高通 CEO 克里斯蒂亚诺·安蒙谈 AI 驱动的计算变革

大家好，欢迎来到下午的会议。我们的超级明星 CEO Cristiano 来了。我们非常欢迎 Cristiano 和高通来到 Cadence LIVE。我们与高通有着长期的合作伙伴关系，而在 Cristiano 的带领下，这一关系变得更加紧密。我们有很多共同点，包括对汽车的热爱。你们得问问 Cristiano 关于他的车的事。他甚至在修复一些车，我可达不到那个水平，好吧。还有我们对 F1 的热爱。我想我们也在拉斯维加斯见过面。所以非常高兴欢迎高通和 Cristiano。请。

Hello everyone. Welcome to the afternoon session. Our superstar CEO, Cristiano, is here. We really want to welcome Cristiano and Qualcomm to Cadence LIVE. We have a long-standing partnership with Qualcomm, and it has really gotten even better with Cristiano. We share a lot of things together, including love of cars. You have to ask Cristiano about his cars. He's even restoring some cars, which I'm not in that league, okay. And also our love for F1. I think we met in Vegas as well. So it's a real pleasure to welcome Qualcomm and Cristiano. Please.

这个介绍让我很有压力。叫我超级明星。我跟人说，在当今科技界，你必须有盈利。如果没有盈利，你就必须有增长。或者你有个明星 CEO。我们没有明星 CEO，所以我们必须追求盈利和增长。我们只有这两个选择。

That was a very intimidating introduction. To call it superstar. I tell people, in today's tech world you have to have earnings. If you don't have earnings, you have to have growth. Or you have a celebrity CEO. We don't have a celebrity CEO, so we have to chase earnings and growth. Those are the only two options we have.

是的，你什么都做了。我想在座的多数人都知道，高通在过去的 10-15 年里的许多演进和革命中处于核心地位。但你也在改变高通，对吧？我们合作了一段时间，但现在公司已经变了，并且还在向前发展。所以也许你可以告诉我们，高通是如何改变的。

Yeah, you're doing all the things. So, I think most people in this audience know, but Qualcomm is kind of central to a lot of the evolution and revolutions that have happened in the last 10-15 years. But you're also changing Qualcomm, right? We worked together for a while, but now the company has changed, and going forward. So maybe you can tell us about that, how Qualcomm is changing.

哦，很乐意。首先，我想再次说我很高兴来到这里。感谢有机会与你和 Cadence 活动的观众交流。我要说高通正处于转型过程中，我认为这对我们公司来说一直如此。我们公司一直处于转型过程中。大多数人可能知道高通以及我们在无线领域所做的工作。整个移动革命——对于不了解的人来说，高通代表质量通信。这就是我们名字的含义。人们知道我们在 CDMA、3G、4G、5G 方面所做的工作。我认为那一直是关于连接。它始于一个想法：创造一种利用无线电波的技术，让每个人都能拥有手机，我们能把它连接到互联网吗？我们能创造宽带吗？当我们进入转型——我喜欢说从 3G 到 4G 的转型——当时我们很清楚，当我们将宽带带到手机时，手机将变成电脑。我们将创造一个新的计算空间，用这个新术语来说。电脑需要能放入手掌中。大多数人可能不记得，但我们是第一家真正在电池供电的设备（手机）中交付主频 1 GHz 的 CPU 的公司。这在以前是不可能的。所以那改变了高通。那是第一次重大转变。从一家专注于无线通信的公司转变为一家计算公司。这推动了许多技术，这些技术构成了今天移动行业中的骁龙。那是从移动革命转向重新定义连接和处理。这就是我们现在所处的位置：我们构建非常高效的计算能力，因为从工程角度来看，手机非常苛刻。手机有三个不能妥协的领域：电池寿命，它不能发热——你不能在手机里放风扇——而且外形尺寸非常小。因此，我们为手机设计的计算技术，我认为我们开启了这一新转型，我们称之为重新定义连接处理。我们认为这些技术会有新的终端市场，这推动高通从一家主要为手机做连接处理器的公司进入 PC 领域，思考下一代 PC、汽车行业的数字化转型、新的 XR 类型应用和工业。我认为我们看到连接和高性能、低功耗计算有机会改变任何行业。然后随着我们走向未来，真正的是实现智能计算无处不在。我想我们都知道新的计算形式是 AI。英伟达所说的加速计算确实是处理信息的新方式。我们在边缘 AI 流行之前就开始谈论它了。所以我认为这就是高通的故事。

Oh, happy to. First of all, I just want to say again I'm very happy to be here. Thank you for the opportunity to talk to you and to the audience of the Cadence event. I'll say Qualcomm is in a process of transformation, and I think that has been true for our company. Our company has always been in the process of transformation. Most people probably know Qualcomm and what we have done in wireless. The whole mobile revolution—for people that don't know, Qualcomm stands for quality communications. That's what our name is. And people know what we have done in CDMA, 3G, 4G, 5G. And I think that has been about connecting. It starts with an idea of creating a technology for use of radio waves so everybody can have a cell phone, can we connect it to the internet? Can we create broadband? And as we get to the transition—I like to say the transition from 3G to 4G—it became clear to us at that point in time as we bring broadband to a cell phone that the cell phone will become a computer. We'll create a new computing space, to use this new terminology. The computer needed to fit in the palm of your hand. Most people don't remember this, but we were the first company to actually deliver the very first CPU with a clock of 1 GHz in a battery-powered device that was a phone. It was not possible before. So that changed Qualcomm. That was the first big change. From going from a company that was focused on wireless communication to becoming a computing company. And that drove a lot of the technologies that are part of what is today the Snapdragon in the mobile industry. And that was about moving from the mobile revolution to redefining connecting and processing. And that's where we are right now: we build computing capabilities that are very efficient because from an engineering perspective, the phone is very unforgiving. The phone has three areas that you can't compromise: battery life, it cannot be hot—you cannot put a fan in the phone—and the form factor is very small. So with the computing technologies we design for the phone, I think we initiated this new transition which we call redefining connected processing. We thought that there would be new end markets for those technologies, and that drove Qualcomm from a company that's primarily doing connected processors for phones into entering the PC space with next-generation PCs, the digital transformation of the automotive industry, new XR type of applications, and industrial. And I think we saw that connectivity and high-performance, low-power computing had an opportunity to transform any industry. And then as we go into the future, it's really about enabling intelligent computing everywhere. I think we all know the new form of computing is AI. The way Nvidia communicates about it—accelerated computing—is really the new way of processing information. And we started talking about AI at the edge before it was popular. So I think that's kind of the story of Qualcomm.

确实。我记得甚至你们出色的芯片总是有调制解调器、CPU，从长远来看很棒，然后还有 GPU。高通的 GPU 在移动形态中总是令人惊叹。所以也许你可以——你如何看待与 Cadence 的合作？你能谈一谈吗？

Exactly. I remember even your remarkable chips would always have the modem, the CPU, which was great long term, and then GPU. Qualcomm GPUs are always amazing in a mobile form factor. So maybe you can—how do you view the partnership with Cadence? Can you talk a little bit about that?

是的。看，我认为我们——这是一个非常广泛的合作伙伴关系。我要说，我会把它定义为高通真正的战略关系。如你所知，它非常广泛。但了解高通是好事。当我们考虑构建产品时，我们开发不同的 IP 引擎，无论是蜂窝调制解调器还是 Wi-Fi，无论是 CPU 还是 GPU、NPU、图像信号处理器。所有不同的 IP，我们设计它们以从——你可以想象从手表和耳机一直到汽车上具有数据中心类型性能的东西，例如，或在边缘计算解决方案中。我们构建不同的 SOC。我们与 Cadence 的合作伙伴关系非常普遍，因为我们开发 IP 并构建这些 SOC。我认为这是一家庞大的公司，我们喜欢随着 Cadence 的发展，我们的合作伙伴关系也在发展。我认为我们一直在用你们的 AI 引擎做的事情——我们真的很喜欢利用 AI 进行设计的演变。

Yes. Look, I think we're—it's a very broad partnership. I'll say it's a relationship which I would define as a really strategic relationship for Qualcomm. It's very broad, as you know. But it's good to understand Qualcomm. When we think about building our products, we develop these different IP engines, whether it's a cellular modem or Wi-Fi, whether it's a CPU or GPU, an NPU, an image signal processor. All of the different IP, and we design them to scale from—you can think about it from a watch and an earbud all the way to something that you have a data center type of performance on a car, for example, or in an edge computing solution. And we build different SOCs. Our partnership with Cadence is very pervasive as we develop the IP and we build those SOCs. I think it's a broad corporation and we like that as Cadence is evolving, our partnership is evolving, too. I think what we've been doing with your AI engines—we really like the evolution of leveraging AI for design.

我认为我们在验证方面所做的工作，我很高兴地报告，设计验证和仿真对于显著缩短设计周期时间、并真正使我们具备这种扩展能力至关重要。你应该这样想：如果没有那些类型的合作伙伴关系，我们就无法同时为这么多行业构建 SOC。我认为这正是与 Cadence 合作使我们得以实现的机会。

I think what we're doing in verification, and I'm happy to report that both design verification and emulation have been fundamental to significantly reduce the design cycle time and actually enable us to have this ability to scale. You should think about it this way: if we didn't have those types of partnerships, we wouldn't be able to build SOCs for so many industries all at the same time. And I think this is the opportunity that working with Cadence has enabled us to do.

是的，谢谢你，Cristiano。这是一次很棒的合作。实际上，即使在今天的 Cadence Live 上，高通也有几篇论文，讨论包括验证中的 AI 和设计中的 AI 等不同方面。所以我们非常珍视。而且，正如你所说，高通拥有全系列的设计类型和产品，对吧？模拟——你们在模拟方面很出色——数字、不同的调制解调器、CPU、GPU、NPU，当然，你们在世界各地也有团队。所以我们很高兴在美国合作，在印度合作，在所有其他地区合作。所以谢谢你。

Yeah, thank you, Cristiano. This is a great partnership. Actually, even today at Cadence Live, Qualcomm has several papers talking about different aspects including AI in verification and AI in design. So we truly value it. And also, Qualcomm has a whole range, like you said, of design types and products, right? Analog—you are great in analog—digital, different modem, CPU, GPU, NPU, and of course, you also have teams across the world. So we are glad to partner in the US, partner in India, all other geographies. So thank you for that.

哦，谢谢。非常感谢。

Oh, thank you. Thank you very much.

现在，考虑到高通的广度，然后行业的广度，还有地域的广度，对吧？你基本上在世界各地运营，涉及各种高科技行业。那么哪些趋势最让你兴奋？你知道，有很多事情在发生，但哪些趋势——

Now, in terms of given the breadth of Qualcomm and then the breadth of industries and also the breadth of geographies, right? You basically operate all over the world in all kinds of high-tech industries. So what trends excite you the most? You know, there's a lot of things going on, but what trends—

看，有几个趋势，我认为它们某种程度上与我们看到的扩大高通终端市场数量的机会相关。我们很早就开始的一个趋势，我们对此持乐观态度，我们继续投资，这只是时间问题，那就是物理世界和数字世界的融合。我认为这就是你今天看到的混合现实、虚拟现实、增强现实。我们 100% 确信这不再是硅的问题。我认为最终你会在光学方面获得规模经济，但特别是当你看到拥有另一个计算空间的能力时，也就是眼镜，我们认为这是一个巨大的机会，这推动了我们在这个领域所做的很多事情，我们与 Meta 的合作，与 Google、三星和许多其他公司的合作。我们真正喜欢的另一个趋势，而且——你知道，这对任何人来说都不应该感到惊讶——是 AI 获得规模的能力。而这个 AI——我们非常专注于设备上的 AI。设备上的 AI 的演变方式将与 AI 在云端的演变方式不同。我们真的很喜欢这个机会，这正在改变我们对计算的思考方式以及计算机在许多其他设备中的使用方式。所以另一个我喜欢的趋势是——有时汽车制造商不会那样谈论，但从我们作为芯片公司的角度来看，汽车正在成为另一台计算机，对吧？这是一个趋势，汽车是另一个计算空间，这就是——我喜欢这样传达，那就是我们过去认为，你的计算空间是你的台式机和笔记本电脑。然后突然手机出现了，我记得当人们想，哦，手机会取代 PC。现在，当然不会。它是一个新的计算空间。汽车将成为另一个计算空间，人们将在车里娱乐，在车里工作。所以我们对这个趋势感到兴奋。我认为另一个，也是公司的另一个增长领域，是——这是一个我认为需求无限的趋势。问题是有太多的摩擦，有太多不同的事情需要发生，但那就是所有企业的数字化转型。你知道，我喜欢举例。你可以选任何一个。你可以选任何垂直行业。如果你选能源，你选零售，你选医疗保健，你会看到这些行业如何被技术转型。就像我喜欢提供一些简单的例子。能源：你有一家公司做发电，你有一家公司做输电，然后你有配电，你有集中式发电。现在你到一个家庭，你有太阳能，你有电动汽车，你插上电，你有一个完全不同的能源管理解决方案。它们需要变得智能。这不仅仅是连接一个电表。你一直走到零售。现在每家商店都是电子商务。你需要管理货架上的库存，就像你用智能摄像头看到的那样。所以我们认为这是企业数字化转型的另一个大趋势。所以这些就是现在正在发生的事情，这就是为什么我们对半导体行业以及技术的作用，尤其是在边缘，持乐观态度。

Look, there's a number and I think they're somewhat associated with this opportunity that we saw to expand the number of end markets for Qualcomm. One that we started early, we're bullish about it, we're continuing to invest and it's just going to be a matter of time, is the measure of physical and digital worlds. I think it's what you see today with mixed reality, virtual reality, augmented reality. We're 100% convinced that it's no longer a silicon problem. I think eventually you're going to get the economy of scale in optics, but when you especially look at the ability of having yet another computing space, which is glasses, we see that as a big opportunity and that drives a lot of the things we're doing within the space, our partnership with Meta, our partnership with Google, Samsung and many other companies. The other trend that we really like and I will—you know, it should not be a surprise to anybody—is the ability of AI to get scale. And this AI—and we're very focused on AI on device. AI on device is going to evolve different than the way that AI evolves in the cloud. We really like that opportunity and that's changing how we think about computing and the use of computers in many other devices. So another one that I like is—sometimes the automakers don't talk about it that way, but from our perspective as a chip company, the car is becoming another computer, right? It's the trend which is the car is another computing space and that's—I like to communicate it that way, which is we used to think about it, your computing space was your desktop and your laptop. And all of a sudden the phone arrives and I remember when people thought about, oh, the phone's going to eliminate the PC. Now, of course not. It's a new computing space. The car is going to be another computing space and people are going to get entertained from the car, they're going to work from the car. So we're excited about that trend. And I think the other one, which is yet another growth area for the company, is—this is one I really consider unlimited demand. The problem is there's so much friction and there's so many different things that need to happen, but is the digital transformation of all the enterprises. You know, I like to provide examples. You can pick anyone. You can pick any vertical. If you pick energy, you pick retail, you can pick health care, and you see how those industries are being transformed by technology. And like I like to provide some simple examples. Energy: you had a company doing generation, you have a company doing transmission, then you have distribution, you have a centralized generation. Now you go to a home, you have solar, you have an EV, you plug in that, you have a whole different energy management solution. They needed to become intelligent. It's not about just connecting a meter. You go all the way to retail. Every shop now is an e-commerce. You need to manage inventory at the shelf as what you see with smart cameras. So we think that that's another big trend of digital transformation of enterprises. So those are the things that are happening right now and that's why we are optimistic about actually the semiconductor industry and the role of technology, especially at the edge.

是的。实际上，看到高通处于这么多趋势的中心，真是令人瞩目。你知道，有这么多都是大趋势，对吧？你知道，你提到了四五个真正巨大的趋势，对吧？现在，就 AI 本身而言，好吧，当然它在过去 12 到 18 个月里占据了社区、世界的意识。那么你对 AI 以及它将如何塑造，尤其是消费者体验，有什么看法？

Yes. Actually, it's remarkable to see how many trends Qualcomm is in the middle of. You know, there are so many of these are all big trends, right? You know, you mentioned four or five really massive trends, right? Now, in terms of AI itself, okay, that of course it has taken the community, you know, the consciousness of the world over the last, you know, 12 to 18 months. So what's your view on AI and how it will shape, you know, especially consumer experience, right?

是的，我们对此非常有主见。我认为，对于像高通这样的公司以及我们的许多合作伙伴来说，令人兴奋的是，正如我之前所说，AI 现在正与云端并行发展。我认为设备现在有两个非常不同的机会。一个机会是我们所谈论的，我想我们甚至写了一些关于这个的东西，那就是混合 AI。那就是通过运行、利用边缘设备上存在的计算能力来卸载数据中心，或者给数据中心一个先机，甚至能够结合工作，因为你在设备上做的事情，无论是你的 PC，还是你的手机，它们对你试图做的事情有更高的精度，因为它们是你周围的上下文。所以我们看到了云上发生的事情与设备合作的大机会。另一件事是设备上的 AI 正在以完全不同的方式演变，为用户提供完全不同的用例。所以我想我们实际上有一个视频来展示一些你在 AI 上看到的用例，例如在设备上。但让我带你了解一些例子。想想这个 AI 现在无处不在运行。AI 只是在运行。你开始看到它变为现实。现在，我们可以看到手机谈论我有这个功能，那个功能。但令人着迷的是，我们过去看到大约 10 个用例。现在人们在谈论数百个。它将演变成数千个。但你应该这样想，AI 试图预测你的生活移动。它只是在运行。简单如你在文本上写的东西可以成为模型的提示。

Yes, we're very opinionated on this. And what is exciting, I think, for a company like Qualcomm and for many of our partners is I think what I said before, AI is developing now parallel to the cloud. And I think the devices they have now two very distinct opportunities. One opportunity is what we talk about it. I think we even wrote a number of things about this, which is hybrid AI. And that is the ability to offload the data center by running, leveraging the computing power that exists on the devices at the edge, or giving the data center a head start, or even to be able to work in combination because things that you do on the device, whether it's your PC, whether it's your phone, they have a lot more precision about what you're trying to do because they're context around you. So we see this huge opportunity of what is happening on the cloud to work in partnership with the device. The other thing is the AI on device is evolving completely different with completely different use cases for the user. So I think we actually had a video to show about some of the use cases that you see happen on AI, for example, on device. But let me walk you through some of those examples. Think about this AI now running pervasively. The AI is just running. And you started to see that come to life. Now, we can see phones talking about I have this feature, this feature. But it was fascinating is we used to see like 10 use cases. Now people are talking about hundreds. It's going to evolve to thousands. But the way you should think about this is that the AI is trying to predict your life moving. It's just running. Something as simple as you text what you write on the text could be a prompt for the model.

模型会说：“好的，我没什么可做的。哦，我有事可做了。我可能需要问你愿不愿意分享这些数据。我可能需要帮你找到这份文档。你正在谈论它。你可能想叫 Uber。我可能需要你导航。我可能想做不同的事情。”这是助手尝试预测你行为的一种方式。但还有其他事情，因为 AI 正在无处不在运行。我喜欢它的一点是，既有我提到的混合方面，也有基于对你独一无二的东西、你自己的私有数据、所需可靠性的构建方面。我可以提供许多不同的例子。

And the model says, 'Okay, there's nothing for me to do. Oh, there's something for me to do. I may need to ask you if you want to share this data. I may need to find this document for you. You're talking about it. You may want to call an Uber. I may want you to do a navigation. I may want to do different things.' That's one way an assistant tries to predict your movement. But there are other things as the AI is running pervasively. And what I like about it is there's both the aspect that I mentioned about hybrid, but I also the aspect of building on things that are going to be unique to you, your own private data, the reliability that is required. And there are a number of different examples I can provide.

例如，汽车领域。我们与汽车客户进行的一次令人兴奋的讨论：即使你有一个模型，该模型在 Snapdragon 数字座舱中使用 RAG 运行。你实际上可以在该车型的整个数据库上训练该模型。然后你有一个问题，它远远超出了拥有一个数字操作手册的能力，该手册能解释你的问题并告诉你它在手册中的位置，甚至可以说这是你的车里发生的情况，这是你可以采取的行动。如果你想预约，我们会为你做。就像我们做的一件事，我们有一些有趣的演示，将生成式 AI 模型与在汽车数据库上训练的模型相结合。然后我们集成了 Salesforce，它会立即连接到 OEM 公司，告诉你预约的可用性并采取行动。你可以考虑这一点；导航是另一个例子。

For example, automotive. One of the exciting discussions we have with our automotive customers: even if you have a model, the model gets running in the Snapdragon digital cockpit using RAG. You can actually train that model on the entire database of that vehicle model. And then you have a question that goes way beyond having the ability to have a digital operating manual that interprets your question and tells you where it is in the manual, even to say this was happening in your car, this is the action you can take. If you want to schedule an appointment, we're going to do it for you. Like one of the things we did, we had some interesting demos combining a gen AI model with the training of the model on the database of the car. And then we had the integration of Salesforce that will actually immediately connect to the OEM company and tell you availability of appointment and take action and work in that. You can think about that; navigation is another example.

但我想回到这个混合环境。想想这个：你在做我们通常谈论的文本方面的事情。我想给你一个图像方面的例子。假设你在做设计。你在设计 PowerPoint 或创建图片。你正在创建数字图片，你进入构思阶段，进行许多不同的交互。如果你们所有人都在其中一些事情上使用 AI，你可能会进行几次交互，只是为了看看你是否喜欢；你可以在设备上做到这一点，而不是在设备上运行模型。然后当你时间到了，说“我要去生产了”，然后你结合云来做。当你考虑混合时，提供服务的公司的经济状况会改变，而且当你自己进行交互时，查询会变得更精确一些。所以我们非常清楚地看到这是一个巨大的机会。这就是为什么我们正在构建能够在手机上运行数十亿模型参数的芯片，当然还有 AI PC、汽车等等。

But I like to go back to this hybrid environment. Think about this: you're doing something we usually talk about text. I want to give you maybe an example of image. Let's say you're doing a design. You're designing a PowerPoint or you're creating a picture. You're creating a digital picture and you go to your ideation phase and you're doing a lot of different interactions. If all of you use AI for some of those things, you'll probably do a couple interactions just to see if you like it; you can do that on the device rather than the model running on the device. And then when it's time for you to go, 'I'm going to go to production,' then you do it in combination with the cloud. The economics to the company that's providing the service change when you think about hybrid, as well as the query becomes a little more precise as you go to your interactions yourself. So we see that very clearly as a huge opportunity. That's why we're building chips that can do multiple billion model parameters running on phones, on of course the AI PC, automotive and so forth.

在整个 AI 对话中，我认为这会改变这一点；设备端 AI 将改变许多行业的计算。这就是为什么它使高通甚至能在个人计算行业这样成熟的行业占据一席之地。

And on that whole conversation of AI, I think it's going to change this; on-device AI is going to change the computing of many industries. And that's why it enables Qualcomm to even take a position on industries well established like the personal computing industry as an example.

是的，这非常令人兴奋。你知道，首先，有所有这些新闻说今年和明年将是 AI 手机，手机上会有所有这些应用。今年早些时候也有一些使用你们芯片的产品问世，对吧？在亚洲，这些手机发布了。我知道它们都使用高通芯片。所以韩国对这些新手机的反应非常热烈。所以看看手机行业将如何因 AI 而改变会很有趣。

Yeah, it's super exciting. You know, first of all, there's all these news that this year and next year will be the AI phone, all these applications on the phone. And there is some that came out also using your chips earlier in the year, right? In Asia, these phones came out. I know they're all using Qualcomm chips for that. So it's a super exciting response in Korea and all for these new phones. So it'll be interesting to see how the phone industry is going to change with AI.

这只是开始。这只是开始。我喜欢用这个例子，因为我见过。只是因为我们已经看到，特别是因为你在谈论移动，我举了汽车的例子，但因为你谈论移动，我们在那个世界生活了很多代。我看待现在发生的事情的方式，想想从功能手机，比如你的诺基亚手机、黑莓手机到智能手机的转变。你第一次看到智能手机时，你可能有 10 个或 15 个应用，现在我们有数十万个应用。你应该以同样的方式思考一些生成式 AI 用例。现在，我们只是开始。“哦，看这个。这是新的，我可以有这个应用”，我们开始看到这些用例。如果我们一年后有这样的对话，那对话会非常不同。

And it's just the beginning. This is just the beginning. I like to use this example because I've seen it. I just because we've seen especially because you're talking about mobile, I give an example on car, but you because you talk about mobile, we lived in that world for many generations. The way I see what's happening right now, think about the transition from your feature phone, let's say your Nokia phone, your BlackBerry phone to a smartphone. And the first time you saw the smartphone, you probably had 10 apps or 15 apps and now we have hundreds of thousands of apps. You should think about the same way about some of those gen AI use cases. Right now, we're just at the beginning. 'Oh, look at this. This is new and I can have this app' and we started to see those use cases. If we have this conversation 1 year from now, that conversation's going to be very different.

确实。实际上，我一个月前在韩国，这个产品出来了，这款使用你们芯片的 AI 手机，我问他们人们觉得最兴奋的是什么，结果是实时翻译，从一种语言到另一种语言。但还会有很多其他应用。但你最近也发布了这些令人兴奋的 PC 和笔记本电脑芯片。性能惊人，集成了调制解调器、CPU、GPU 和 AI。所以整个行业本身，我们称之为笔记本电脑或个人计算，正在经历大规模转型，而高通正处于其中，提供极其新颖的解决方案。所以，我知道这可能是你们的举措之一。那么你对这个超级令人兴奋的市场有什么看法？

Exactly. Actually, I was in Korea a month ago and this thing came out, this AI phone using your chips, and I asked them what's the most exciting thing people find, and it was like on-the-fly translation from one language to the other. But there'll be so many other applications. But you also recently launched these exciting chips for PCs, for laptops. It's like amazing performance integrated with modem and then CPU and GPU and AI. So that whole industry itself, let's call it laptop or personal computing, is going through massive transformation and Qualcomm is right in the middle of it providing extremely new solutions. So, and I know this is probably one of your initiatives. So what's your take on this super exciting market?

看，对我们来说，这是一个新市场。对我们来说，这是一个新市场。我们知道这是一个成熟的市场。有一个成熟但持续演进的市场。但我们对这个市场如此兴奋的原因，除了显而易见的一点——这是一个我们可以销售更多芯片并扩展进去的新市场——我们喜欢它是因为很少有像这样的机会，我们看到三个不同的顺风同时发生。这就是 Snapdragon X Elite 诞生的原因。我要在这里发表一个挑衅性的声明，但这是我们作为该领域新进入者的意图。我认为 Snapdragon X Elite 创造了一个机会，将性能领导地位恢复到 Windows 生态系统。我认为我们对 Snapdragon X Elite 感到无比兴奋，它基本上是当前笔记本电脑上存在的最快处理器。这就是其中一个顺风。它为行业创造了一个新的性能基准，并持续演进。第二个顺风是设备端 AI。以及普遍运行设备端 AI 的能力，特别是当你考虑带有 Copilot 的 Windows 生态系统时，我认为微软有非常雄心勃勃的计划，以及在设备上运行其中一些用例的能力。正如我们所说，设备上发生的全新功能。加上它如何与云中的 Copilot 协同工作。这是第二个顺风，我们有能力展示我们可以在每瓦性能方面做出不同的东西。

Look, for us, this is a new market. Like for us it's a new market. We know it's an established market. There's a market which is mature but continues to evolve. But the reason we're so excited about this market, besides the obvious thing which is a new market we can sell more silicon and expand into that market, we like it because there are few opportunities like this that we see three different tailwinds all happening at the same time. And that's what led to Snapdragon X Elite. I'm going to make a provocative statement here, but that's our intent as the new entrant into the space. I think Snapdragon X Elite creates an opportunity to restore the performance leadership to the Windows ecosystem. I think we are incredibly excited about Snapdragon X Elite, basically the fastest processor that exists for a laptop right now. And that is one of those tailwinds. It creates a new benchmark for performance in the industry, which continues to evolve. The second tailwind is on-device AI. And the ability to run on-device AI pervasively, especially when you think about the Windows ecosystem with the Copilot, which I think Microsoft has very ambitious plans, and the ability to run some of those use cases on the device as well. As we talk about it, completely new features that happen on the device. Plus how that works together with the Copilot in the cloud. That's the second tailwind, which we have the ability to show that we can do something different in terms of performance per watt.

我们能运行这个吗？如果能在手机上全天运行，那么也能在 PC 上全天运行而不影响电池续航。我认为第三点是移动端与 PC 的融合。这一点还没有被充分理解。让我用一个非常简单的方式来描述。今天，我相信我可以为在座的各位描述一个场景。你拿着手机，有些事情——也许你收到一封需要回复的邮件，而且会非常复杂。于是你说：“我要等到回办公室或回家。我有安静的时间，用键盘和屏幕来写这封又长又复杂的回复。”你不会在手机上做这件事。也许一些简单的事情你会在手机上做。人们有不同的偏好。像 Copilot 为你起草邮件这样的功能，在有手机的情况下反而更有用。所以，移动端与 PC 的融合将推动越来越多的端侧 AI 用例，这关乎可靠性、响应时间以及补充云端的能力。当我们审视这三点时，这就是我们设计 X Elite 的原因，我们对此持乐观态度。

Can we run this? If I can do it on the phone, running all day on the phone, we can do it running all day on a PC without compromising battery life. And I think number three is the convergence of mobile and PC. This is not well understood. Let me give you a very simple way to describe this. Today, I'm sure I'm going to describe a scenario for everybody here in the room. You have your phone and there are certain things—maybe you get an email that you need to respond to and it's going to be very complicated. So you say, "I'm going to wait until I get to my office or my home. I have a quiet moment with my keyboard and my screen. I'm going to write this long complicated response." You're not going to do it on your phone. Maybe some simple things you do on your phone. People have different preferences. Things like the co-pilot drafting an email for you—it's way more useful in the situation where you have a phone, for example. So what's going to happen is this convergence of mobile and PC is going to drive more and more of those on-device AI use cases, which is about reliability, response time, and the ability to complement the cloud. So when we look at those three things, that's the reason we went and designed X Elite and we're optimistic about it.

是的，这真的很了不起。有几件事。一件有趣的是，微软如此大力支持，对吧？因为软件生态系统必须跟上。然后，我和所有这些——我相信你对这些趋势再熟悉不过了。但有一点要指出，我和所有这些谈论这些模型的人交流过，对吧？LLM 模型会怎样？它们会变大还是变小？很多 AI 公司认为模型可以做得更小，但仍然有足够的保真度和准确性，尤其是对于端侧 AI。因为 X Elite 已经可以运行巨大的模型，但模型不需要像 1 万亿或 6000 亿参数那么大。所以这将是非常强大 AI 的自然融合，可以直接在本地计算机上运行，对吧？

Yeah, it's truly remarkable. I mean, a couple of things. One thing is interesting to see—Microsoft supporting it so vigorously, right? Because the software ecosystem has to come with it. And then, I was talking to all these—I'm sure you're more than familiar with all these trends. But one thing to point out is talking to all these people about these models, right? What's going to happen to LLM models? Will they get bigger? Will they get smaller? And a lot of these AI companies think that the models can be made smaller but still have enough fidelity and accuracy, especially for on-device AI. Because X Elite can already do huge models, but the models don't need to be as big as 1 trillion or 600 billion. So it'll be a natural merger of very, very powerful AI which can just run locally on the computer, right?

是的，我想说的是，有很多技术正在开发中，我们的研究团队在这方面投入了大量时间，包括自动化。但有很多技术正在为这些模型的量化而创造，你可以实际上量化其中一些模型而不损失准确性，并创建一个机制，最终就像你在手机上使用应用一样。就像你有一个应用，然后最终你会从应用商店收到新版本的应用推送。这将成为常态。你将拥有自动量化的模型。它们现在在设备上运行。在某些情况下，你可以根据想要运行的设备来决定模型的大小。然后你能够不断更新，我认为这就是这些事物的发展方式。我喜欢用汽车的例子来描述这一点。尽管我们已经看到一些模型在设备上运行并保持准确性的惊人结果，但我想给你一个不同的思路。当你想到汽车，我们一直与许多不同公司合作使用他们的技术栈，但我们也有自己的辅助驾驶和自动驾驶技术栈。例如，我们正在与宝马联合开发技术栈。我们还有许多其他公司合作利用我们的技术栈。在汽车领域，你可以有不同的自动驾驶应用。你可以决定你要一级、二级、二级加、三级、三级加，你可以从例如很多人想要的高速公路自动驾驶到完全自动驾驶。有不同的应用。所有这些不同的应用都有市场。所以我们开始意识到，端侧 AI 的机会在于训练有素、最新且量化的模型，你将能够拥有非常精确的模型，但也有很多针对特定应用的不同模型。人们说：“我不需要所有这些。我只想要这个模型在这个数据上训练，非常快、非常可靠，数据隐私，而且我想非常经济地做到这一点。”所以这就是为什么我们看到，例如，GenAI 的工业案例也在以非常快的速度发展。

Yeah, I think what I will say about this is there's a lot of technology being developed and we have been spending—when our research team—a lot of time on this, including automation. But there's a lot of technologies that are being created for quantization of those models in a way that you can actually quantize some of those models, not lose accuracy, and create a mechanism that eventually is going to be like what you do with an app on your phone. Like you have an app and then eventually you get pushed from the app store a new version of the app. And this is going to be like business as usual. You're going to have models that get automatically quantized. They are now running in devices. In some cases you can decide how big you want the model to be depending on what device you want it to run. And then you have the ability to constantly update and I think that's how those things are going to evolve. And I like to use an auto example to kind of describe this. Even though we have been seeing incredible results of some of those models when they run on device and maintaining accuracy, I want to give you maybe a different thought process. When you think about auto and we have been partnering with many different companies using their stack, but we also have our own stack for assisted driving and autonomy. As an example, we're doing jointly with BMW developing the stack. We have a number of other companies we're working together leveraging our stack. And in automotive you can have different applications of autonomy. You can decide that you're going to have level one, level two, two plus, three, three plus, and you can go from, for example, a lot of people want to do highway autopilot to all the way to fully autonomous self-driving. And there are different applications. There's a market for all of those different applications. So what we're starting to realize is the opportunity for on-device AI goes to model that very well trained and up-to-date and quantized, and you're going to have the ability to have that model to be very precise, but also you have a lot of different models for specific applications. People said, "I don't need all of this. I just want this model to be trained on this data and to do this very fast, very reliable, with privacy on the data, and I want it to do that very economically." So that's why we're even seeing, for example, industrial cases for GenAI evolving at a very fast pace as well.

确实。说到汽车，实际上，首先，我们都喜欢的东西。我看到你们现在在汽车领域有巨大的积压订单和渗透率。几乎所有的——当我在世界各地旅行时，几乎所有的汽车公司都在与高通合作。哦，谢谢。这是个好消息。是的，这与几年前相比是个巨大的变化。老实说，这是汽车行业的一次显著转变，所以我认为看到这种大规模的——那么，你如何看待——我的意思是，你在整个汽车行业有前排座位吗？他们传统上比 PC 或数据中心慢。所以，你现在对汽车有什么看法？他们会一样快吗，或者是什么——

Exactly. And talking about autos, actually, first of all, something we both love. I saw that you have massive backlog and penetration in the auto sector now. Almost all—when I travel around the world, almost all auto companies are working with Qualcomm. Oh, thank you. That's great news. Yeah, this is like a big change from a few years ago. Honestly, it's a remarkable transformation of the auto industry, and so I think it's good to see this kind of massive— So, but how do you see the—I mean, do you have a front row seat on the entire auto sector, and they have traditionally been slower than like PC or data center. So, what's your take on the auto now? Are they going to go as fast, or what is the—

是的。看，这是个好问题。我喜欢谈论这个。但然后我可能从评论你所说的开始。当我回顾高通的历史和旅程时，我认为汽车是我们公司如何受益并在非常短时间内真正解决市场的最佳例子。我们受益于我们从移动传统中发展出的一些核心能力。很简单。当你看看今天的汽车，看看完全沉浸式数字座舱的整个计算体验。当你看看现在能够连接到云端，这个与云端不可分割的数字孪生，你看看运行自动驾驶栈所需的计算能力，包括 AI。你不能把服务器放在汽车后备箱里，因为那会减少续航里程。所以你必须有一个解决方案，具有服务器类型的计算能力，但具有移动类型的能效，我认为这就是高通的秘密武器，以及为汽车公司构建同时包含芯片和软件的解决方案的能力。现在你问题的第二部分实际上是最令人兴奋的部分。

Yeah. Look, it's a great question. I love talking about this. But then I'll start maybe commenting on what you said. As I go back and I talk about the history and the journey of Qualcomm, I think the automotive is the best possible example of how we in the company benefited and we actually addressed the market in a very short period of time. We benefited from some of the core competencies we developed coming from a mobile heritage. It's very simple. When you look at a car today and you look at the whole computing experience of a fully immersive digital cockpit. When you look at the ability to be now connected to the cloud to this digital twin inseparable from the cloud and you look at the computing power including AI that is required to run autonomy stacks. You cannot put a server on a trunk of a car because you're taking away from the range. So you have to have a solution that has the server type computing power but it has the mobile type power efficiency and I think that has been the Qualcomm secret sauce and the ability to build a solution with both silicon and software for the car companies. Now the second part of your question is actually the most exciting part.

汽车公司正在经历大规模的技术转型，以及客户群的转型。今天的汽车消费者与过去的汽车消费者截然不同。公司需要改变，它们需要成为数字公司。未来非常成功的汽车公司将是科技公司。变化在于：这个行业过去注重长期规划，而现在衡量标准是功能开发的速度。公司的价值将更多取决于你的车在拥有期间能提升多少。假设你买了一辆车，打算开 5 年或 7 年，我能为这辆车增加多少价值，而不是平台开发有多少个不同阶段。公司需要敏捷。我认为这意味着这些公司需要合作伙伴关系，并且需要拥抱这种新客户。甚至包括它们与客户的互动方式——我们与 Salesforce 等公司建立了非常成功且自然的合作伙伴关系，例如，现在已内置于高通骁龙数字座舱中。汽车公司过去没有与客户的直接关系，那是通过经销商完成的。但现在通过车内的那些屏幕，汽车公司始终与你同在。你的账户是什么？你在做什么？你想做的事情是什么？这些正在改变动态，我认为这恰恰为技术创造了机会。

The car company is going to a massive amount of transformation for technology as well as a transformation of its customer base. Consumers of cars today are very different than consumers of cars of yesterday. And the companies need to change. They need to become digital companies. Car companies that are going to be very successful are tech companies. And what a change is: you have an industry that was geared towards planning things over a long period of time, and now their measure is by the speed of feature development. Companies are going to be more valuable by how much your car gets better during the ownership. Let's say you're going to buy a car and you're going to keep it for 5 years or 7 years, how much more can I add value to the car versus how many different stages of the platform development. Companies need to be agile. And I think that means that those companies need partnerships and they need to embrace this new customer. Even how they interact with their customers—the reason we had a very successful and natural partnership with companies like Salesforce, for example, that is now built into the Qualcomm Snapdragon digital cockpit. The car company didn't have a direct relationship with the customer; that was done by the dealership. But now through those screens of the car, the car company is there with you all the time. What's your account? What are you doing? And what are the things you want to do? Those things are changing the dynamic and I think that exactly creates opportunity for technology.

是的，绝对如此。然后，你知道，我们谈了很多大趋势，对吧？比如高通的转型、AI、笔记本电脑、桌面计算，那是巨大的转型，还有汽车。但当然，你知道，高通一直处于移动领域的核心，对吧？所以如果我们处于 5G 时代，并且有所有这些关于 5G 到 6G 迁移的讨论。那么你如何看待这一点？

Yeah, absolutely. And then, you know, we talked about a lot of big trends, right? You know, talk about Qualcomm transformation, you know, AI, laptop, desktop computing, that's a huge transformation, automotive. But of course, you know, Qualcomm has always been central to the mobile, right? So if we are in the 5G and there's all this talk of 5G to 6G migration. So how do you see that?

是的。嗯，这将会发生。5G 之后，6G 肯定会到来。每十年发生一次——不是双关语，但移动行业的节奏是每 10 年有一代新的无线技术。所以 6G 计划在十年末，大约 2030 年。我们将看到试验的到来。我认为我们对我们在标准方面做出的许多贡献感到非常高兴，标准制定正在按计划进行。我喜欢以稍微不同的方式回答这个问题。每一代无线技术——这就是我们在高通思考的方式。尽管有所有这些计算和 AI，我们仍然是一家无线公司，做我们在蜂窝领域需要做的事情。但我们把每一代无线技术看作一个需要解决的问题。所以，2G 从模拟到数字的问题是，我能确保地球上的每个人都有手机吗？3G 的问题是，我能把手机连接到互联网吗？4G 的问题是，我能把宽带带到手机上吗？5G 的问题是，我将要进入一个无限数据速率的环境，我需要支持关键任务应用。我需要它能够从尽力而为变为有保证的连接，这就是 5G 带来的很多东西。6G，这是一个新概念。这是一个全新的范式，关乎网络本身是活的，并与网络中连接的一切相结合。那就是网络，你为网络增加了感知能力。这就是我们将看到 AI 的惊人体现的地方，因为网络将变得自主，你将看到大量感知能力在网络本身中在广域内发生。想想广域感知网络。这是描述它的一种方式。还有很多可以谈论的，但我认为我们对 6G 的兴奋程度不亚于我们对每一代新无线技术的兴奋。

Yes. Well, it's going to happen. Definitely after 5G, 6G is coming. It happens every decade—no pun intended, but the cadence of the mobile industry is every 10 years you have a new generation of wireless. So 6G is scheduled to the end of the decade, about 2030. And we're going to see trials coming in. I think we're very happy with a number of the contributions we have made to standards, how the standards development is on track. And I like to answer this question a little bit different. Every generation of wireless—and that's how we thought about this in Qualcomm. And with all this computing and AI, we're still a wireless company doing what we need to do in cellular. But we look at every generation of wireless with a problem to be solved. So, the problem of 2G from analog to digital is, can I make sure everybody on Earth has a phone? Problem of 3G is, can I connect that phone to the internet? Problem of 4G is, can I bring broadband to the phone? The problem of 5G is, I'm going to move to an environment that is unlimited data rates, and I need to support mission-critical applications. I need it to be able to go from best effort to a guaranteed connectivity, and that's a lot of the things that came with 5G. 6G, it's a new concept. It's a whole new paradigm, which is about the network is in itself is alive and combining with everything that is connected in the network. That's the network, and you add sensing capabilities to the network. And that's where we're going to see an incredible manifestation of AI as the network's going to become autonomous, and you're going to see a lot of sensing capabilities happening within the network itself in the wide area. Think about the wide area sensing network. That's one way to describe it. There's a lot more to talk about it, but I think we are as excited about 6G as we've been excited about every new generation of wireless.

哇，太棒了。那么，Cristiano，在个人方面，或者你的领导哲学，你知道，因为高通是一家巨大的公司，对吧？而且在世界各地工作，你涉足如此多不同的技术，从移动 to 汽车到 AI 再到桌面计算。那么，你对观众、工程师和管理者的建议有什么想法？你经营这样一家重要科技公司的领导哲学是什么？

Wow, great. So, Cristiano, on the personal side, or your leadership philosophy, you know, because Qualcomm is a huge company, right? And working in all parts of the world, you're involved in so many different technologies from mobile to automotive to AI to desktop computing. So, what's your thoughts on advice for people in the audience, engineers, and managers? What's your leadership philosophy to run such an important technology company?

瞧，我不知道我是否能提供好的建议。我就从前提开始吧：既然我免费提供建议，也许它没那么好。但我会告诉你一些在我自己的旅程中对我是有效的，并且对公司和我们公司的文化也是有效的事情。第一个是我总是用完全相同的答案回答这个问题，因为我一直这么觉得。而且我认为它确实对我的职业生涯有帮助。我认为其中之一是让你周围都是比你更优秀的人。你必须让你的团队中有一些人，他们做的事情比你自己能做的更好。我认为管理者的工作是真正指引人们朝着正确的方向前进，并让团队能力的集合发挥作用。我认为另一件事是，在高通，我们非常自豪拥有协作的文化。我们公司内部没有很多相互争斗的派系。我认为大家同赢同输。其他事情：我总是过度沟通。我认为我做的事情我不知道是好主意还是坏主意，但比如当我们谈论战略时，我们确保每个人都理解战略。有时我们甚至把一页纸的战略分发给每一位员工。这样你可以思考自己如何融入那个战略，以及在你做的每件事中如何与公司保持一致，因为来自每个人的想法和能力的力量，尤其是如果你有一个有很多聪明人的组织，它是巨大的。还有几件事。即使我们做得很好，我们内部也倾向于认为我们做得不好。所以在我们行业变得自满是非常危险的。我经常在采访中提供这个例子，我会重复它，只要我在高通，我就会说那就是例子。我告诉我们的团队，我们处于角斗士行业。我不知道人们是否听过我谈论这个。就像你去斗兽场，有三种可能的结果，尤其是在科技世界。你总是必须重塑自己。所以你可以赢，你可以输，或者你们都输，对吧？所以这就是结果。

Look, I don't know if I can provide good advice. I will just start with the premise: since I'm providing advice for free, maybe it's not that good. But I'll tell you some of the things that have been valid for me in my own journey, and have been valid for the company and the culture we have in the company. First one is one that I always answer this question with exactly the same answer because that's how I always felt about it. And I think it did help me in my career. I think one is surround yourself with people that are better than you are. You have to have people working in your team that are going to do things better than you'll be able to do yourself. I think a manager has the job of actually pointing people in the right direction and getting the collection of the capabilities of the team to perform. I think the other thing is in Qualcomm, we take a lot of pride in having a culture of collaboration. We don't have a lot of warring clans within the company. I think everybody wins and loses together. Other things: I always over-communicate. I think I do things which I don't know if it's a good idea or bad idea, but like when we talk about the strategy, we make sure that everybody understands the strategy. Sometimes we even distribute a strategy on a page to every single employee. So you can kind of think about yourself how you fit within that strategy and how in everything you do, you align with the company because the power of ideas and capabilities that come from every single individual, especially if you have an organization that has a lot of smart people, it's enormous. A couple other things. Even when we're doing well, we tend to think inside that we're not doing well. So it is very dangerous in our industry to become complacent. I have often in interviews provided this example, and I will repeat it and as long as I'm in Qualcomm, I'm going to say that is the example. I tell our teams that we are in the gladiator business. I don't know if people have heard me talking about this. Like you go to the coliseum, there are three possible outcomes, especially in the tech world. You have to always reinvent yourself anyway. So you can win, you can lose, or you both lose, right? So that's kind of the outcome.

如果你赢了，你得到的唯一东西就是再进一次斗兽场的资格，对吧？就是这样。事情就是这样。我认为我们必须始终推动自己、重塑自己。而高通——希望当我们回顾公司历史时——我热爱这家公司，我非常偏心，我从 1995 年就在那里，当时是以工程师身份加入的。我认为公司经历过很多次濒死体验，而我们重塑了自己，我认为我们把这当作文化的一部分、力量的一部分。

And if you win, the only thing you get is the right to go to the Coliseum one more time, right? So that's it. That's what happens. And I think we have to always be pushing ourselves, reinventing ourselves. And I think Qualcomm — hopefully, when we look at the history of the company — I love the company, I'm very biased, I've been there since 1995, I joined as an engineer. I think the company has been through a lot of near-death experiences, and we reinvented ourselves, and I think we take that as part of our culture, part of our strength.

哦，太棒了，Cristiano。我没听过斗兽场那个比喻，但这个说法很好。感谢你精彩的分享。

Oh, that's great, Cristiano. I hadn't heard the gladiator one, but that's a good one. Well, thank you for your insightful comments.

哦，谢谢。也感谢与 Cadence 的出色合作。你知道，这种广泛的合作。你们的工程团队非常出色。能向你和你的工程团队学习很棒。他们在全世界都是世界级的。所以，谢谢你们。也感谢你来到 Cadence。

Oh, thank you. Also, thank you for the great partnership with Cadence. You know, this broad-ranging partnership. You have amazing engineering teams. And it's great to learn from you and your engineering team. They're world-class all over the world. So, thank you for that. And thank you for being here at Cadence.

谢谢。感谢你给我这个机会。非常感谢。谢谢。哇！

Thank you. Thank you for the opportunity. Thank you so much. Thank you. Woo!