AI workloads are shifting toward platforms that can handle both high-performance training and low-latency inference

In this session, Harshdeep Banwait, Director of Product at CoreWeave, looks at trends in GPU-accelerated infrastructure, mixed-use deployments, and the strategies shaping next-generation AI platforms, including:

  • The shift toward infrastructure that supports both training and inference workloads
  • Methods to improve inference performance through right-sizing, quantisation, and optimisation techniques
  • How infrastructure strategies are adapting to increasingly specialised AI hardware
  • Market signals guiding future AI infrastructure planning