Bloomberg Intelligence Head of Technology Research Mandeep Singh is joined by Nicole Hu, a Silicon Valley technology veteran and GLG expert, to explore the implications of Google’s TurboQuant paper and the evolving economics of AI infrastructure. As hyperscalers look to improve the efficiency of AI workloads, advances in quantization are redefining the tradeoffs between memory and compute, with far-reaching implications for cost, latency, and datacenter architecture. They examine how new approaches to model optimization and inference could reshape hardware requirements, deployment strategies, and the next wave of AI investment.