Geo Chen
· Fidenza Macro
· April 15, 2026 at 04:31
· ⏱ 7 min read
| Read on Substack ↗
Summary
The AI supercycle is accelerating due to exponential demand for AI tokens from agentic AI, creating a flywheel of compute investment and cost reduction. Meanwhile, the K-shaped economy persists with the top end benefiting from AI and the bottom end struggling under liquidity and energy pressures. The article argues that value will shift from commoditized products to human-centric, relational goods and services, but offers no explicit trade ideas.
•Cost per AI token has fallen by a factor of 10 to 1,000 depending on model quality.
•Anthropic's annual revenue run rate has surpassed OpenAI's despite raising less capital.
•Starbucks, with a $112 billion market cap, exemplifies a highly standardized product that is easy to automate.
•LinkedIn’s Skills on the Rise 2026 report found surging demand for communication and creative thinking; job postings mentioning 'storytellers' doubled over the past year.
•Snowflake's CEO predicted in late 2025 that as execution becomes commoditized, strategic thinking and vision will separate high-performing organizations.
•The article cites Dario Amodei (Anthropic) for a framework on how AI capex can rise alongside accelerating revenue and demand for tokens.