The Question of Gen AI End-User Usage
End-user usage is the Trillion dollar Gen AI question. If the AI bubble is to persist, users need to use GPT the way teenagers use TikTok. I highlight GPT because OpenAI (along with its carnival barker CEO) is the AI Bubble poster child. Here’s what we know, and what we don’t know:
We know there is demand for AI chips with which language models are built.
We know there is demand for models from the likes of Cursor, Microsoft (MSFT) (together more than 20% of Anthropic revenue), Google (GOOGL), Amazon (AMZN) and soon Oracle (ORCL) as the cloud services companies like to offer Gen AI as a feature to business customers (OpenAI said it has 1 million business customers). In the case of Microsoft and Google, they have jammed OpenAI and Gemini into all of their products.
We don’t know what Microsoft Copilot and Google Gemini user uptake is as it relates to using Gen AI features across each company’s product suite. However, I can guess that Microsoft Copilot uptake is weak because initially reported Copilot numbers from over a year ago were weak before MSFT stopped disclosing Copilot Pro figures. Importantly, lack of uptake has not stopped Microsoft and Google from increasing prices. Yes, MSFT and GOOGL are enjoying Revenue “lift” from Gen AI features, even if user uptake is not necessarily there.
We know that OpenAI says it has 800 million weekly users. We don’t know what repeat usage looks like, nor do we know the number of “premium” users.
We know that ChatGPT app usage went slightly down in September and flatlined in October. Nobody seems to want to acknowledge this fact.
We know that Sam Altman has said that OpenAI will achieve a $20 billion annualized revenue run rate by year end 2025 and grow to hundreds of billions by 2030. We don’t know what inputs are factored into these figures. For all I know, OpenAI may count cash infusions from investors as revenue.
My guess is that OpenAI will lose $40 Billion this year. The company has lost approximately $24.6 billion through the first 9 months of 2025. A $15 billion loss in Q4 after Q3’s $12 billion loss is hardly a stretch.
If OpenAI does go public next year, Sam Altman won’t be the CEO. Altman talks way too much, he is not a technical founder, he’s never built an LLM, and he’s untrustworthy.