Despite fears that cheaper, open-source models would commoditize the market, the opposite is happening. While token usage for cheaper models is rising, the actual share of economic value (wallet share) is increasingly flowing to expensive frontier labs like Anthropic and OpenAI.

OpenAI and Anthropic form a powerful duopoly at the "frontier" of AI, commanding premium prices like Apple. A second, commoditized tier of open-source and lagging models exists, where value is captured through compute and services, not the model itself. This creates a clear market separation between premium and "good enough" AI.

Decagon's CEO explains a paradox: while open-source AI usage grows, its market share shrinks. This is because open-source is ideal for scaled, defined tasks, but most enterprise AI is still in the experimental phase, where powerful, flexible frontier models are preferred.

Vercel data shows open source models have flipped to over 60% of AI token volume, indicating mass adoption for high-volume tasks. However, analysts predict closed, frontier models will still capture the vast majority of economic value, as premium intelligence for critical tasks commands a significant price premium.

An RBC analyst predicts an "80/20 world" for AI, where 80% of workloads can be handled by older, cheaper, or open-source models. However, the largest portion of the total addressable market (TAM) in terms of dollars will remain concentrated in the 20% of complex tasks that require cutting-edge frontier models.

The AI model market has two clear segments: expensive, high-IQ frontier models for critical tasks like cybersecurity, and small, cheap, fast models for high-volume, simple tasks. Mid-tier models are struggling to find a clear product-market fit, as users gravitate to either extreme.

The market for AI models is bifurcating. Users either pay a premium for top-tier frontier models for high-stakes tasks like cybersecurity or use extremely cheap, small models for high-volume, simple tasks. Mid-tier models struggle to find a viable use case, getting squeezed from both ends.

The greatest value in AI won't be captured by frontier labs alone. Instead, companies in the "applied layer" are incentivized to build routing systems that use expensive frontier models for high-level orchestration while deploying cheaper open-source models for bulk tasks, creating a more efficient, barbell-shaped cost structure.

Contrary to commoditization fears, the rise of powerful open-source AI models actually enhances the value of leading frontier models. The most advanced models become 'orchestrators,' leveraging armies of cheaper, specialized AIs, making their superior intelligence even more valuable for complex tasks.