Find XO: Why VCs Are Shifting from AI Apps to AI Infrastructure: An Inference Layer Playbook for Korean Founders

- RadixArk, commercializing SGLang, raised $100 million in seed funding, and Gimlet Labs, which builds a multi-silicon compiler, secured $80 million in Series A funding in the same week. - The trend is that more money is flowing to the foundational layer serving pre-trained models more cheaply than flashy apps. - Korean teams have little reason to compete in chip manufacturing. - Instead, there is a good chance for operational services that integrate different clouds and inference engines while explaining costs.