“And so, as we -- let's call it fast forward a year, I would say we would have a fairly balanced portfolio between training and inference.”

In context

“estion. Joe Moore: I wonder if you could talk about the $5 billion of AI revenue, how that breaks down between training and inference, if you're able to assess that and just I know you sort of started off with most of the traction in inference, but you've seen some traction on the training side. Can you just talk about what that split looks like going forward as well? Dr. Lisa Su: Sure, Joe. So certainly, from the $5 billion that we're talking about, the early traction has been primarily with inference just given the strength of the product portfolio. MI300 is like very, very well optimized for inference given the memory capacity and memory bandwidth capabilities. But we have had some training adoption, and we expect that, that will continue to grow as we go through the next few quarters. And so, as we -- let's call it fast forward a year, I would say we would have a fairly balanced portfolio between training and inference. Operator: And the next question comes from the line of Vivek Arya with Bank of America Securities. Please proceed with your question. Vivek Arya: I had two. So, Lisa, for the first one, how do you address this investor argument that MI is off to a great start, but spec-wise, remains kind of one year behind the industry leader, right? You're shipping something comparable to Hopper while they are starting to ship Blackwell next year when you are at MI350, they will be on Blackwell Ultra or Rubin. So how do you see AMD closing that gap? And can you really gain share until that gap is closed? Dr. Lisa Su: Yes. Vivek, I actually don't see that. So maybe let me state it in another way. I think MI300, when we launched it was behind H100, H100 was in the market for a longer time. And we have with”