NVDA is the biggest winner in the Meta AI supercluster, since the GPUs, NVLink networking and ConnectX network interface cards are all Nvidia's domain; co-designing with Meta should let Nvidia apply what it learns to other customers too, spreading the benefit beyond this one buildout.

Meta and Nvidia just teamed up to build one of the craziest AI supercluster out right now. The first phase is complete with hundreds of thousands of Nvidia GPUs.

Distributed machines is what Jensen has actually been talking about for a long time. and these are the types of networking solutions that Meta is working with that Nvidia is working with and I'm glad to see some in action right now

Obviously, the big winner in something like this is Nvidia, right? The GPU domain is Nvidia Envy Link. Networker network interface cars typically are Nvidia Connect Xnix,

We also see the increasements of rack power. The GB300 is listed at roughly 135 kilowatts. The upcoming platform that is going to be using their San Miguel platform and it's going to be the Vera Rubin is listed above 170 kilowatts per rack and will use fully liquid cool compute trays.

Now for Nvidia, I think obviously the big win here is they're co-designing with Meta and and trying to learn from this and whatever they learn from this, they'll be able to push out into other customers as well.