Andrew Keys found Ethereum before it launched, when ETH was around 30 cents. Eleven years later, he’s betting on a very different layer of the technology stack: AI infrastructure.

In Ohio, a former paper mill is being repurposed for AI compute. Keys says the first transaction represents roughly 10,000 GPUs, while Alignment Engine has disclosed $1.17 billion in four-year AI infrastructure orders. Amanda and Andrew get into why AI agents may need Ethereum to transact, whether open-source models crush proprietary AI economics, why ETH still has a value-capture problem, and whether ETH treasuries justify their costs versus staking-enabled ETFs.

Timestamps

00:00 Finding Ethereum Before Launch

02:54 Does the Machine Economy Need Ethereum?

04:44 x402 and Standards for AI Payments

06:37 AI Slop and the Provenance Problem

08:31 Music, IP and Onchain Attribution

10:53 Is AI Breaking the IP Regime?

13:11 Open Models vs. Proprietary AI

14:35 OpenAI, Anthropic and Open-Source Economics

18:50 The AI Infrastructure Spending Race

21:19 $1.17B in AI Infrastructure and 10,000 GPUs

24:50 Why AI Compute Could Need Ethereum

27:09 Turning GPU Compute Into a Market

31:19 If Ethereum Wins, Does ETH Capture Value?

36:11 Why the Ether Machine SPAC Was Terminated

39:28 Is the ETH Bear Market Over?

40:51 What Ethereum Can Learn From Hyperliquid

41:57 ETFs vs. ETH Treasuries

43:54 AI + Ethereum: Two Halves of One Economy

Stay Connected

https://www.youtube.com/@endgamepodcast

https://x.com/amandacassatt

https://x.com/serotonin_hq

◽️Apple Podcasts & Spotify

Endgame with Amanda Cassatt | Podcast on Spotify

https://podcasts.apple.com/us/podcast/endgame-with-amanda-cassatt/id1801809440

#Ethereum #ArtificialIntelligence #AndrewKeys #AIAgents #Endgame