AI demand for GPUs is exploding – and most of that capacity is locked inside underused data centers.
In this episode, I talk with Mark from Aethir, a decentralized GPU cloud that aggregates idle, enterprise-grade GPUs into a global network. We discuss how Aethir feels like AWS on the front end but works like “Airbnb for data centers” behind the scenes, why compute demand outpaces supply, and how they keep latency low across 90+ countries.
Mark also explains Aethir’s token and revenue model, their work with EigenLayer, and why he believes solo founders now have superpowers in an AI-native world.
Nothing in this episode is financial or investment advice.
Key timestamps
[00:00:00] Intro: Sam introduces Mark and Aethir’s decentralized GPU cloud.
[00:01:00] Mark’s journey: From oil and gas infra and biotech to building GPU infrastructure for AI.
[00:04:00] What Aethir is: AWS-style GPU cloud on the front end, “Airbnb for data centers” on the back end.
[00:06:00] Enterprise-only GPUs: Why they only use data-center-grade hardware and no consumer devices.
[00:07:00] Exploding demand: GPU demand 6–8x supply, with inference-heavy apps driving the next wave.
[00:14:00] Global coverage: 90+ countries and routing users to nearby nodes for low latency.
[00:31:00] Business model: 20% protocol fee, 80% to GPU hosts, plus token rewards and staking for large clusters.
[00:39:00] Solo founder era: Why one-person AI-native companies will be extremely powerful.
[00:41:00] Mark’s message: Focus on projects with strong fundamentals and keep building through cycles.
Connect
https://www.linkedin.com/company/aethir-limited/
https://www.linkedin.com/in/markrydon/
Disclaimer
Nothing mentioned in this podcast is investment advice and please do your own research.
It would mean a lot if you can leave a review of this podcast on Apple Podcasts or Spotify and share this podcast with a friend.
Get featured
Be a guest on the podcast or contact us – https://www.web3pod.xyz/