Foundation Model Usage-Based Subscriptio Diagram

Strengths

• Superior model performance and multi-modal capabilities, resulting in spillover productivity for end-users
• Consumption-based pricing is naturally tied to usage; higher volume leads to better unit cost efficiency
• Strong market share lock-in effect with high migration costs

Weaknesses

• Inference costs represent a heavy portion of revenue, posing a risk of losses amid price wars
• Revenue is highly correlated with the client's own AI usage, leading to significant volatility

Opportunities

• Surge in Agent use cases drives exponential growth in API calls, creating space for token volume expansion
• Enterprise AI adoption is shifting from experimentation to core business, opening a window for stable subscription revenue

Threats

• Substitution by open-source models and domestic models offering better pricing and flexibility
• Systemic price reductions or subscription discount wars driven by potential regulation and competition