Open and closed models are on different exponentials
Interconnects · Nathan Lambert · 2026-06-01
Nathan Lambert argues that open and closed AI model ecosystems are operating on fundamentally different economic growth trajectories, with closed frontier labs like Anthropic and OpenAI commanding durable premium pricing through hardware-software integration while open models will dominate a larger but lower-margin fragmented enterprise market.
Appears in
Extraction
Topics: open-source-aiclosed-modelsai-market-structurecoding-agentsai-economics
Claims
- Closed frontier labs will maintain durable advantages over open models through deep hardware-software integration that generates returns open model ecosystems cannot replicate.
- Coding agents have demonstrated strong product-market fit at the top of the knowledge work market, with users unwilling to settle for open models once they experience frontier-quality agents.
- OpenAI and Anthropic could reach $2–10 trillion valuations in 5–10 years, forming an oligopoly analogous to today's cloud hyperscalers.
- Open model builders will eventually stop competing with closed models on benchmark indexes and instead serve lower-cost enterprise niche use cases at commodity pricing.
- Claims that recursive self-improvement will give closed labs an unassailable advantage over open models are overblown.
Key quotes
I would pay $2000/month for the tools today, especially knowing they'll get much better.
In this mental model, the frontier labs as businesses, will look like new, reimagined forms of a mix of Apple and Microsoft.
The key is that the open and closed model economies are operating on different exponentials.