Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Applications, Applied AI

Why It Matters

This underscores a critical inflection: as AI moves from models to productized inference, specialized infrastructure vendors like Base10 can capture significant value by reducing latency, increasing reliability, and enabling multicloud deployment—challenging general cloud incumbents and shaping AI economics.

Summary

Tuhin, co-founder and CEO of Base10, recounted his path from finance and early ML research to launching Base10 in 2019 to build production inference infrastructure. Base10 now powers latency- and reliability-sensitive AI apps—examples include WhisperFlow (speech-to-text keyboard) and Abridge (ambient clinical scribe)—by delivering multicloud, optimized inference and a developer platform. He argued inference demand is about to surge and that while most token spend today goes to frontier models, product differentiation and profitable AI businesses will depend on custom or post-trained models running efficiently. Base10’s thesis is that specialized inference infrastructure will be essential for scaling real-world AI applications.

Original Description

For more information about Stanford’s graduate programs, visit: https://online.stanford.edu/graduate-education

This seminar covers applications, applied AI, and agent monetization.

Follow along with the course schedule: https://mse435.stanford.edu/

Guest Speaker: Tuhin Srivastava, Founder / CEO, Baseten

Instructor: Apoorv Agrawal is a Stanford Adjunct Lecturer in Management Science and Engineering and a partner at Altimeter Capital.