The AI landscape is rapidly evolving, with a recent focus on the capabilities of open-weight models from China, such as Alibaba Qwen, DeepSeek R1, V3, and V4, and Moonshot Kimi K3, compared to their closed-weight counterparts from the US, including OpenAI GPT, Anthropic Claude, and Google Gemini. This shift in attention has sparked a debate about the potential impact on the revenue streams of American and European AI model makers, particularly in the context of the substantial investments in datacenters and hardware required to support GenAI aspirations. The article delves into the spending patterns on AI models and platforms, highlighting the rapid growth in GenAI model revenues, which have slowed significantly in 2025 and are projected to continue decelerating in 2026. The author questions the accuracy of Gartner's forecasts, suggesting that the actual revenues for GenAI models and platforms may be lower than expected, given the significant investments made by companies like OpenAI and Anthropic. The article also explores the market dynamics of domain-specific and specialized GenAI models, which are growing twice as fast as broader foundation models, and the maturity of AI development platforms, which are already a larger market than the models themselves. The author argues that the market for GenAI platforms is growing slower than for the models, possibly due to API-heavy revenue streams or companies' desire to control their GenAI fates. The discussion then turns to the potential threat of fully open-source models, particularly the Nemotron 3 foundation models by Nvidia, and the implications for the long-term growth of GenAI model licensing and subscriptions. The author concludes by questioning the future of AI model makers' businesses, given the disparity between actual revenues and the investments required to support GenAI aspirations.
