Newsletter Archive
A running archive of research on AI infrastructure, model competition, and the bottlenecks that matter most across the stack.
Bits barely moved. Prices jumped roughly 60%. That gap is what a physical bottleneck looks like in an income statement.
New essays are published here with durable links, clear publication dates, sources, and related research. Substack remains the easiest way to get each one by email.
Free to read in your inbox or here · No spam · Unsubscribe anytime.
Three essays that frame the archive: what constrains AI, how model economics are shifting, and why task cost matters more than token price.
The most important question in AI investing is what the system can physically support.
The price per token no longer tells you what a task costs.
Demand is real. Spending is extreme. The missing pieces are not what most people think.
Earlier newsletter essays, presented as durable research notes instead of a generic feed.
The capability gap is holding. The market willing to pay for it is narrowing.
Demand is real. Spending is extreme. And the missing pieces aren't what most people think.
Seven frontier AI models in 18 days. The price per token no longer tells you what a task costs.
What Anthropic vs. OpenAI means for enterprise software, compute demand, and how you work.
Why the most important cost driver in AI isn't compute, and what happens when it gets compressed.
The most important question in AI investing is what the system can physically support.