian Horne
Christian Horne, another LessWrong writer, shared his view that if whoever controlled the most compute controlled the most powerful AI, then the future of AI would be decided by the deepest pockets. The two incorporated Vast.ai on June 28, 2016, to connect underused GPUs in gaming rigs, mining farms, research labs and small data centers with people who could not afford big cloud providers' rates. "A world where compute flows freely to thousands of independent researchers is a fundamentally different world than one where it is locked behind the pricing walls of AWS, GCP, and Azure," Cannell said in 2016, as quoted by the company.
Launching the Marketplace on Reddit
The founders spent about two years building host onboarding, search, pricing and Docker-based instance management before launching in September 2018 to friends, family and a post on Reddit. Vast.ai says 2024 was its biggest year, with 310 percent growth, and it now lists more than 17,000 GPUs from over 350 independent hosts in more than 40 data centers. It opened its first office, in Los Angeles, in July 2024 and a second in San Francisco, California, in 2025, and it has more than 40 employees, according to its press kit.
