AMD is betting big on AI internally in its goal of increasing chip production, CEO Dr. Lisa Su told an audience at SXSW on Monday. That includes her. Because she uses it to manage her calendar and send emails.
In a conversation with podcaster Ryan Patel, Hsu outlined the different ways AI is being used and said it will become virtually ubiquitous within a few years. Here are the most important takeaways from her conversation:
“The best thing to keep in mind is that I think AI is the most important technology to emerge in at least the last 50 years,” Hsu said.
You can watch the conversation here.
AMD is betting big on AI internally
AMD ships AI hardware in accelerators and PC processors, but it also uses AI hardware to accelerate its own business.
“So at AMD, we want to be at the forefront of leveraging AI in every aspect of our business,” Su said. “We use it to design chips, and we also use it to design faster chips, improve reliability, build better software… It’s a productivity tool. I Our goal is to tell our engineering teams that we can use AI to increase the number of products we can launch each year.”
Su said AI is also being used in human resources processes, financial processes and customer service.
It’s not perfect, Su acknowledged, and one way AMD is thinking about “responsible AI” is to train the AI the right way. “The AI thing we have to be careful about is how we train our models while making sure that the data we’re using isn’t biased in any way,” Su says.
“But what I want to say is, for those who are a little worried about what AI will do, the answer is no, go slower,” Hsu added. “Okay, that’s definitely not the answer.”
AMD’s GPU supply will increase
Su, representative of AMD’s AWS cloud services, said in response to questions from the audience that AMD’s GPU supply will increase. Sue laughed and promised to help.
YouTube
“In the chip industry, we’re really going through cycles where demand outstrips supply,” Su said. “One thing we can say for sure is that we’re very good at increasing supply. So we’re seeing significant increases in GPU supply. Our job is to build capacity there.”
Mr. Su also uses AI
“I definitely use it. [Microsoft] Copilot is one of the things I use to do really simple things like summarize meetings or track action items,” Su added.
What doesn’t she do? Use it to write emails. “That doesn’t work very well,” she said.
Who will win with AI? Local AI or the cloud?
AI consists of two aspects: machine learning (or training) and testing (also known as inference). The latter is what we think of when we talk about LLM, AI chatbots, and AI art. However, it is faster to run in the cloud. It may always be faster to run in the cloud, but it will make AI more pervasive to clients like AI PCs.
“When you think about inference, inference is basically asking questions and getting answers, and almost all of that is happening in the cloud today,” Hsu said. “And frankly, that’s part of the delay I talked about. It takes a few seconds to respond.
“I think a huge percentage of it is going to happen at the edge, whether it’s at the client or closer to the humans. It’s just that you’re going to be looking at AI and all these different places,” Sue said. he added. “Over the next year or two, we think inference will surpass training. So that’s where we see the AI trends heading.”
What did AMD need? Focus
Mr. Hsu is about to celebrate his 10th anniversary as CEO in October. Looking back on her previous career, she said that what AMD needed was focus.
“When I first became CEO, we probably did too much, like any company that doesn’t have enough resources,” Su said. “And it was very important for us to decide what we would be good at. And what we would be good at is building the highest performing computing capabilities. So , we’re a high-performance processing company. That’s what we spend our time doing. And it turned out not to be a bad bet.”
AMD
When asked to describe herself in one word, Sue said she is a “doer.” “And our goal, my goal, is to set very ambitious goals and help the team achieve them,” Sue said. “That’s why I call myself a doer. I love going to the lab. I love seeing the latest and greatest happenings, and I love spending time with our customers and partners. I love it.”
Memorable career memories
Having worked as both an engineer (he received his undergraduate, master’s and PhD degrees from MIT) and a manager, Sue said he has learned lessons from both.
“If you’re an engineer, you can say that with certainty. [the lesson learned was from] This is the first product I worked on,” said Hsu. “I was working at IBM at the time, and I can tell you that we announced a product that day and nothing worked. Like the chip wouldn’t turn on. It’s a scary time when that happens.”
Hsu said engineers are “very innovative and creative, and where there’s a will there’s a way.”
As for business lessons, it was “the first business I ever ran,” Sue said.
“I felt like I went from being an engineer to being a business person,” Sue said. “No offense, but I thought it would be relatively easy. I didn’t expect there to be such a big transition between engineering and business. I mean, in the first business I ran, literally the first We missed about 50% of our annual operating plan. I was devastated.
“My boss told me, Lisa, That’s terrible. you are terrible how can i do that? And I said, ‘You know what?’ He was right,” Sue added. “I could have done better, so I learned from that and learned how to ask more questions next time.”
What is Lisa Sue’s workout routine?
It was a strange personal question, but Sue answered it politely. “I have a great guy who comes to my house every day or every few days, and it’s boxing and strength training,” she said.
When asked whether she preferred F1 racing or golf, Sue said she preferred the latter.
