Dan Niles: You need to have a viable open source community
Speakers
Dan Niles
— Founder & Portfolio Manager, Niles Investment Management
Summary
Dan Niles argues that open source AI models will commoditize the proprietary model layer, shifting value to cloud platforms and semiconductors. He highlights accelerating revenue and margin growth at AWS, Azure, and Google Cloud and states his portfolio is fully positioned in infrastructure names.
- Open source models cut token costs by 90% and reduce reliance on top-tier proprietary models.
- The AI model layer (Anthropic, OpenAI) will become commoditized over time.
- Value accrues to infrastructure players—cloud platforms and semiconductors.
- AWS, Azure, and Google Cloud all showed accelerating growth and margin expansion in recent quarters.
- Niles is fully allocated to cloud platforms and semiconductors.
- Companies must keep proprietary data proprietary to avoid competitive risks.
Ideas
Cloud and semiconductors benefit as models commoditize
As open source models proliferate and reduce token costs by 90%, workloads will be routed to the most appropriate model for each task, commoditizing the proprietary AI model layer. Value will accrue to infrastructure players, specifically cloud platforms like AWS, Azure, and Google Cloud, and semiconductors. Recent earnings from AWS, Azure, and Google Cloud show accelerating revenue growth and expanding margins, supporting this view. Niles is fully positioned in these infrastructure names.
