Archive — US — techcrunch
Why the rise of open source AI isn’t hurting Anthropic … yet
A new theory by Decagon CEO Jesse Zhang argues that open source AI and frontier models, like those from Anthropic, are not direct competitors. Instead, both coexist as different phases of AI deployment. Although open-source usage grows, Anthropic maintains over half of AI spend due to a growing market. The dynamics suggest a stable two-tiered AI economy where frontier labs focus on discovery while open-source dominates production.
Key points
- Jesse Zhang's theory suggests open source AI and frontier models serve complementary roles, rather than competing.
- Mature deployments are opting for lighter open source models, but overall spending on frontier models has not decreased significantly.
- Data from Vercel shows that while usage of models like DeepSeek has increased, Anthropic still retains a majority of the spending.
- The rise of open source may be due to an expanding market for AI tasks, allowing frontier labs to maintain their status.
- The model appears to indicate a stable two-tier economy where frontier labs focus on early discovery, while open source models are used in production.
Why it matters
- The trend of transitioning to lighter models correlates with the broader 2025 IT agenda, emphasizing adaptability and cost-effectiveness.
- Different phases of AI development reflect historic shifts in technology adoption where initial investments pave the way for scalable alternatives.
- The notion of frontier labs acting as commodity inputs for application layers is aligning with current trends where foundational technologies support diverse use cases.
Despite the rise of open source AI, frontier models like those from Anthropic hold a significant market share, indicating a durable coexistence in a rapidly evolving AI ecosystem where both innovation and practicality thrive.