Yahoo Finance Video
One of the biggest concerns AI players like Nvidia (NVDA) have to contend with is increased energy use. Dion Harris, Nvidia's head of data center product marketing, joins Catalysts to discuss the heightened energy demand and how the energy efficiency of Nvidia's new Blackwell chip. "Blackwell was purpose-built from the ground up to really make sure it could deliver the most performance and efficiency," Harris tells Yahoo Finance. He notes that the Blackwell processor is coupled with the Grace CPU to offer two times more energy efficiency than the standard x86 CPU. He adds, "And then of course, we also looked at how do they communicate to each other. So we have a proprietary communication protocol called NVLink, which allows GPUs to connect to each other at very high speeds, and that eliminates a lot of the bottlenecks and therefore drives more efficiency." Thus, Harris explains that the whole system is designed to deliver "incredible performance" at 25 times more energy efficiency than Nvidia's previous generation. When it comes to energy, he believes there are two parts of the equation: short-term and long-term. In the short-term, many of Nvidia's customers do not have the means to quickly build efficient data centers. Thus, Blackwell chips provide "a huge uplift in terms of the productivity they can get out of their existing power assets." In the long-term, Nvidia is working closely with partners and energy providers to ensure that data centers are being built as efficiently as possible and that grid infrastructure is resilient. "AI is going to be critical to enabling sort of the next wave of infrastructure. We talk about the grids as one sort of key use case, but it will also play out in manufacturing and aeronautics and a number of heavy industries," Harris notes. For more expert insight and the latest market action, click here to watch this full episode of Catalysts. This post was written by Melanie Riehl
Yahoo Lifestyle 
