So, You Want to Build an (AI) Supercomputer? Lessons from Global HPC and AI Infrastructure Projects

This talk explores what it really takes to procure, operate and justify high performance computing (HPC) and AI infrastructure successfully. Drawing on lessons from hundreds of HPC and AI procurements worldwide, it examines where organisations go wrong, why projects fail to deliver expected outcomes, and how these pitfalls can be avoided. The session will explore approaches to procurement strategy, highlighting the importance of defining requirements, engaging stakeholders and structuring procurements to achieve long-term success rather than short-term compliance. It will examine different approaches to cost modelling, moving beyond capital expenditure to consider Total Cost of Ownership, Total Cost of Operation and Total Cost of Solution, helping organisations understand the true value of their investment. Finally this presentation will discuss how organisations can balance risk and value, develop robust business cases and secure executive or senior support for major HPC investments. Attendees will leave with practical insights, real-world examples and proven frameworks for making better HPC investment decisions and delivering greater value from their computing infrastructure.