it’s much worse than stated here

starting point is to put user numbers on an S-curve that by 2030 reaches 3bn, “equivalent to 44 per cent of the world’s adult population” ex China.

while article says OpenAI has 800m users (who’ve used service as curiosity), its statement of 40m paid users is extremely exaggerated. 5m paid users is a high end estimate. There is ample free AI use/access.

The problem with all LLM makers is that it is an extreme competition field, and there is 0 moat to switching providers if users are smart (ie. all tools that use LLMs can switch to a different one under the hood, and you should use those front end tools or basic web interface). Google, Anthropic released better models this week, and last month Qwen and Kimi released better models than gpt 4, and other US models.

Unlike MS Office that has a 3 year release schedule, with UI and file format lock in, these models are replaced every 3-4 months, with no switching obstacles, and old model essentially thrown away.

It’s simply too much to ask investors/world that a company with no money is going to perpetually dominate a field where rest of world has far more money and talent than it does. Even if one of its releases does dominate, the reign will last a matter of weeks, and models are often used to train other cheaper models.

The AI companies with extortion power are TSMC, memory makers, and for now, NVIDIA. Downstream has no competitive advantage. Even your AI girlfriend will get swapped for a newer hotter model.