Why the Rise of Open Source AI Isn't Hurting Anthropic Yet
Key Takeaways
- ▸Anthropic maintains pricing power despite open source competition, capturing over 50% of AI spending on major platforms
- ▸Open source models are winning on token volume but losing on revenue, with Opus 4.8 commanding 23x higher token prices than DeepSeek V4 Flash
- ▸Frontier and open source models occupy different lifecycle phases—frontier labs own discovery, open source owns production
Summary
Despite the surge in open source AI models like DeepSeek and GLM-5.2 capturing significant token volume share on platforms like Vercel and OpenRouter, frontier AI labs—particularly Anthropic—continue to dominate overall spending. Anthropic accounts for more than half of AI spend on Vercel's platform, with its Opus 4.8 model commanding prices roughly 23 times higher than competing open source alternatives. According to Decagon CEO Jesse Zhang's analysis, frontier and open source models represent two phases of the same lifecycle rather than direct competitors. Frontier models are used by companies to prove out new use cases, which then migrate to cheaper open source alternatives as they mature; simultaneously, new use cases continuously emerge, ensuring steady demand for expensive state-of-the-art models.
- The rapidly expanding market of AI-addressable tasks is large enough to support both premium frontier models and commodity open source alternatives
Editorial Opinion
Anthropic and other frontier labs appear to have found a durable moat through their position in early-stage AI deployment discovery. Rather than being commoditized by open source, they're systematically moving up-market to premium use cases that require frontier capabilities—a positioning that looks remarkably resilient. This suggests the AI economy is maturing into a sustainable two-tier structure rather than collapsing into commodity pricing, which validates the long-term viability of premium frontier model providers.
