00:00 Speaker A
According to reports from AMD, its stock price has skyrocketed. What did you think of the print?
00:03 Speaker B
I mean, I was writing something a few months ago, right? By some, I mean two.
00:10 Speaker B
Basically, we talked about CPUs being back in fashion, right? And lo and behold.
00:14 Speaker A
And there were skeptics.
00:15 Speaker B
there were. they were. Yeah. they were.
00:17 Speaker A
And you know what I say to that?
00:18 Speaker B
Hmm.
00:19 Speaker B
there is nothing. I don’t know.
00:21 Speaker A
Look at the chart. To that you say:
00:22 Speaker B
Read, read, read the results. Please read and cry.
00:24 Speaker A
Please read and see the chart.
00:25 Speaker B
Um, yeah, they actually, they totally blew it. Well, guess what number Lisa Sue was talking about? The 70% growth in the second quarter is what she’s expecting from the CPU server side business, which is 70% year-over-year growth. They also said that they are seeing growth every year and are expected to grow by 18% over the next 3-5 years. That has now jumped to 35%. So this is just part of the overall effort into agent AI, where these systems perform tasks that CPUs are good at. Therefore, the importance of the CPU will inevitably increase. Will this last forever? That’s a great question, you didn’t just ask me, but I’d like to say, I thought you did.
01:07 Speaker B
We don’t know because we’re continually seeing different combinations of AI traders and AI infrastructure. So the CPU was left by the wayside and now it’s hot again. GPU, oh no one needs it anymore. Because we’re going to do more training, rather more inference. Well, wait a minute. GPUs are great for training, but what about Google’s CPUs or ASICs like Amazon’s Trainium chips? Therefore, it is constantly changing and evolving. So I don’t know how long this will last. They say where is the demand for this, at least until 2027? And I think this also goes back to what I was saying. We don’t necessarily know how fast these data centers are ramping up and what that means for the semiconductor companies themselves. In short, III believes that CPUs will remain a popular product, at least for now.
01:54 Speaker A
How competitive do you think AMD’s AI roadmap is with Nvidia’s right now? Of course, Nvidia’s earnings will be released on May 20th.
02:02 Speaker B
I think Nvidia is still the 800 pound gorilla in the room. So they’re four trillion, five trillion, five trillion, now that I think about it, there’s a billion dollar, billion dollar gorilla in the room. I think they will become even more competitive with the upcoming release of rack servers. It’s called Helios. That’s coming later this year, and basically Nvidia is talking about rack servers and NVL 72, right? That’s their big villain on stage. Now it’s about 1 to 2 tons, I forget what Jensen said. Alas, AMD didn’t have such a feature, but now Helios is going to have such a feature. It will have 72 GPUs. It will basically be functionally the same as Nvidia’s. Doing so will make them even more competitive in this situation. What does that mean for Intel? Well, they’re still a CPU giant, but they don’t have the GPU side. So we’re lucky to have AMD on both ends of this, but we also have Arm. These Arm chips are here, and more are starting to appear. That’s Nvidia’s Arm CPU. Oh, and Qualcomm is moving further into data centers. That is Arm CPU. So a lot of competition is starting to break out not only between GPUs but also CPUs.
