Semiconductors sit at the center of the AI transformation, driving a critical shift from training-focused infrastructure to high-performance inference at scale. SambaNova, led by CEO Rodrigo Liang, addresses this demand by deploying efficient, air-cooled hardware that enables trillion-parameter models to run in single-rack configurations, significantly reducing power and space requirements compared to traditional GPU clusters. This approach supports the growing need for low-latency, agentic AI workflows in both metropolitan and edge environments. As enterprises increasingly prioritize sovereign, private AI models to protect intellectual property and sensitive data, the industry is experiencing a notable repatriation of infrastructure to on-premise solutions. By focusing on premium inference performance and capital efficiency, organizations can move beyond simple cost-saving measures to create differentiated, high-value AI services that sustain long-term competitive advantages in a rapidly evolving global market.
Sign in to continue reading, translating and more.
Open full episode in Podwise
