EPISODE · Mar 13, 2026 · 1H 24M

Is AI a Bubble?

from Lumida Wealth : Non-Consensus Invest Beyond the Ordinary · host Justin Guilder

(01:55) Anthropic Trillion Talk (03:12) Mainframe AI Mindset (04:38) Value Based Pricing (07:56) Local Agents Shift (11:23) Rent Seeking Economics (15:14) Diffusion Reality Check (17:10) RPO Debt Spiral (20:02) Hyperscaler Capex Trap (21:45) Bailouts And Mergers (24:23) Nvidia Knock On Risk (35:15) TSMC Discipline Factor (36:49) China Taiwan Risk (38:37) Memory Trade Returns (40:05) Micron Valuation Debate (43:10) Consumer Memory Squeeze (45:35) Upstream Supply Winners (45:48) Semis Look Crowded (46:12) Power Gating Excuse (47:19) Utilities Winners Losers (50:57) Energy Demand Secular Bull (52:17) AI Agents Personal Compute (53:57) Open Claw Origin Story (59:01) What OpenAI Bought (01:04:57) Elon SpaceX Valuation Doubts (01:08:29) Space Data Centers Hype (01:12:17) China EV Manufacturing Edge Connect with Us Online:Lumida NewsLumida DealsLumida WealthLumida ETF Watch the video on Youtube: https://www.youtube.com/@Lumida_Wealth🌐 Website: https://www.lumidawealth.com🐦 Twitter Follow us on https://twitter.com/LumidaWealth🎵 TikTok: https://www.tiktok.com/@lumidawealth📸 Instagram: https://www.instagram.com/lumidawealth📘 Facebook: https://www.facebook.com/lumidawealth 

Episode metadata supplied by the publisher feed · Published Mar 13, 2026

Embed this episode

Attribution link and audio player

What this episode covers

Ram Ahluwalia interviews Michael Parekh about whether today’s AI boom is becoming a bubble. They question Anthropic and OpenAI leaders forecasting trillion-dollar revenues and very fast GDP growth, arguing the timelines may be too optimistic, costs are extremely high, and a slowdown could lead to defaults on large “revenue performance obligations” owed to firms like Microsoft and Oracle. Parekh says AI is still in a “mainframe stage” where a few big labs assume they will control most computing, but new local AI agents like the open-source “Open Claw” show more computing can move onto personal devices, which could reduce data-center demand assumptions.