Vipul Prakash, Together AI | theCUBE + NYSE Wired: AI Factories - Data Centers of the Future

In this episode of theCUBE + NYSE Wired: AI Factories – Data Centers of the Future, John Furrier interviews Vipul Prakash, co-founder and CEO of Together AI, discussing the rapid evolution of AI factories and their significance in modern enterprise infrastructure. Prakash highlights the astonishing growth of AI-native applications, emphasizing a shift in computing demands that necessitate a re-evaluation of compute, storage, and networking systems. He shares insights on how companies are segmenting traffic across closed APIs and open-source models, as well as the importance of data adjacency for optimizing performance. The conversation also delves into the organizational structures that support AI deployment, including the role of a chief AI officer, and Together AI's approach to overcoming real-world constraints, particularly related to power requirements for new AI factories being established in locations like Maryland and Memphis. Prakash outlines the technical advancements in AI architecture and the operationalization of AI-native applications, targeting efficiency in both training and inference stages.

• Growth of AI-native applications and the acceleration of computing needs.
• Importance of data adjacency and parallel storage systems for AI performance.
• Challenges in AI infrastructure, particularly regarding power and speed of deployment.
• The role of chief AI officers in enhancing organizational efficiency and technology adoption.
• AI-scale architecture trends and evolving transformer models.
• Practical examples of enterprise use cases, including AI applications in regulated industries.
• Continuous testing and evaluation methods for generative AI systems.
• Together AI's plans for integrating training and inference processes on shared infrastructure.