last week, openai dropped gpt-5 at $10 per million tokens

at almost 10x cheaper than anthropic’s flagship and on par in performance benchmark, it’s openai’s declaration of its willingness to destroy the margins in “training and delivering” inference.

Ethan goes on to describe the “four kingdoms of profit” for tech companies with a graph showing advertising, cloud infrastructure, productivity software, and systems of record.

you pick one kingdom. you defend it with your life. you burn everything else to defend it.

“inference was supposed to be the next AWS” - and Anthropic and Claude, especially for code, was/is outperforming OpenAI, so maybe they are pivoting?

A good read no matter what you think of the conclusions.

ChatGPT has become the noun and the verb for the entire industry, much like “Google” before them.

And Google very much is playing catchup, although owning both web search and the productivity suite gives them a built in distribution.

Looking at the “four kingdoms”, the productivity space has Microsoft at $69B, Adobe (???) at $18B, and a very large “Others” at $45B.

This is where Notion sits and a bunch of other software that is at least nominally designed to help you be efficient.

How much of this gets eaten by generalist [[malleable software]]?

At enterprises, might malleable software grow out of the “system of record” software? They all claim to do this already (eg Salesforce apps).

All in all this still makes me want to fight so that not everything gets sucked into the big tech clouds and/or becomes filled with advertising.

This productivity space is the main area to focus on if we want to provide software that people & businesses will pay for and be built in a way that supports user agency.