As enterprises rapidly scale AI from experimentation to production, a new challenge is emerging: the cost of AI is no longer simply an infrastructure or procurement issue - it’s becoming an architectural and engineering problem. AI workloads introduce highly variable consumption patterns across tokens, models, compute, inference, storage, and data movement, making traditional approaches to application architecture and technology cost management increasingly difficult to apply.
This is changing how AI systems must be designed. Decisions around model selection and routing, workload placement, caching, context and token usage, GPU utilization, inference optimization, and the use of smaller or specialized models can have a significant impact on both technical performance and operating economics. As AI adoption expands across the enterprise, organizations must architect systems that balance performance, reliability, scalability, and user experience with the growing requirement to control consumption and demonstrate measurable business value.
Topics of discussion will include, but not be limited to:
Jay Dhar, Director, Emerging Capabilities & AI Enablement, U.S. BANK
Shrikant Sonparote, Lead Engineer, LOWE’S COMPANIES
Navita Singh, Principal Data Scientist (Research Engineering), AUTODESK
Moderator: Gaurav Basra, Chief Executive Officer, BASRA CONSULTING SERVICES
