Baseten CEO Tuhin Srivastava on Custom Models, and Building the Inference Cloud

AI inference serves as the critical frontier for scaling artificial intelligence, as businesses increasingly transition from generic APIs to custom, in-house models. Tuhin Srivastava, CEO of Baseten, highlights that the application layer remains vital because companies derive competitive moats from proprietary user signals and specialized workflows that frontier labs cannot replicate. Despite a persistent, multi-year supply crunch for high-end compute, the market is shifting toward a multi-chip, multi-cloud future where operational reliability and software-defined infrastructure are paramount. The integration of inference and post-training creates a self-reinforcing loop, where lower costs and higher performance enable more complex agentic workflows. As intelligence becomes a commodity, the ability to secure capacity and execute at scale defines the dominant players, forcing enterprises to prioritize operational maturity and first-principles engineering to maintain their competitive edge.