The AI Opportunity: Silicon to Systems | Intel Business

Thank you.

Yeah, good question. First of all, thank you for inviting me back. You're welcome. And I assume I'm doing a good job so you invite me back. Of course. So I think couple of things. One, some of them you can apply from Cadence to Intel. First of all is I always want to hear the truth and so that you'll know where we are and then where we were and also how to improve. So I usually spend a lot of time with the employees and then you pretty much can decide which are the one tell me the truth. Yeah. And which I want to try to you know sugar coating it. Yeah. So I always focus on the people that telling the truth and I like bad news. Tell me all the bad news. And so I think that's very important to hear likewise to hear from the customer and so they tell you very candidly where you fail where you at the area to improve that's very valuable and then secondly is very important, you know, I'm a basketball player in the college so I always like to have a team that I can count on to build a great team that can win it take times so I think it start with the culture and how to hold the people accountable in whatever position they are in. So if you screw up, they take full responsibility. The one that I do really don't like is starting to point finger to others. I like to have people own the problem and then we can work on that. And then the other part is very important to have that transparency that you know don't talk to me nicely and then behind me tell me something bad. Usually those people I distance myself. So I think all in all I think it's very important to hear the bad news really focus on it and then really drives customer success. I call it delight customer. So go extra mile with the customer and listen very carefully what they tell you and then really make the changes and then I think compared to Intel a little bit more complex a much bigger team and when I took over is over 100,000 people and then I had to bring it down to 78,000 and so drive efficiency drive accountability, level down instead of the 10 12 layer you bring down to four to five layer so that you really can really get a good feel from the team and then the one part is difficult is the one that the manufacturing and that's critical for Intel on the foundry and the most advanced packaging like EMF that we can drive the leadership there. So I think a little bit more complex, more capex, more people and more operations and rather than just pure software.

Yeah, in my personal view, you know, I went through the whole multiple cycle. I remember the silicon days competitive carrier and then same thing with the internet. It's all great and I think you have to embrace them. I think the AI is much bigger than the internet impacts and so some correction is kind of natural and I don't overreact to it but something that you have to pay attention we live in the Silicon Valley is a bubble and you know we all get so excited about AI when you go to different part of the country and you found that there's a very strong anti-AI sentiment so I think it's very important for us as a leader and the driver and then of AI is to really have more positive narrative. Everything new is always have the negative side but we have to really explain the negative side you know like security is very important make sure the data not compromise privacy not compromise and meanwhile embrace the new technology. So just give you an example at Intel I completely embraced the AI. We have a AI transformation using the Gemini, Anthropic, Perplexity and really change the way we do business in term of design chip manufacturing chip and then really drive productivity and more efficiency on that and then we can create more value going forward.

Yeah, being a venture capitalist as you know even though I run Cadence I run Intel I still continue my investment. Yeah. Because it's very important to really plug into the startup community and also the venture community and so that you won't miss out some of the major wave and that's what I convinced my board at Cadence and convince my board at Intel that I need to be plugged into the startup community I was early investor in Habana Labs. Yes, same as early investor in Anapurna Labs. So both company and you know, Anapurna Labs we acquire by you know my good friend Amazon and then they embrace a team and then all the founders still there still running it and so on contrast when the Habana Labs I sold to Intel and then the Intel didn't keep the team the founders left and he started a company called Element Labs doing very well and So again it's just mistake that you have to empower the founders the entrepreneur and make sure that they're successful and they find it as a home and then you know to really drive success together rather than well I know it all I don't need you and then they one by one living and that you lose the talent. So I think something we learned from that and in term of your second question about AI plan and clearly the CPU right now in high demand. So the in reinforced learning agentic AI I think the in term of control plane orchestration CPO general purpose actually is very useful and so in fact I'm challenging quarter by quarter they will only provide some percentage of customer one many of CEO friend of mine call me up I need more and so I think that's a good news but meanwhile we also quietly building our own AI capability and strategy is very important. This is fast growing. If you look at semiconductor growth to one trillion and going to be two trillion coming up and so it's very important to really focus on the next AI movement. So we are delighted you and you know 10 years ago I decided you know GPU is great but power hungry we all know and so I decided to back SambaNova and I decided to back Cerebras and one is doing the data flow architecture one is doing the wafer scaling and then basically able to provide the performance at one of the power so I think those are the thing that you should really look at and how to drive that and then SambaNova right now we are starting to take off now and then the main reason is clearly you can drive the performance towards better and also able to work together with Nvidia is kind of more disaggregated inference and a multiple customer want to have that alternative. So I think all in all I think that is great and in fact the future even better and if you look our brain and how much token we are using but the power is much much lower so almost like 10,000 lower and so with that I think there's some new initiative that I'm starting to look at in term of the brain chip and then the neuromorphics and then NPU and so there's a lot of new thing that we can work on and also CPU memory how do you drive and you know drive performance and more efficient use of memory those I think is a great opportunity going forward.