Mitesh Agrawal, Positron AI | Robotics & AI Infrastructure Leaders

In this episode of theCUBE, host Dave Vellante interviews Mitesh Agrawal, CEO of Positron AI, who discusses the future of AI inference in robotics infrastructure. During their conversation, Agrawal outlines Positron AI's innovative custom silicon design aimed at enhancing performance-per-watt efficiency, enabling scalable and cost-effective inference solutions. He explains how Positron AI offers compelling alternatives to market giants like Nvidia, particularly in edge and token-processing environments, emphasizing the importance of application-specific optimization as demand for AI inference continues to surge. The discussion also delves into the competitive landscape of AI infrastructure and the strategic advantages of flexible, efficient inference.

Key Topics Covered:

• Overview of AI inference and its growing significance in robotics.
• Positron AI’s approach to performance-per-watt efficiency in chip design.
• Comparison of Positron AI's offerings with Nvidia's existing solutions.
• The shift in AI workloads toward inference and its implications for infrastructure.
• Unique aspects of Positron's ASIC design and memory architecture.
• Insights into product-market fit and scaling strategies in a young company.
• Challenges and opportunities in the AI and robotics landscape.
• The importance of energy efficiency amid rising demand for AI applications.