英特尔 CEO 陈立武谈文化与工程转型

欢迎回到舞台，Cadence 的总裁兼首席执行官 Anerude Devgan 博士。大家好，希望你们今天过得愉快。我们还有一场激动人心的环节。我很荣幸欢迎另一位无需介绍的人——Intel 的首席执行官 Lip-Bu Tan。Lip-Bu，欢迎回家。

Welcome back to the stage, Cadence's president and chief executive officer, Dr. Anerude Devgan. Hey everyone, hope you're having a good day. We have another exciting session. It's my pleasure to welcome another person that needs no introduction. CEO of Intel, Lip-Bu Tan. Lip-Bu, welcome home.

是的，很高兴回家。

Yeah, nice to be home.

哦，当我听说 Lip-Bu 成为 Intel 的 CEO 时，我非常高兴，因为 Intel 是如此标志性的国际品牌，是国家的瑰宝，而没有人比 Lip-Bu 更适合来转型这家公司。所以恭喜你。

Oh, I was so happy when I heard that Lip-Bu is becoming CEO of Intel, because Intel is such an iconic international brand, a national treasure, and there's no better person than Lip-Bu to transform that company. So congratulations.

是的。我以为我从 Cadence 退休后，可以享受和孙子孙女在一起的时光，结果最后一刻我决定接受这个职位。这太棒了。

Yes. I thought that I retired from Cadence, I could just enjoy time with my grandchildren, and then last minute I was decided to do this. It's fabulous.

对整个行业来说太棒了。Lip-Bu，当然我们合作了这么久，这是我职业生涯中的荣幸。我们总是谈论，我知道对你来说非常核心的是文化，对吧？你多年来领导的 Cadence 文化及其转型。当然，你与初创公司合作的历史悠久，文化非常重要。那么现在你加入 Intel，这家公司正在经历一些动荡，你对 Intel 文化的评估是什么？需要做些什么？

Fabulous for the entire industry. And Lip-Bu, of course we work together for so long. This is the pleasure of my professional career. And we always talk about, I know it's very central to you, is the culture, right? The culture and the transformation of Cadence culture that you led over the years. Of course, you have this long history working with startups, and culture is so important. So now as you join Intel, this company going through some turmoil, what's your assessment of Intel culture and what needs to be done there?

是的。首先，我今天非常非常高兴来到这里。Cadence Live 是我最喜欢的活动之一，我一直很期待。我很兴奋来到这里，一走进来就看到这么多熟悉的面孔。我本想给他们一个拥抱，但没时间了。我们有时间，Lip-Bu，我们可以做。是的。你知道，在 Intel，我清楚地为公司设定了重点。然后，就像大约 15 年前我在 Cadence 所做的那样，真正打造最好的产品，然后回归到对产品的关注，让客户满意，不仅仅是满足客户要求，而是让客户满意，为客户多走一步，更重要的是，与客户建立信任。如果你还记得，我们有很多客户会议。关键是真的建立信任，让客户相信我们能交付他们期望的东西。所以我认为这极其重要。文化，同时增强两者兼顾的能力。

Yeah. First of all, I'm very, very happy to be here today. Cadence Live is one of my favorite events, I always look forward to it. I'm excited to be here, see so many familiar faces when I just walked in. I thought I would give them a hug, but there's no time. We have time, Lip-Bu, we can do it. Yes. You know, at Intel, I think clearly I laid out the focus for the company. And then, like almost 15 years ago when I was doing the same at Cadence, really create the best products and then back to that focus on the product and delight the customer, not just meet the customer requirement, but delight the customer, go that extra mile for the customer, and even more important, build trust with the customer. If you recall, we had many customer meetings. The key thing is really building the trust that the customer can trust us to deliver what they expect us to do. So that, I think, is extremely important. Culture and also enhance the ability to do both.

是的。我的意思是，客户至上总是至关重要的，对吧？我记得，我实际上要告诉你一件非常独特的事情。你几周前刚加入，对吧？Intel。是的。第一天，同一天是 GTC，对吧？是的。所以首先，Lip-Bu 在 GTC，看到你认识 Jensen 真是太棒了，我知道你们有很好的合作关系。Jensen 今天早上在这里，对吧？是的。然后我遇到一个非常重要的客户，可能是 Intel 最大的客户之一。我不在这里提名字。是的。这是 Lip-Bu 的第一天，我和这位非常有名的人交谈，他说：“是的，是的，我和 Lip-Bu 共进晚餐。”所以我知道在 Lip-Bu 作为 Intel CEO 的第一天，你就已经见到了你最重要的客户之一。所以很高兴看到 Lip-Bu 和我认识的那个人一样。这太棒了。但另一件事，当然客户至上至关重要，而且 Lip-Bu 是篮球队的坚定信徒，你知道，一个团队，身边要有合适的团队。但另一个关键点是工程文化，对吧？研发创新，紧迫感。而且有一些媒体报道说，当然你在 Intel 董事会待了一段时间，你不满意，或者 Intel 的工程文化需要加速。那么你如何看待这一点，让公司变得更以工程为中心？因为我知道这对你很重要。我们总是努力做研发，与客户进行研发合作。那么你如何看待 Intel 的情况？

Yeah. I mean, customer first is always critical, right? I remember, I actually have to tell you something which is quite unique. So you just joined a few weeks ago, right? Intel. Yes. And the first day, the same day is GTC, right? Yes. So first of all, Lip-Bu was at GTC, which is amazing to see you know Jensen, and I know you have a great partnership. Jensen was here this morning, right? Yes. And then I meet some very important customer, probably one of Intel's biggest customer. I won't mention the name here. Yes. And this is Lip-Bu's first day, and I was talking to this very famous person also, and he says, "Yeah, yeah, I'm having dinner with Lip-Bu." So I know on Lip-Bu's very first day as Intel CEO, you already met one of your most important customers. So it's good to see that Lip-Bu is no different from the person I know. So this is great. But the other thing, you know, of course customer first is critical, and also Lip-Bu is a big believer of basketball team, you know, one team to have the right team around you. But the other key thing is, of course, engineering culture, right? R&D innovation, urgency. And there's some reports in the press that, of course, you were on Intel board for some time, that you were not happy, or the Intel engineering culture needs to be turbocharged. So how are you looking at that to become like a more engineering-focused company? Because I know that's important to you. We always try to do R&D, R&D collaboration with customers. So how do you see that with Intel?

是的。所以我认为我做的第一件事就是回归我的重点：推出最好的产品。所以我决定让所有工程领导直接向我汇报，这样我就能真正深入钻研。我总是深入钻研，正如你记得的，这就是我了解需要改进之处的方式。同时，我现在刚过了六周。是的。所以我挺过了这六周。然后很明显，我真正专注于倾听客户，同时关注工程以及我们的不足之处。所以我问客户：“你给我的产品打几分？”结果我得到了一些 C、一些 F，就像我刚开始在 Cadence 时一样，我铭记在心。另一部分是，这家公司是如此标志性的公司，有着如此悠久的历史。多年来，它建立了许多汇报层级。所以除非你往下走六到七层，否则你得不到真实的数据。而一些真实的数据，你知道，在你向董事会汇报后——我在董事会待了两年——你开始看到一切都是绿色的，但现实并不那么好。所以我认为这是一种信息失真。你真的需要加倍、三倍努力去理解问题所在，这样你才能理解客户的挫折和反馈，然后我们才能纠正。所以我认为这对我来说非常非常重要：倾听客户的声音，然后你开始意识到你必须走多远。我是一个篮球迷。所以如果你真的想在篮球中打得好，除非你和勒布朗·詹姆斯一对一，否则你不知道你必须提升多少才能打好比赛。所以如果你没有那种体验，你总是对自己拥有的感到满足，但那不是一支冠军球队。所以对我来说，另一个非常重要的部分是拥有谦逊的文化，倾听客户。有时如果你太成功，你会变得非常骄傲，不再倾听。所以确保你倾听非常重要，同时你努力组建一个团队，随着时间的推移，这个团队可以像篮球队一样真正推动成功，你可以赢得冠军。这就是我的理念。然后慢慢地、慢慢地组建合适的团队，然后才能真正满足客户需求。

Yeah. So I think the first thing I did is basically back to my focus: get the best products out. So I decided to have all the engineering leaders report to me so that I can really do deep dives. I always do deep dives, as you recall, and that's how I learned where we need to improve. And also, at the same time, I just passed six weeks now. Yeah. So I survived the six weeks. And then clearly, I'm really focused on listening to the customer and also focusing on engineering and where we come short. So I asked the customer, "How do you grade my products?" So I got some C, some F, just like Cadence when I started, and I took it to heart. And then the other part is, the company is such an iconic company with such a long history. Over the years, it has built up a lot of layers of reporting. So unless you go down six or seven layers down, you don't get the real data. And some of the real data, you know, after you present to the board — I was on the board for two years — you start to see everything green, and the reality is not that good. So I think this kind of lost in translation. You just need to really double, triple down on understanding where the issue is, so that you can understand the customer frustration and feedback, then we can correct that. So I think that to me is very, very important: to hear from the customer, and then you start to realize how far you have to go. So I'm a big basketball fan. So if you really want to play well in basketball, unless you do one-on-one with LeBron James, then you know how much you have to step up to play the game. So if you don't have that, you're always very content with what you have, but it's not a championship team. So the other part very important for me is to have that culture of being humble, listen to the customer. Sometimes if you are too successful, you become so proud, you are not listening. So it's very important to make sure that you listen, and at the same time, you try to put together a team, and over time that team can really drive success as a basketball team, you can win championships. So that's kind of my philosophy. And then slowly, slowly putting the right team together, and then really can address the customer requirement.

我知道你已经，我在报道中看到，你已经扁平化了组织，对吧？是的，立即行动。所以那天下午我和一些人聊天，问他们：“我应该问 Lip-Bu 什么？”所以我收到一个请求，我转达给你，Lip-Bu，这与工程文化有关。另外，就像你提到的，如何——我相信你专注于重新招聘或招聘一些顶尖人才，对吧？这始终是你的一个重点。当你审视 Intel 的工程文化时，对吧？是的。在某种程度上，过去六周，我的大部分时间，首先，我打破了自己的记录。所以我一天有 22 个会议。哦，哇。还有两顿晚餐，有时三顿。

And I know you already, I saw in the reports, you already flattened the organization, right? Yes, immediately. So then I was in the afternoon talking to some people, asking like, "What should I ask Lip-Bu?" So one request I got, I passed it along, Lip-Bu, which is connected to the engineering culture. Also, like you mentioned, how to, I'm sure you're focused on rehiring or hiring some of the top talent, right? That's always a big focus of yours. As you look at the engineering culture of Intel, right? Yeah. In some way, the last six weeks, most of my time, first of all, I broke my own record. So I have 22 meetings a day. Oh wow. And two dinners, and sometimes three dinners.

我这么做的根本原因是，我向英特尔所有高管、前高管都问过：你们认为哪些关键人才——架构师、工程师——真正厉害？他们每个人都给了我一份需要我去接触的名单。我问了大约 20 个人，这 20 个人给了我各种不同的名单。于是我开始汇总所有人的意见，整理出一份综合名单。然后我计划逐一拜访他们，这就是为什么我安排了 22 场会面。是的。所以我认为，确保留住优秀人才并坦诚相待非常重要。这些年来我们流失了一些优秀人才，所以现在首要任务是保住现有团队。另一部分工作则是花大量时间从外部、非英特尔体系招募人才，这样你就能清楚知道哪些领域——架构或工程方面——我还没有合适的人选。也许现有的人选不错，但要么太年轻，要么经验不足，那我就需要引进一些真正有经验、来自不同文化背景的人，这样我才能推动我所说的“从零到量产”。是的。

And the main reason I'm doing that, clearly, you know, I asked all my top executives from Intel, past executives, what are the key talents, you know, architects, the engineers they are really good at. And then they all give me the list of people that I need to really reach out to them. So I asked about 20, and then 20 people they give me all the different lists. So I start to compile a list as a combine everybody. Then I start to program to meet every one of them. That's why I have 22 meetings. Yes. And so I think it's very important to make sure that you hold on to the good talent and to be candid with you. We lost some good talents over the years, and so right now it's basically try to keep what I have. And the other part is also spend a lot of time trying to recruit the talent from outside, the non-Intel, and then outside, so that you know clearly identify where are the area architects or where the area in the engineering that I don't have the right one. You know, maybe a good player but they are either kind of young or maybe less experienced, then I need to bring in somebody that are really proven and then from the different culture so that I can drive that I call it the zero to productions. Yes.

你知道，一次流片成功。我以前做风险投资，甚至在担任 Cadence CEO 期间，董事会也允许我继续投资。在英特尔也一样，我仍然想做这件事，因为这样你才能意识到正在发生的大浪潮——AI、云、移动，所有这些大浪潮。你必须真正与外部人才接触，这样才能吸引一些人才回流，帮助你开发下一代产品。但更重要的是，除了设计，你还必须流片。我认为，建立“一次流片成功”的文化纪律非常关键。是的。你不想流片失败，这就是为什么 palladium 验证平台非常重要。你要真正验证你的设计，确保在流片前没有 bug。就像创业文化一样，每次流片，尤其是先进制程，成本高达 5000 到 6000 万美元。所以你必须一次成功，对吧？我认为英特尔过去失去了这种文化。所以我必须把它带回来。当然，要带回这种文化，你就得引进一些有“一次流片成功”记录的新人才，来真正强化这一点。

You know, first time pass. You know, like I being a venture capital before, even during the time I was a CEO of Cadence, the board still allow me to do investment. And likewise in Intel, I still want to do that because then you realize what are the big wave is happening. You know, the AI, the cloud, the mobile, all this big wave. You got to be really involved with the talent outside, so you really need to have that so that you can really attract some of the talent to come back in, yes, and to help you to develop the next generation product. But more important, beside designing it, you got to have a tape out. And that's where I think it's very important to have the discipline of the culture that the first time pass. Yes. You know, you don't want to tape out the chip and that's why the palladium is very important. And then you know really verify what you design, make sure that there's no bugs before you go to tape out. Like startup culture, every tape out, especially the advanced node, is 50, 60 million. So with that, you have to really the first time pass, right? So that culture over the time I think Intel lost that. So I had to bring that culture back. And then of course, when you want to bring the culture back, you have to bring in some new talent that have that track record of first-time pass to kind of bring in to really strengthen that.

我的意思是，一方面是成本，就像你提到的流片成本，但如果迭代次数太多，损失的时间也是无价的，对吧？错过了窗口期。

I mean, one is the cost, of course, like you mentioned of the tape, but also if there is too many iteration, then the time lost is invaluable, right? Miss the window.

是的，错过了窗口期。对。那可能价值连城。

Yeah, it's missed the window. Yes. And that could be worth a lot.

那么关于 AI，这当然是个大话题。我们知道所有那些数字，比如预测：半导体本身将达到 1.2 万亿、2 万亿，超过 1 万亿。没错。系统市场，电子系统，到 2035 年将达到 5 万亿，半导体可能达到 2.5 万亿。好的。其中很大一部分是由 AI 驱动的，它正在扩展到不同的细分领域，不仅仅是云端，还有边缘端，比如笔记本电脑和汽车。当然，英特尔在所有这些都是关键参与者。那么你怎么看英特尔的 AI 战略？

So in terms of AI, okay, that's a big topic, of course. And we know all these numbers would be like the projections: semiconductor itself will reach 1.2, 2 trillion, more than one trillion. That's right. The systems market, electronic systems, will be like 5 trillion by 2035, semiconductor maybe 2.5 trillion. Okay. And a lot of it is driven by AI, which is stretching in different parts of the segment, not just on the cloud, on the edge, as you know, on laptops and automotive. And of course, Intel is a key player in all of these things. So what do you think about the AI strategy of Intel?

对。这当然是一个非常重要的话题。我知道你肯定在头六周花了很多时间在这上面。是的。是的。是的，我认为很明显，AI 正在推动整个半导体行业的增长。工作负载已经更多地转向 AI。所以我认为英特尔重新定义我们的 AI 战略非常重要。在我加入英特尔之前，我在 AI 领域投资了很多。是的。不仅是在芯片方面，还有我称之为软件 2.0 的方面，即 AI 驱动的新计划，以及一些应用。然后因为海量数据，如何利用 AI 来驱动更准确的数据、结果，并优化解决方案。所以我认为这部分，英特尔是缺失的。要称赞 Jensen，他做得非常出色。我以前认为 Jensen 和 Nvidia 是一家 GPU 公司，但现在他真正成为了 AI 的领导者。我认为这值得称赞。他还构建了 CUDA 软件，整个行业都离不开 CUDA。所以基于这种理解，审视我们已有的东西非常重要。我们有 CPU，我们在 PC 客户端有一些优势 CPU，我们仍然拥有 70%、68% 的市场份额。在数据中心方面，尽管我们正在失去一些机会，但我们仍然有 50% 的市场份额。生态系统就在那里，是的，但我们只需要思考如何拥抱 AI，使其成为更好的解决方案。所以某种程度上，我专注于如何拦截一些客户需求。然后在整个智能体式 AI、物理 AI 方面，我想你谈了很多。我们如何拥抱这些，然后使用系统方法和软件方法，我称之为软件 2.0，来定义芯片，并满足一些大型平台的需求。我们希望成为他们的计算平台，为他们的解决方案进行优化。

Right. It's such an important topic, of course. I know I'm sure you spent a lot of time on that in the first six weeks. Yeah. Yeah. Yeah, I think clearly, you know, the AI is driving all the semiconductor growth. The workload have changed a lot more into the AI. And then, so I think it's very important for Intel to redefine our AI strategy. And before I came over to Intel, I invest a lot in the AI. Yes. And not only in the silicon side, and also on the I call it the Software 2.0, on the AI driving the new initiative, and also some of the application. And then because of the massive data, and then how do you use AI to drive more accurate data and then result and then optimize for the solution. So I think that part, I think Intel is missing that. Give credit to Jensen, he done a fabulous job. And then I used to look at Jensen and Nvidia is a GPU company, and then now he really driving become the leader of the AI. And that I think give him credit. And he also built the CUDA software, and then the industry, nobody can go without the CUDA. So I think with that understanding, it's very important to look at what we have. We have the CPU, we have some of the benefit CPU for the PC client, we still have 70%, 68% of the market share. And in term of the data center, even though we're losing some opportunity, but we still have 50% market share. The ecosystem is there, yes, but we just need to how do we embrace the AI to make it better as a solution. So in a way, I kind of focus on how do we intercept some of the customer requirement. And then in the whole agentic AI, physical AI, I think some of them you talk quite a lot about it. And then how do we embrace that, and then using the system approach and the software approach, I call it the Software 2.0, define the silicon, and also meet up some of the big platform. We want to be their compute platform that we optimize to their solutions.

举个例子，早期我投资了一家叫 Annapurna Lab 的公司，现在当然是你们的大客户，后来它成为了亚马逊的一部分。那个团队，我支持的那个团队，他们现在还在亚马逊。很明显，他们真正推动了为亚马逊解决方案优化的专用芯片。所以现在有几个大型平台，他们都迫切需要一些专用芯片。我们能扮演这个角色吗？从系统层面定义它，同时考虑软件需求，然后芯片针对他们的应用进行特定优化。这些就是我们看到的机遇。另外，我认为可能利用我们的代工业务来真正推动一些效率提升。这些是我感兴趣的方向。你知道，我们真的需要为 AI 制定一个游戏计划，对吧？所以尽管才五六周，我已经开始组建团队，我正在招募团队来做这件事。所以请拭目以待。当然，Cadence 可以帮助我推动一些效率提升，比如光学解决方案，我知道解决方案能服务于这个目的。

So give you an idea, and then earlier days I invest in the company called Annapurna Lab, of course a big customer of you now, and then you become part of Amazon. And then the team, the same team that I backed, they are still at Amazon. And then clearly, you know, they really drive that whole purpose-built silicon optimized for the Amazon solution. So again, right now there's a couple of big platform, they are all dying to have some purpose-built silicon. Can we play that role? And then from the system level and define it, and then also looked at the software requirement, and then the silicon optimize specific optimize their application. So those are the opportunity I think we see that. And also I think potentially using our foundry to really drive some of the efficiency. I think those are kind of my interest. You know, we really had to have a game plan for the AI, right? So even though it's five, six week, I already starting, I putting a team, I'm recruiting the team to really do that. And so I think stay tuned. And then of course Cadence can help me to drive some of this efficiency, the optical solution, I know solution to serve the purpose.

我认为 Lip-Bu 你说的非常有趣，因为你已经有了 x86 这个特许经营业务，对吧？是的。而 AI，你知道，这跟英特尔以前做的略有不同。所以，你的意思是英特尔更加灵活，能够与顶级客户合作，因为他们正在自己做芯片。

I think Lip-Bu what you say is very interesting because you already have a franchise business with x86, right? Yes. And AI, you know, and this is, you know, slightly different than what Intel has done before. So, what you're saying is Intel is more, let's say, flexible to work with the top customers because they're doing some of their own chips.

但如果 AI 需要特定用途、特定领域的活动，那么英特尔会对此持开放态度，尤其是与大客户合作，因为这与过去相比是一个巨大的变化，对吧？

But if AI requires a purpose-built, domain-specific activity, so Intel will be open to especially with the big customers to do because this is a big change from the past, right?

是的。在某种程度上，这是利用我们的实力和市场地位来推进一些 AI 计划。我是 Habana 的早期投资者，然后把它卖给了英特尔。现在我们有了 Gaudi 芯片。我们如何利用和修改它，以便率先进入市场？这非常重要。你不想开发一款需要两三年才能推出的芯片。有一条更短期的路径至关重要。另一部分是我在 AI 计划中的一些投资。我们可能只是与他们合作、授权技术，甚至最终收购他们，使他们成为我们未来的解决方案。目标基本上是每年推出一款 AI 芯片，驱动不同的性能和不同的应用。这是我的重点：推动效率。在某种程度上，我试图创建一个独立的团队，基本上就像英特尔内部的初创公司，其文化与业务的其他部分隔离，这样他们可以更快地行动、更快地进入市场，并有更多纪律来推动效率。当你采用初创公司模式时，你行动得更快；没有那么多评审或那么多人来决定这是否正确。你必须下一些赌注，并确保它们是真正的好赌注，然后整体上处理我们的产品组合，并优化我们的解决方案以满足客户需求。非常令人兴奋。

Yeah. In some way, it's using our strength and market position to move into some of the AI initiative. I was an early investor in Habana, and then I sold it to Intel. Now we have the Gaudi chip. How can we utilize and modify it so that we are first to market? That's very important. You don't want to develop a chip that takes two or three years to come out. It's very important to have a shorter-term way to get there. Then the other part is some of the investments I make in AI initiatives. We may just partner with them, license the technology, and eventually even acquire them, so that they become our solution going forward. The goal is basically to have one AI chip coming up every year, driving different performance and different applications. That's my focus: drive efficiency. In some way, I try to create a separate team, basically like a startup within Intel, with a culture isolated from the rest of the business so they can move faster and get to market faster, with more discipline to drive efficiency. When you have a startup model, you move much faster; you don't have so many reviews or so many people deciding whether this is the right one. You have to take some bets and make sure they are really good bets, then holistically approach our portfolio and optimize our solutions to meet customer requirements. Super exciting.

对，很好。

Right. It's good.

另一个大话题是，当英特尔如此庞大且对许多领域至关重要时。我想和你讨论的另一个大话题是芯片设计、EDA、IP 和硬件。但在那之前，我相信每个人都想知道的大话题是：你对英特尔代工业务有什么看法？有哪些独特的机会？我知道你上周举办了一个重要的代工日。谢谢你邀请我。很高兴能去。那么你对英特尔代工业务有什么看法？

Another big topic when Intel is so big and critical to a lot of areas. The other big topic I want to talk to you about is chip design, EDA, IP, and hardware. But before we get there, the other big topic I'm sure everybody wants to know is: what are your thoughts on Intel Foundry? What are the unique opportunities? I know you had a big Foundry Day last week. Thank you for inviting me. Glad to be there. So what are your thoughts on Intel Foundry?

首先，我感谢你来到我的舞台，一起参加英特尔愿景代工活动。我知道你有个电话，但你设法过来支持我。我对此表示感谢。客户至上。没错。谢谢。显然，我认为我们有一个独特的机会。我们在英特尔代工中拥有唯一的美国国内先进技术工艺设计、研发和制造。我认为这对我们来说是一个很好的优势。此外，客户正在寻找灵活、有弹性、安全的代工厂，我们可以扮演这个角色。我很高兴看到特朗普政府真正支持美国技术和制造领导地位。所以我认为我们有支持来做到这一点。我的核心重点是确保我们的 18A 工艺改进到可生产状态，因为我们的内部客户需要它来生产 Pentalake。我还想建立一个值得信赖的代工厂，我称之为系统代工厂，采用系统方法。我也拥抱行业标准的 EDA 和 IP。每个客户都有自己独特的设计风格、设计方法和最喜欢的 EDA 工具。你不能告诉客户他们必须更换 EDA 工具。你要拥抱他们一直在使用的东西，随着时间的推移，你可以引入一些更好的想法。与其构建我们自己的定制 EDA 工具，我应该拥抱行业标准。这就是为什么我联系了所有 EDA 厂商来支持我。作为值得信赖的合作伙伴，满足客户要求更为重要。我们想成为一个服务型企业。你需要非常有效地服务客户，按时交付，并提供可靠且可预测的良率。我们必须学习这一点。这是一个服务行业，我们必须支持这一点。总体而言，客户的反馈相当积极。我们另一个非常擅长的领域是独特的先进封装技术，比如 EMIC 和 3D 堆叠。这些将是我们可以为客户提供的有趣选择。我们可以提供高质量、可靠的良率，以便他们可以使用。我们正在朝这个方向努力，再次强调，我们可以合作支持我们的客户。

First of all, I thank you for coming to my stage together at the Intel Vision Foundry event. I know there was a call, but you managed to come over to support me. I thank you for that. Customer first. That's right. Thank you. Clearly, I think we have a unique opportunity. We have the only United States domestic advanced technology process design, R&D, and manufacturing in Intel Foundry. I think that is a good advantage for us. Also, customers are looking for a flexible, resilient, secure foundry, and we can play that role. I'm really pleased to see that the Trump administration really supports American technology and manufacturing leadership. So I think we have the support to do that. My laser focus is to make sure our 18A is improved to production viable, because our internal customer needs that for Pentalake. I also want to build a trusted foundry, what I call a system foundry with a system approach. I also embrace industry standard EDA and IPs. Every customer has their own unique design style, design methodology, and favorite EDA tool. You don't tell the customer they have to change their EDA tool. You want to embrace what they have been using, and over time you can introduce some better ideas. Rather than build our own custom EDA tool, I should be embracing the industry standard. That's why I reached out to all the EDA players to support me on that. It's more important to meet the customer requirement as a trusted partner. We want to be a service business. You need to service a customer very effectively, on time delivery, with reliable and predictable yield. We have to learn that. This is a service industry, and we have to support that. Overall, the feedback from customers is quite positive. Another thing we are very good at is unique advanced packaging technology, like EMIC and 3D stacking. Those will be interesting options we can bring to customers. We can deliver high-quality, reliable yield so they can use that. We are working towards that, and again, something we can collaborate on to support our customers.

你提到的一个关键点是：代工业务是基础技术，比如 18A，晶体管工作得如何，背面金属工作得如何，但还有赋能：工具、IP、针对 ARM 核心或关键 IP 或 GPU 优化的流程。我们的台湾朋友在 15 年里做得很好，优化了整个流程。那么你怎么看？这是否是英特尔需要更多做的事情，当然要有标准流程，但真正投资于生态系统？我想这就是你更关注它的原因，但很高兴听听你的想法。

One thing you mentioned which is key: the foundry is the base technology like 18A, how well the transistors work, how well the backside metal work, but also the enablement: the tools, the IP, the flows optimized for ARM cores or critical IPs or GPUs. Our friends in Taiwan have done a great job over 15 years optimizing this full flow. So how do you see that? Is that something Intel needs to do more of, of course have standard flows but really invest in the ecosystem? I guess that's why you are putting more focus on it, but it'll be good to get your thoughts on that.

是的，绝对。我认为我们必须非常专注。事实上，Cadence 对我们来说是一个很好的合作伙伴。在某种程度上，在 Cadence 的 15 年也给了我优势。

Yeah, absolutely. I think we just have to be really focused. In fact, Cadence is a great partner for us. In some way, the 15 years at Cadence also give me the advantage.

所以某种程度上，我了解专利识别、客户偏好的路线以及他们的设计方法，这样我们才能真正合作，更好地服务客户。归根结底，对于代工厂来说，你需要建立可靠性、良率性能和量产能力，这对客户至关重要，因为客户的收入依赖于我们的生产。无论我们用 EDA 工具和 IP 设计什么，最终都要在代工厂交付。如果我们良率不达标或延误进度，他们的收入就会受影响。所以我们这边责任重大，这也是为什么我一直保持谦逊，向客户学习，最重要的是按时、按质交付，建立信任——这是我们强调的。

So in some way I know the patent recognition where are the customer prefer route and their way of u design methodology so that we can really work together and then serve the customer better and then uh I think overall at the end of the day for foundry you really need to kind of build that kind of reliability the yield performance for volume production because it's critical for the customer and the customer uh their revenue depend on our production and so in some way whatever we design using the EDA tool and IP at the end of day that they come to the foundry we have to deliver if we miss the yield or we miss the schedule they get affected by the revenue and so those are a lot of responsibility for our side and that's why I always want to be humble and learn from the customer and most important deliver on time on quality and that kind of building the trust us that we emphasize that.

哦，有道理。那么，Lip-Bu，还有一件大事——我们听你讲过 AI 战略和代工厂，但执行当然非常关键，对吧？PPA 做得怎么样，进度如何，无论是内部项目、大客户项目还是代工厂产品。你当然非常熟悉芯片设计流程，而且这方面有很多创新，比如 AI、新工具 Millennium、Palladium、Cerebras、Vericium。你怎么看待采用这些工具？Cadence 和 Intel 如何合作来加速这种文化变革、聚焦最佳产品？在你看来，设计流程有多重要？

Oh, makes sense. So bibu also uh you know one thing which is big you know so we we heard from you on the AI strategy you know about the foundry but of course the execution of that is super critical right how well the the PPA is there how well the schedule is there for all these whether it's internal projects or project with major customers or foundry products so of course you are very familiar with the chip design flow right and you know there's a lot of innovation in that with AI with new tools like Millennium with Palladium you know Cerebras Vericium so how do you see that to adopt these tools you know and how can Cadence and Intel work together to turbocharge this cultural change this focus on best product and how important is the design flow in your mind that yeah

我觉得这是个很好的问题。首先，我想祝贺你们用 AI 提升工具生产力。我还记得我在的时候，用 AI 机器学习驱动 PPA 提升了 15% 到 20%。我想你们所有产品都有 AI 五大支柱吧？是的。那么请教我们、分享经验，我们如何用这种方式更好地服务客户？我认为我们有很多合作机会，可以互相学习，你们也能帮助我们。我一直记得，对于最先进的开发，你不能直接把设计扔给代工厂，指望他们就能搞定。我认为需要与客户建立三方合作伙伴关系，让客户成功。

I think that's a very good question you know first of all I want to congratulation to you you know you're using AI to drive some of your tool productivity. I still remember when I was there it's like 15 20% improvement in PPA using the AI machine learning to drive those uh I think you have across all the product have the AI five major yes and then please teach us and share with us how can we serve our customer better that way and so I think there's a lot of opportunity for us to work together right and then we can learn from each other and then you also can help us and I always remember you know for the most advanced development. You cannot just throw your design and then to the foundry they can just make it work and I think that three-way partnership with the customer to make a customer successful.

绝对。我们双方合作很重要，我称之为“首次通过成功”。如果我们合作更紧密，机会就更大，然后加速高性能结果、按时交付，这对客户至关重要。所以总的来说，封装方面也一样——你们有很多技术我们可以学习，以及如何帮助我们提升良率、改进设计，服务内部客户和我们要支持的客户。我认为这对双方是双赢。

Absolutely. that is important for both of us to work together and I think that's uh I call it is the first time pass success and the opportunity is higher if we both work closer together and then uh and then accelerated high performance uh result on time delivery that's critical for the customer so I think all in all I think uh same thing with the packaging side yes you have a lot of uh technology that we can learn from you and then how we help for us to improve our yield, improve our design and then the internal customer and also the customer we want to support. I think that's a win-win for both.

是的。我的意思是，就像你提到的，有很多领域，比如 Palladium 首次正确。当然，用 AI 数字流程、验证流程来改进产品 PPA，如果优化得当，效果几乎可以媲美一个技术节点。封装也有很丰富的历史，我知道你们有几个——我和另一个大客户聊过，他们和你们做了很多封装。封装的一大块就是所有这些仿真：热仿真、电磁仿真，不仅仅是集成和 Allegro。今天你们发布了 Millennium，可以加速所有这些——是的，谢谢，甚至已经有订单了，很棒。所以我认为有很多可以做的，来革新像 Intel 这样的公司内部设计流程的执行。然后你谈到系统代工厂，它超越了晶圆，更像一个全栈平台。

Yeah. Yeah, I mean there like you mentioned there a lot of areas like like palladium first time right you know uh of course using AI you know digital flow verification flow to improve the product PPA it can be almost as as big as one technology node right if the right optimizations are done and also packaging has such a rich history right and I know you you have several I I talked to one other big customer they they're doing lot of packaging with you and you know packaging one big thing as you know is like all this simulation of of thermal simulation electromagnetic simulation so not just integrate and allegro and you know today we launched millennium which can accelerate all these yeah thank you even have orders yeah yeah it's great on the spot so so I think there is lot that can be done to kind of uh revolutionize the internal execution of the design flow right that at a company like Intel and then you know you you you talk about uh you like system foundry right that goes beyond wafers to like a full stack platform.

那么你怎么看这个？在产品方面，我们可以做很多改进，当然有些东西在代工厂方面是通用的。但你怎么看整个 3D 堆叠系统代工厂？在代工厂方面需要什么来支持它？这类似但又不同，比如 IP 要求不同，我们需要参考流程和示例。

So how do you see that on the so this is on one on the product side right we can do a lot of improvement and of course some of the things are common on the foundry side but what do you see this whole 3D stacking system foundry and what is needed to enable that on on the foundry side which is which is similar but different like IP requirements are different and you know we need to have a reference flow we need to have examples yeah yeah

这又是一个我认为我们可以合作的领域，更好地服务客户——内部客户和外部客户。另一部分，显然我们可以共同推动一些新技术，比如 18A，我们非常依赖 EDA 支持。我们还有一些非常独特的 RibbonFET 和全环绕栅极晶体管技术，以及背面供电。这对 Cadence 来说是一个独特的局面，但我们确实需要帮助客户用你们的工具设计得更好，这样我们最终能共同服务客户。所以我认为这是很好的合作机会，真正让客户受益。

Again that's an area that I think we both can collaborate and serve our customer better our internal customer also external customer and then the other part you know clearly you know we can really drive some new technology together you know the like the you know the 18A we really depend on the EDA support and then we also have some really unique uh ribbon fat and uh get all around transistor technology and also the backside power and then that can be a very unique situation that we have at Cadence and then But we really need to help customer to design better and with your tool and so that we can really you know at the end of the day serve our customer together and then so I think that great opportunity that we can work together and then uh so that we can really benefit the customer.

你知道，Lip-Bu 在 Cadence 有过几句名言。其中一句著名的就是“一个 Cadence，一个团队”，对吧？Lip-Bu 非常信奉“一个团队”的理念。我们以前也这样做过，就像你说的，与客户、代工厂三方合作。我认为这是一个独特的机会，我们可以成为一个团队，尤其是在产品方面，当然在代工厂方面也是，与 Intel 组成一个团队，共同争取大客户。

You know, Libu had several uh sayings, you know, quotes at cadence. You know, one one famous quote just is uh you know, one cadence, one team, right? So, Libu is a big believer of one team concept, right? And I think with and we used to do that like you said three-way with with some customers with the foundry and the customer and I think this is a unique opportunity. We can be one you know especially on the definitely on the product side but also on the foundry side one team with Intel as you go after these big big customers.

没错。我认为协作和伙伴关系是非常重要的成功法则。另一部分，回到我自己的 Intel 这边，销售和工程团队必须作为一个团队并肩工作，而不是互相指责、过度打折、说对方产品糟糕——那永远不会发生，对吧？永远不会。所以我认为必须作为一个团队工作，我现在正在 Intel 团队中推行这一点。每次与客户开会，我都要求工程和销售团队同时向我汇报进展，这样工程方面知道要交付什么进度、期望什么样的质量和性能，销售方面则要“低承诺、高交付”，让客户满意。最糟糕的是过度承诺，然后产品糟糕，进度还一再延误。那就会有大麻烦。

That's right. I think the collaboration and partnership is very important success formula and then the other part you know just back to my own intel side is very important sales and engineering they work side together as a team and then rather than pointing finger at each other you discount too much and then you are and your product suck that never happens right never happen exactly so I think it just have to work as one team and right now I'm impost that into my Intel team. You know this every every meeting with the customer I want the engineering and the sales team they both tell me where we are so that we make sure that we engineering side know what the schedule going to deliver what kind of quality performant we are expecting and so the sales people also going to be presenting under promise overd delivering that kind of delight the customer. The worst thing you want is you overpromise and then your product suck and also the schedule you keep missing. Then you have a big problem you know.

太好了。我们期待做更多、更多的事。

It was great. We look look forward to doing a lot more lot more.

我们等待这个机会很久了，希望能与英特尔做更多事情。我非常高兴。好的一面是，你了解行业趋势，知道什么有效，也知道其他玩家在做什么。而且，你仍然与所有公司保持友好关系，无论是合作伙伴还是竞争对手，你都和他们保持着友谊，所以你知道他们在做什么。这太棒了。

We're waiting for this opportunity for a long time to do a lot more with Intel. I'm so delighted. And the good thing is you know the industry trends, you know what works, you know what the other players are doing. And the good thing is that he still is friends with all the companies, whether they are your partners or competitors, you're still friends with all of them, so you know what they're doing. This is great.

现在我想稍微换个话题。你提到过，无论是担任 Cadence CEO 还是现在担任英特尔 CEO，你仍然拥有投资者和风险投资家的视角。当然，你总体上是最成功的投资者之一，尤其是在半导体和电子系统领域。那么你看到了什么？从初创企业生态和风投生态来看，有哪些新兴技术？这些技术又如何与英特尔未来的方向保持一致？

Now I think changing a little bit. You mentioned that when you were Cadence CEO and now Intel CEO, you still have the vantage point of an investor and VC. And of course you are one of the most successful investors in general, especially in semiconductors and electronic systems. So what do you see there? What is the emerging technology from the startup ecosystem, from the VC ecosystem, and then how does it align with Intel's direction going forward?

是的，这非常令人兴奋。显然，人工智能正在改变格局。你很少能遇到一个大浪潮，有时浪潮非常大，你不想错过它。如果你错过了一个浪潮，你就错过了整个事情。所以我认为密切关注地平线上正在发生的事情非常重要。你可以清楚地看到，人工智能和生成式人工智能正在飞速变化。看看 OpenAI，他们多快就达到了 1 亿用户，现在几乎有 6 亿用户。采用新技术的速度太惊人了。最近我和一个朋友聊天，他们谈到了整个智能体式 AI 和推理模型，这正在发生。你有智能体，以及智能体之间的交互。这样一来，它确实能帮助你处理很多日常事务，甚至在某种程度上，你越来越不需要依赖程序员来编程了。你可以真正使用智能体来做这件事。所以我认为这是颠覆性的，是一个非常大的浪潮。而且现在它与软件密切相关。所以我认为你必须审视整个格局的变化：我们如何从硅片的角度、从工具的角度切入，如何推动这些变化？每天都有海量数据产生，你如何利用这些数据更快地为自己谋利，推动你想要的结果，并更快地转化结果？我想你从你的工具中看到了，你在使用人工智能和机器学习方面取得了多大的进步，不仅提高了内部效率，还帮助客户提高了效率。我认为这是一个巨大的机会，尤其是当我们从关税转向代币的时候。最近在一次电话会议上，人们谈到了从关税到代币。是的，代币是真正的新方式，那是更重要的 T 字词。所以我认为这很令人兴奋。我在风险投资行业干了 40 多年，见证了许多变化。我参与了 500 家初创公司，而且非常幸运：我有 145 次 IPO。我认为没有哪个风投能声称做到这一点。这太惊人了。我还投资了 251 家出色的半导体公司，这让我有了一些经验。当然，现在我试图将这种纪律带到英特尔。所以我们很乐意与你合作，共同推动一些成功，更重要的是，创造几个我们都能享受的大平台。我绝对乐意这样做。

Yeah, this is very exciting. Clearly, AI is just changing the landscape. Only a few times do you see a big wave, and sometimes the wave is very big, and you don't want to miss the wave. If you miss a wave, you miss the whole thing. So I think it's very important to keep an eye on what is on the horizon, what is developing. You can clearly see that AI and generative AI are changing so fast. Look at OpenAI, how quickly they got to 100 million users, now it's almost 600 million users. The speed of adopting new technology is so amazing. Recently I was talking to a friend of mine, and they were talking about this whole agentic AI and the reasoning model, and it's happening. You have agents and agent-to-agent interaction. And in that way, it really helps you drive a lot of your daily routine, and even in some way, you less and less depend on programmers to program now. You can really use the agent to do that. So I think this is game-changing, and it's a very big wave. And it has a lot to do with software now. So I think you just have to look at this whole horizon change: how can we intercept from the silicon point of view, from the tool point of view, how do we drive these changes? There's so much data, massive data generated every day, and how do you use that for your own benefit in a quicker way to drive the results you want and convert the result much quicker? I think you see that from your tool, how much improvement you're using AI and machine learning to drive not just your own internal efficiency, but also help your customer drive efficiency. I think that's a tremendous opportunity, especially as we move down the path from tariff to tokens. Recently in a conference call, people talked about tariff to tokens. Yes, token is the real new way, that's the more important T-word. So I think that's something exciting. Being in the venture business for more than 40 years, I've seen a lot of changes. I've been involved with 500 startup companies, and I've been very lucky: I have 145 IPOs. I don't think any VC can claim that. It's amazing. And I've invested in 251 fabulous semiconductor companies, so that gives me a bit of flavor. Of course, now I try to bring that discipline to Intel. So something we would love to work with you and really drive some of the success together, and more importantly, create a few big platforms that we all can enjoy. Absolutely love to do that.

有一点，正如你所说，你对这个领域、投资领域以及整个初创企业非常了解。我最近注意到，大量资金正在涌入机器人和物理 AI，对吧？不仅在硬件方面，还有像世界模型这样的人工智能模型。就连英伟达和 Jensen 也在谈论这些新型大语言模型，为物理世界打造的新 AI。我们讨论过这个，我知道有一段时间了：汽车、机器人、无人机。它可能和云部分以及 PC 部分一样大，甚至更大，我知道这对英特尔很重要。就像你提到的，数据中心和 PC，没问题。但与此同时，还有这个新的物理 AI。而且我知道有 Mobileye，对吧？那是一个大项目。那么你有什么看法？从投资角度和英特尔角度，你对物理 AI 这部分有什么看法？

One thing, you know this space, the investing space and the whole new startups, very well like you said. One thing I noticed recently is that there is big funding going into robotics and this physical AI, right? Even on the hardware side, but also on these AI models like these world models. Even Nvidia and Jensen talks about it, these new kind of LLMs, new AI for the physical world. And we have discussed this, I know for some time: cars, robots, drones. It could be as big or bigger than the cloud part and the PC part, which I know will be important to Intel. Like you mentioned, the data center and the PC, okay. But at the same time, there's this new physical AI. And I know there's Mobileye, right? That's a big one. So what's your thoughts on that? Is that something you are, both from an investing standpoint and also an Intel standpoint, what's your views on the physical AI part of it?

好问题。显然，从生成式 AI 到智能体式 AI 已经在发生。我认为物理 AI 是下一个浪潮。这很有道理，因为过去机器人只是每次教它做一个动作：拿起东西，放到别处，精确执行。但现在，机器人几乎可以像另一个人一样。我们称之为类人机器人。这样一来，你拥有所有 AI 软件、智能体式 AI，并赋予它智能。它们可以推理，可以进行强化学习，可以执行多项任务，而不仅仅是一个任务。因此，机器人可以像人类一样学习并执行不同的功能。我唯一还没看到的部分是情感部分。情感部分可能会在以后出现。但现在，因为有些人类也没有情感部分——这是真的。无论如何，我认为潜力巨大。你可以看到特斯拉是一个很好的例子；他们有很多类人机器人。而且你我都与 CMU 有联系。CMU 在机器人领域非常强大，所以有很多新东西涌现出来。一些初创公司非常令人兴奋。从风投方面，我们也在该领域进行了三项投资，涉及一些工业应用和其他不同应用。另一个我非常兴奋的领域是医疗。看看 Intuitive Surgical，我们的好朋友是那里的董事长。他们如何以精确的方式推动整个心脏外科手术——这太惊人了。人工智能和机器人能在多大程度上帮助人类。有时可能不能完全由它们自己完成；你需要有人类的判断。但它在帮助人类更准确、解决问题方面有很大作用。所以我认为这部分是我非常兴奋的一个大领域。我对此非常兴奋，以至于我加入了 UCSF 的执行委员会。

Good question. Clearly, from generative AI to agentic AI is already happening right now. And I think physical AI is the next wave. It makes a lot of sense because it used to be that robotics just taught the robot to do one trick each time: pick up something, put it somewhere else, with precision. But now, robotics can be almost like another human being. We call it humanoid. And in that way, you have all the AI software, agentic AI, and give it intelligence. They can reason, they can do reinforcement learning, and they can do multiple tasks, not just one task. So in that way, robots can learn and do different functions just like human beings. The only part I don't see yet is the emotional part. That emotional part may come later on. But right now, because some humans don't have the emotional part either — that is true. Anyway, I think the potential is huge. You can see that Tesla is a good example; they have a lot of humanoids. And you and I are involved with CMU. CMU is very strong in robotics, and that's why there's a lot of new stuff coming out. It's so exciting, some of the startups. From the venture side, we also made three investments in this area, in some industrial applications and other different applications. The other part I'm very excited about is medical. If you look at Intuitive Surgical, our good friend is the chairman there. How they drive that whole thing with precision in terms of heart surgical operation — that is amazing. And how much AI and robotics are able to help human beings. Sometimes it may not be completely done by themselves; you need to have human judgment calls. But it helps a lot in terms of humans being more accurate and addressing the issue. So I think this part is a big area of excitement for me. I was so excited about it that I joined the executive council at UCSF.

这确实给了我很多机会，我投资了一些公司，比如 Altera AI。你知道，我们所有人年纪大了以后，显然都会面临前列腺癌的问题。如果你看 PSA 测试、4K 测试，准确率只有 35%。所以我决定支持一位教授兼肿瘤学家，他是前列腺癌的专家。然后基本上能够实现 92% 的准确率，无论是诊断方式、治疗方法还是治疗频率。你不想因为过度治疗而影响生活。所以我觉得这些事情非常令人兴奋。在不到一年零三个月的时间里，他们获得了国家指南的批准，这简直是闻所未闻。不到两年，我们就让医疗保险支付每次 750 美元的测试费用。今年我们期待获得 FDA 批准。而且他们并没有止步于此。现在他们正在研究乳腺癌、肝癌、结肠癌以及头颈癌。对我来说，我认为这只是一个巨大的平台，而且他们拥有数据。

And then it really gave me a lot of opportunity and I invested in some companies like Altera AI. You know, all of us when we get older, I think is clearly going to have some prostate cancer issue. And then if you look at the PSA test, 4K test, it's only 35% accurate. So I decided to back a professor and oncologist, an expert in prostate cancer. And then basically able to deliver 92% accurate in terms of the way you diagnose it, the way you treat it, and how frequently you treat it. You don't want to be overdosed and affect your life. So I think those things are so exciting. In less than one year and three months, unheard of, they got the national guidance approval. In less than two years, we got medical insurance paying $750 per test. And now we're looking forward this year to FDA approval. And they are not stopping there. Now they're doing breast cancer, liver cancer, colon cancer, and neck and head cancer. To me, I think it's just a huge platform and they have the data.

同样，另一个让我更兴奋的公司叫 Echo Neuro Technologies，仍处于隐身模式。我和其他几个好朋友一起投资了。这是 Eddie Chen，他是一位顶尖的神经外科医生。我认识了他，我们成了很好的朋友，他是我的邻居。他说，我希望有一天我不需要来看你，但你很迷人。他说，人们来我的实验室看看我有什么。然后基本上利用 AI 机器学习，能够帮助帕金森患者以 80% 的准确率进行交流，而不用像霍金那样不停地敲键盘。所以基本上是用半导体、传感器，通过脑电波来解读大脑，理解大脑如何交流。接下来，在很短的时间内，我把他介绍给所有半导体公司，帮助他构建那个系统。他说，Lip-Bu，我要构建，你能帮我雇一个 AI 首席架构师吗？我说，你需要 AI 首席架构师做什么软件？他说，Lip-Bu，我要构建大脑的解码器，作为大脑的基础模型。我说，你可以有一个与我的大脑镜像的基础大脑，然后我如何与你交流可以更准确地数字化。这是我从未想过可能发生的事情，而正是因为有了海量数据，以及利用 AI 机器学习的进步，感谢 Jensen 和他能够生成的所有算力，才能实现这一点。而且他没有止步于此。现在他正在研究大脑的不同部分。这就是为什么我学到了很多。这部分可能负责交流，这部分负责视觉，另一部分负责推理。所以他说，我可以帮助抑郁症患者，我要帮助阿尔茨海默病和癫痫。我说，天哪，你会获得诺贝尔奖的。他四十多岁，仍然令人惊叹。这让我非常兴奋。我知道你收购了一家叫 Open Eye 的公司，很多制药公司都在用它。加倍下注，因为我认为这是世界需要的。而且在某种程度上，除了做生意，你还在帮助人类。我们都参与其中，正如我们多次讨论的 AI 的三个阶段：基础设施阶段、物理 AI 阶段，然后科学就是 AI，其中很多将是生命科学。是的，将是生物学。我的意思是，也会有其他科学，但那是最大的潜力。还有那些我们现在用 AI 做模拟的事情。我的意思是，很多都是薛定谔方程。是的，它们需要大量算力。是的，当然，在电路中，我们做麦克斯韦方程等等。薛定谔有时甚至更复杂。所以随着 AI 的发展，算力需求要高得多，然后对社会产生巨大影响，就像你说的那样。

And likewise, the other one even more exciting for me is called Echo Neuro Technologies. Still in stealth mode. I invested with a couple of other good friends. And this is Eddie Chen. He is a top neurosurgeon. I got to know him and we became very good friends. He's a neighbor to me. And he said, I hope one day I don't need to come to visit you, but you are fascinating. And he said, people come to my lab, take a look at what I have. And then basically using AI machine learning, able to address helping Parkinson's patients to 80% accurate to communicate without keep punching the keyboard like Stephen Hawking did. So basically using semiconductor, using sensor, doing brain wave to interpret the brain, how the brain communicates. Then the next thing I know, in a very short time, I introduced him to all the semiconductor companies to help him build that system. And he said, Lip-Bu, I'm going to build, can you help me hire a chief architect for AI? I said, what do you need a chief architect for AI for software? He said, Lip-Bu, I'm going to build the decoder of the brain as a foundation model of the brain. I said, you can have a foundation brain mirror image to my brain, and then how I'm going to communicate you can more accurately digitally. This is something that I never thought could happen, and just because of the massive data we have and using AI machine learning advances, thanks to Jensen and all the compute power that he is able to generate, and then able to help that. And he's not stopping there. Now he's moving into different parts of the brain. That's why I learned a lot. This part maybe for communication, this part for vision, and the other part for reasoning. So he said, I can help depression patients, I'm going to help Alzheimer's and epilepsy. I said, boy, you're going to be a Nobel Prize winner. And in his late 40s, still amazing. This is something that got me so excited. And I know you bought a company called Open Eye, and a lot of pharmaceutical companies are using it. Double triple down on it because this is something that I think the world needs. And also in some way, besides doing business, you help the human population. We are all in, as we discussed the three phases of AI so many times: the infrastructure phase, the physical AI phase, and then the science is AI, and a lot of it will be life sciences. Yes, it will be biology. I mean, there will be other sciences too, but that is the biggest potential. And also those things like we do simulation there now with AI. I mean, a lot of them are Schrödinger equations. Yes, they need a lot of compute. Yes, of course, in circuits, we do Maxwell equations and all that. Schrödinger sometimes even much more complicated. So the compute requirement is much higher along with AI, and then the huge impact like you're saying to society.

事实上，如果我有空闲时间，我会研究其中一个。我有一个非常好的朋友得了脑瘤。哦不。而且是非常侵袭性的那种。然后我开始研究，因为这个朋友对我来说非常珍贵。我了解到，只有 1% 的这类脑瘤患者能存活五年。1%。不。然后他实际上是第 18 个接受 CAR-T 治疗的人。C-A-R-T。这基本上是通过注射改变免疫系统，创造一种不同的方式来治疗和杀死癌症。这让我着迷。而且，在美国只有 18 个人能负担得起这种治疗。我说，天哪，如果这种治疗能让更多人用上，那就太好了。我认为那会很有帮助。所以回到重点，基本上是利用 AI 机器学习，结合你获得的所有活检数据，然后使用巧妙的架构来驱动，我认为那是一个大前沿。从 EDA，从半导体公司，我们能扮演什么角色？所以回到我的观点，你真的需要看解决方案，你要解决什么问题，然后反向推导。然后需要什么样的系统、软件、芯片，以及用什么工具来优化，然后实现它。所以我认为这驱动着我，令人兴奋。

And in fact, one of them, if I have spare time, I work on it. I have a very good friend of mine who has a brain tumor. Oh no. And this very aggressive one. And then I started to study because the friend is very dear to me. And I studied that only 1% of patients with this brain tumor survive in five years. 1%. No. And then he is actually number 18 to get the treatment called CAR-T. C-A-R-T. And then this is basically you inject the immune system, changing and creating a different way to treat and kill the cancer. And it's fascinating to me. And there, this treatment only 18 people in the United States could afford to even pay for this. And I said, boy, it would be great if this one could be made more available for the population. I think that will help a lot. So again, back to the point, basically using AI machine learning and then with all the biopsy that you get, and then using clever architecture to drive that, I think that's a big frontier. From the EDA, from the semiconductor company, how can we play a role? So back to my point, you really need to look at the solution, what problem you're trying to solve, and then work backward. And then what kind of systems, software, the silicon, and then with the tool to optimize for it, and then make it happen. So I think that to me drives me, exciting.

那么 Lip-Bu，现在问一些个人问题。我们时间不多了，但你有这么多事情：我们谈到的长期事情，当然还有短期事情，比如财报电话会议和交付。那么你如何平衡呢？

So Lip-Bu, some personal questions now. We don't have much time, but you have these multiple things: long-term things like we talked about, and then of course there are short-term things like earnings calls and delivery. So how do you balance that?

好问题。每个人都问我这个问题。我知道你每天游泳。我希望你还在每天游泳。我仍然每天游泳，每天去健身房。然后我一个朋友说，我终于发现了你的秘密。我说，什么秘密？能分享给我吗？他说，我知道你有两个双胞胎兄弟。他说，和我一起吃晚饭的那个很安静。那是你的双胞胎兄弟。另一个则有点健谈。所以这就是我的答案。但说正经的，我认为很明显，我们都只有 24 小时。我们不能有 48 小时。所以我认为上帝是公平的。只给了我们 24 小时。所以我认为最好的事情是经营公司或做我喜欢做的事情。非常重要的是你要判断：你需要参加这个会议吗？如果你不参加这个会议，而且你没有价值可以贡献，那就直接取消会议。

Good question. Everybody asks me that. So I know you swim every day. I hope you still swim every day. Still swim every day and the gym every day. And then a friend of mine said, finally I found your secret. I said, what's the secret? Can you share with me? He said, I know that you have two twin brothers. He said, the one I have dinner with is very quiet. That's your twin brother. Then the other one is kind of a little bit more talkative. So that's my answer. But joking aside, I think clearly, we all have 24 hours. We cannot have 48 hours. So I think God is fair. We just give us 24 hours. So I think the best thing is running a company or doing things that I like to do. It's very important you try to judge: do you need to be in this meeting? And if you don't have this meeting and you have no value to add, you just cancel the meeting.

这就是我在 Cadence 的风格，我也把它用到了 Intel。我想确保在别人给你安排一大堆会议之前，你得自己决定哪些会议该参加，其余的可以授权团队去做。这样某种程度上效率更高。另一方面，我一直鼓励年轻一代。做导师时，我觉得很重要的一点是建议他们保持好奇心，持续学习。如果不持续学习和适应，就会停滞不前。我很高兴看到一些榜样，即使从高管职位退休，也在扮演不同的角色，要么帮助政府，要么帮助行业创造新产业。这让我觉得兴奋，也鼓舞着我们所有人。然后保持谦逊，努力工作。

So that's kind of my style at Cadence and I applied that to Intel. I want to make sure that before they load up with a lot of meetings for you, you have to decide which meetings you should attend; the rest you can empower the team to do. So in some way, you try to be more efficient that way. The other part is I always encourage the younger generation. When we do mentorship, I think it's very important to advise them that you want to keep that curiosity in mind and continue to learn. If you don't continue to learn and adapt, then you stagnate. I'm kind of happy to see some role models who, even though they retire from executive roles, play a different role, either helping the government or helping the industry create new industries. That to me is exciting and very encouraging for all of us. And then stay humble, work hard.

我一直很佩服你有 CMU 的博士学位。所以这个周末我终于要拿到荣誉博士学位了。我听说这很了不起。

I always admire you have a PhD from CMU. So finally this weekend I'm going to have the honorary PhD. I heard that's a big deal.

是啊，某种程度上我得了安慰奖，因为我现在太老了，拿不了博士学位。虽然我修完了核工程的所有课程，但因为我的岛屿事故，没有核电站建成，所以没毕业。后来 MIT 邀请我回到顾问委员会，而 CMU 比 MIT 更好，获得荣誉学位非常 prestigious，而且你还要做毕业演讲，对吧？所以长话短说，我没你聪明，但没拿到博士学位却拿到了荣誉学位。

Yeah, so in some way I got the consolation prize because I'm too old to get the PhD now. Even though I finished all my classes in nuclear engineering, but because I decided through my island accident there's no nuclear power plant built, so I didn't finish. Then MIT asked me to come back to the advisory board, and because CMU is better than MIT, that's very prestigious to get an honorary degree, and you're giving the commencement speech, right? So long story short, I'm not as smart as you, but I get an honorary degree without getting the PhD.

太棒了。Limbu，我们可以聊很久，但我知道你今天有董事会会议。所以感谢你抽空过来。这对我意义重大，对 Cadence 也意义重大。我们祝愿 Intel 一切顺利。我们会努力，作为一个团队让 Intel 成功，并与 Cadence 携手共进。

This is fabulous. Limbu, we can talk for a very long time, but I know you had a board meeting today. So thank you for taking time to come. It means a lot to me. It means a lot to Cadence. We wish all the best to Intel. We'll work hard, one team to make Intel successful and together with Cadence together.

是的。谢谢你，Lu。加油。

Yes. Thank you, Lu. Heat.