Diverging Exponential Growth Trajectories for Open and Closed Models

The future balance of power in the AI ecosystem will be determined by economic logic. In particular, the coding agent field is emerging as the first market where users are ready to pay a significant premium for higher intelligence.

The use of coding agents that surpass the Opus 4.5 and Codex 5.2 thresholds is establishing itself as a tool that goes beyond mere convenience to substantially increase output. Users will choose the best model over a 'good enough' one, even if it costs more, for the sake of work efficiency.

Leading closed-model developers such as Anthropic and OpenAI are expected to dominate the market through strategies such as:

• Delaying the API release of top-tier models to protect token supply and prevent distillation
• Focusing on specific high-margin use cases
• Building an integrated technology ecosystem that combines models, hardware, and software

These leading companies are expected to form an oligopoly similar to today's cloud market, with corporate valuations in the $2-10 trillion range within the next 5-10 years. They will dominate the market by optimizing not just for benchmark scores, but for utility per second or utility per watt.