Naveen Rao, founder of Unconventional AI and former VP of AI at Databricks, discusses how —not model training—will become the dominant economic opportunity in AI, potentially reaching a $10 trillion market. He argues that energy will be the biggest constraint on AI scaling, and that chip specialization will prevent commoditization of AI hardware.
Rao also addresses structural shifts in the AI stack: traditional software businesses will matter less as AI capabilities shift investment and value, pure software investors face structural disadvantages, and robotics could become 100 times larger than coding agents. He warns that 80% of AI token flows may move through open models, and discusses why deep-tech venture metrics no longer apply to AI infrastructure companies.
This is an AI-written summary. The original reporting belongs to @20VC. How we write these · Spot a mistake?