Season 1 · Episode full</itunes:episodeType> <itunes:season>1</itunes:season> <itunes:episode>8
Week of June 22nd 2026
The AI Bubble, Translated Into Your Invoice
Five companies — Amazon, Google, Microsoft, Meta, and Oracle — will spend roughly $725 billion building AI this year, and they're losing money on your $20 subscription on purpose. That's great… until someone wants the money back. And the only pocket they can reach is your monthly bill. This week: why the cheap AI got so cheap, why it won't stay that way, and the 30-minute move that keeps any one tool from ever trapping you.
In this episode:
- The $725B bet — 2026 Big Tech AI capex is about $725B, up ~77% in a year. Meta alone guided to $125–145B (Fortune). JPMorgan sees ~$5 trillion flowing into AI infrastructure by 2030 against a sliver of real revenue — the "Grand Canyon gap." The Fed flagged AI as a top systemic risk; Bill Gurley, who called the dot-com top, sees a "reset" coming. Translation: your cheap AI is cheap because investors are subsidizing it.
- The $20 era is ending — Today's ~$20/mo plans are a VC-funded land grab, not a price (Boston Globe). Serving users already eats more than half of the big labs' revenue; analysts (Josh Bersin) expect everyday plans to drift to $25–30+. Both OpenAI and Anthropic already roughly doubled their newest model prices and killed enterprise all-you-can-eat deals. Budget at double today's price.
- Flat-rate is dying — meet "credits" — AI is shifting from flat monthly pricing to usage/credit billing (the same catch we flagged on Chatbase last week). 78% of companies got hit with an unexpected AI charge last year (Zylo), and the average small business now runs ~5 AI tools. Ask: does it bill flat or by the meter, and can I cap it?
The theme: The cheap AI is cheap because someone chose to pay for it — and that won't last. The operators who win the reset aren't the ones with the most AI; they're the ones who can swap any tool without bleeding. Rent the model. Own the data. Keep one hand near the exit.
🔗 Show Notes & Sources
The $725B bet
- Futurum Group — 2026 AI capex sprint (~$725B aggregate, +77% YoY): https://futurumgroup.com/insights/ai-capex-2026-the-690b-infrastructure-sprint/
- Fortune — Meta bumps 2026 capex to as much as $145B: https://fortune.com/2026/04/29/meta-zuckerberg-145-billion-ai-spending-roi/
- AI capex bubble 2026 — JPMorgan ~$5T-by-2030 "Grand Canyon gap": https://tooldirectory.ai/blog/ai-capex-bubble-2026-where-the-revenue-actually-is
- AI bubble 2026 — Fed systemic-risk warning + GPU/capex context: https://medium.com/@svnkrmkr/ai-bubble-2026-is-it-real-capex-fed-warnings-gpu-lifespans-b5db2178d350
- Techloy — Bill Gurley "reset" + the $1T data-center question: https://www.techloy.com/ai-data-center-bubble-2026/
The $20 era is ending
- The Boston Globe — "ChatGPT price increases are coming. Claude, Gemini, and others, too.": https://www.bostonglobe.com/2026/04/29/business/chatgpt-price/
- MindStudio — Why the $20/month era is ending (inference > half of revenue): https://www.mindstudio.ai/blog/ai-pricing-shock-end-of-cheap-subscriptions
- Josh Bersin — "AI Prices Are Going Up, Up, Up": https://joshbersin.com/2026/05/ai-prices-are-going-up-up-up-and-what-this-means-for-enterprise-ai/
- The Decoder — Anthropic/OpenAI raise model prices, drop enterprise discounts amid price war: https://the-decoder.com/anthropic-backs-off-unpopular-billing-overhaul-as-price-war-with-openai-looms/
Flat-rate is dying — meet "credits"
- Zylo — 2026 SaaS Management Index (78% hit by unexpected AI/consumption charges): https://zylo.com/blog/ai-cost
- HubSpot — Buyer's guide to credit-based AI pricing: https://blog.hubspot.com/website/ai-credits-buyers-guide
- SBE Council — SMBs run a median of ~5 AI tools; 82% have invested: https://sbecouncil.org/2026/04/25/the-ai-tools-small-businesses-are-using/
Tool Spotlight — NotebookLM
- NotebookLM plans & pricing (free Standard, Plus ~$7.99, Pro $19.99): https://notebooklm.google/plans
- NotebookLM pricing 2026 breakdown (free-tier limits): https://felloai.com/notebooklm-pricing/
- NotebookLM for Business — 5 use cases (SOPs/onboarding): https://www.itgenius.com/blog/notebooklm-for-business/