Skip to content
Sonic
AI
Sonic
AI
Home
Discover
Ask Sonic
Projects
Use with Claude or ChatGPT
Show me around
Request source or feature
Training AI Models Without a Billion-Dollar Data Center | Steffen Cruz of Macrocosmos, Sonic AI
Home
/
Eye On A.I.
/
Training AI Models Without a Billion-Dollar Data Center | Steffen Cruz of Macrocosmos
Eye On A.I.
Notify me
•
May 25, 2026
•
47:02
Interview
Training AI Models Without a Billion-Dollar Data Center | Steffen Cruz of Macrocosmos
Stefan Kruse
(Co-founder and CTO, Macrocosmos, guest)
Get the full transcript next time Eye On A.I. releases an episode
Summary, key quotes, top claims, and the searchable transcript — emailed automatically. No card needed.
Sign up
Executive Summary
Macrocosmos is building a decentralized compute network called IOTA on the Bittensor blockchain to train large AI models at 10-20% of the cost of centralized providers.
The network aggregates underutilized consumer and enterprise hardware (e.g., Mac Minis, GPUs) through a "Train at Home" program, aiming to scale from 500 to 5,000 compute nodes by mid-year.
The company's initial focus is on the computationally intensive pre-training phase, with a goal to train a 70-billion-parameter model to prove the technology's viability.
Macrocosmos is transitioning from research to commercialization, with commitments from startups to begin using the platform in the second half of the year.
Continue your research
Keep pulling the thread on Stefan Kruse.
Decentralization of AI Compute
Economic Incentives for Distributed Networks
The Bittensor Ecosystem as a Foundation
Or ask anything across 400+ expert conversations
12
quotes
Transcript
Key Arguments
Analysis
Quotes & Entities
12
Related
Loading transcript...
Processed Jun 17, 2026
Daily intelligence brief →
yt-dlp + mlx-whisper + Gemini