AI industry is increasingly shaped by three competing forces: closed frontier-model companies, open-weight model developers and application companies building products on top of AI. Like the three-body problem in physics, the interaction between these forces makes the industry's future difficult to predict because every move by one group changes the incentives and strategies of the others.
The first force is closed frontier AI, represented by companies such as OpenAI and Anthropic. These firms are spending enormous amounts to develop increasingly capable models, but they face pressure to lower prices while recovering the huge costs of training and operating them. At the same time, open-weight models, particularly from Chinese developers, are improving rapidly and offering businesses capable alternatives at much lower costs, potentially weakening the dominance of proprietary APIs.
The third force is the growing ecosystem of AI application companies. These businesses are increasingly using open-weight models to reduce costs and gain control, while also developing proprietary data, workflows and specialized models to differentiate themselves. At the same time, frontier AI companies are moving into applications themselves, creating a two-way convergence in which model companies move downward into products and application companies move upward toward the model layer.
The broader takeaway is that the AI race is unlikely to have one simple winner. Competition should push AI prices lower, encourage greater use of open models and gradually force companies to demonstrate more measurable productivity gains. As models become increasingly commoditized, value may shift toward applications, proprietary data and specialized workflows. The industry is therefore becoming a complex ecosystem where models, open infrastructure and applications continually reshape one another, making the final outcome impossible for any single company or country to dictate.
