“The board mandate should not be to cut the AI bill. It should be to explain it.”
That is the central idea in this week’s Elemental AI: The Briefing.
Chris and I are talking about token math of AI and the paradox that unit costs can fall while total AI spend keeps climbing.
AI spend is shifting from predictable subscriptions to usage-based consumption. That turns AI from a manageable software line item into a highly variable operating cost - one that hits the P&L immediately, while the value takes much longer to prove.
Chris and I sit with that gap for about 25 minutes.
What the episode covers:
• Why unit prices are falling while total AI bills are doubling - Jevons’ Paradox in the enterprise.
• The payroll comparison - why an unmanaged prompt bill behaves like unmanaged headcount that never sleeps.
• Tokenmaxxing - how “everyone should use AI” became a variable cost mandate, and why companies are already pulling back leaderboards.
• Model routing - matching the cost of the tool to the value of the task, instead of hiring a senior partner to do data entry.
• The single best question to ask management in the next board meeting.
→ Full written breakdown: Issue 52 - The Weird Token Math of AI. [SUBSTACK URL]
Let’s get elemental.
