Not All NVIDIA AI Factories Are the Same
Transcript
All NVIDIA AI factories are the same, right? Depending on the OEM, that's all the same GPU, networking, and in some cases, even CPU. Let's get into it. >> [music] >> HPE and NVIDIA invited me to HPE Discover 2026. They paid my way. They're sponsoring this content, but they don't get to see it until you see it. So, this is my honest take on their approach. com, my research site, I compare all of the vendors. Kind of where where are the boundaries?
And I got to see the boundaries for the HPE NVIDIA relationship firsthand at Discover. Where are the two companies coming together and integrating? And where is HPE taking it to the next level? Let's talk about HPE first. HPE's Unleash AI program, I think, is quite frankly, unparalleled on their from a partner perspective. You can basically get everything from the application down to data movement and AI platforms through various partners, pay for via your GreenLake relationship. One hand to shake, which is unique within the AI ecosystem.
Then, their actual platform, their private cloud AI platform, is the technologies that we're going to go into over the next couple of videos, and this is where the handshake between HPE and NVIDIA happens. We'll address three specific areas where I think HPE and NVIDIA are tackling the most challenging problem in enterprise IT of my career, which is a genetic AI. From CPU and GPU, how is Vera Rubin and HPE's approach to Vera Rubin helping you to better uh deploy technologies that take advantage of your intelligent data infrastructure?
Second level level is that intelligent data infrastructure. What's the secret sauce in HPE's solution that's allowing uh KB cache to be uh effectively managed and you get 20x the performance from typical storage solutions? Differentiating, this is a real cost-savings uh mechanism for those of us who have multi- multi-tenant or uh or high-frequency or high-utilization uh inference and agentic AI sessions. And then thirdly, networking. I I've dismissed networking up until HPE Discover for inference. In my mind, the networking solutions for AI were centered around training specifically.
And yes, every percentage of utilization that you can get out of your network performance and efficiency translated into higher efficiency for training. But inference, you know, that that math was a little bit uh different. Agentic, again, changes that calculation for me and what I came out of HPE Discover understanding. You want to learn more about the research? com. You want to learn more about HPE's relationship? There's a link below in the video description. com. Talk to you next CTOAdvisor CTO.
