What you'll learn

• 1 (02:50) **The AI bubble's next phase: token cost shock** - Ed Zitron introduces the episode, noting that the AI bubble's bursting may accelerate due to a few key events like an AI company dying or a hyperscaler cutting capex. He highlights Goldman Sachs analyst Rich Piratovsky's view that spending is just to stay competitive, and that the first hyperscaler to slow spending would be rewarded.
• 2 (04:26) **Enterprise AI token pricing goes live, causing panic** - Both Anthropic and OpenAI began charging enterprise customers (150+ users) the actual cost of token spend as of mid-Q1 2026, ending the era of burning thousands of dollars worth of tokens for a $200 subscription.
• 3 (07:04) **Three months in: enterprises are "screeching like they're being pecked to death by birds"** - The real cost of AI is hitting home, with no plan for ROI or cost management from any company integrating AI.
• 4 (07:21) **The open-source threat: ZPU's GLM 5.2** - Chinese AI lab ZPU's GLM 5.2 coding model is competitive with Anthropic's Opus 4.8 at a quarter to a sixth of the price.
• 5 (09:35) **The brand-name inertia problem** - Customers still prefer paying for name brands like OpenAI and Anthropic, even though their models are constantly broken and unprofitable. A shift to open source would require hyperscalers like Microsoft, Amazon, or Google to offer GLM 5.2 in their foundries.
• 6 (10:28) **The perfect moment for open source to take over** - With everyone freaking out about costs, now is the time to separate genuine AI users from those just doing "cargo cult" AI.
• 7 (13:34) **Open router rankings show a dramatic move toward open source** - The top 10 most popular models are all open source except for Opus 4.7, Opus 4.8, and Sonic 4.6, plus a mysterious free model called Al Alpha.

+ Full timestamped outline available in the app