高通 CEO 克里斯蒂亚诺·安蒙谈 6G 时代与下一代 AI 设备

每一辆车、每一辆公交车、每一辆自行车、每一个行人，一切都在移动。你将创建一个整个世界的数字孪生。

Every car, every bus, every bicycle, every pedestrian, everything is moving. You're going to create a digital twin of the entire world.

欢迎来到可能的边缘。

Welcome to the edge of possible.

你可能没听说过高通，但你肯定用过他们的产品。许多主要的智能设备制造商都使用高通的技术，从我们手机中的物理芯片到连接它们的 4G、5G 以及即将到来的 6G 网络。但是，在快如闪电的科技行业，今天的尖端技术明天就可能过时。高通 CEO 克里斯蒂亚诺·安蒙准备孤注一掷以保持领先。

You may not have heard of Qualcomm, but you have definitely used their products. Many major smart device manufacturers use Qualcomm's technology, ranging from physical chips in our phones to the 4G, 5G, and soon 6G networks that connect them. But, in the lightning-fast tech industry, what's cutting edge today can become obsolete tomorrow. Qualcomm's CEO, Cristiano Amon, is prepared to bet the farm to stay ahead.

如果你看看移动行业的公司，每一代无线技术，都有一个巨大的公司墓地，而我们仍然在这里。

If you look about companies in the mobile industry, to every generation of wireless, there was a big cemetery of companies, and we're still here.

克里斯蒂亚诺将这家 40 年公司的成功部分归功于其重塑文化。在克里斯蒂亚诺担任 CEO 的 5 年里，公司已经从高度依赖智能手机，发展到在汽车、机器人和可穿戴 AI 领域开展工作。

Cristiano credits the 40-year-old company's success, in part, to its culture of reinvention. In Cristiano's 5 years as CEO, the company has evolved from one that's highly smartphone-dependent to one that's working in automotive, robotics, and wearable AI.

珠宝、别针、吊坠，你佩戴的东西，它可以连接到智能体。

Jewelry, pins, pendants, things that you wear, and it can connect it to an agent.

我在高通位于圣地亚哥的高科技总部会见了克里斯蒂亚诺。

I met with Cristiano at Qualcomm's high-tech headquarters in San Diego.

机器人来了。

Robot incoming.

是的。

Yeah.

他给了我们一个小警告。

He's giving us a little warning.

来讨论克里斯蒂亚诺和高通如何打造可能成为我们智能手机之后的下一个主要设备。我是艾莉森·尚特尔，这是《财富》500 强巨头与行业颠覆者。赞助商信息之后，我们将回到克里斯蒂亚诺的访谈。

To discuss how Cristiano and Qualcomm are building what could be our next primary device after the smartphone. I'm Alison Shontell, and this is Fortune 500 Titans and Disruptors of Industry. We'll be back with Cristiano after a message from our sponsor.

《财富》与德勤美国 CEO 杰森·吉尔扎达斯进行了交谈，讨论理解量子计算以及他如何看待其改变行业。

Fortune spoke with Deloitte US CEO, Jason Girzadas, on understanding quantum computing and how he sees it transforming industries.

量子计算在研究圈已经讨论了一段时间，当然也受到商业界的密切关注，但它本质上是一种不同的计算范式，使用物理学原理而不是数学来驱动计算成果。我认为真正的用途最终将围绕非常复杂的优化场景、进一步增强机器学习规模，以及可能与众多不同商业应用相关的非常复杂的模拟。

Quantum computing has been a topic for some time in research circles, certainly closely watched by business, but it's fundamentally a different computing paradigm that uses the principles of physics instead of mathematics to drive the computing outcomes. And I think the real uses will be ultimately around very complex optimization scenarios, further enhancements to scaling machine learning, and also very complicated simulations that could be relevant to a whole host of different business applications.

领导者应该采取哪些步骤来为量子计算准备他们的组织？

What steps should leaders take to prepare their organizations for quantum computing?

现在真正重要的是就绪规划，让组织了解其影响，理解需要哪些技能组合，需要哪些类型的网络安全协议才能在企业环境中可行，并开始思考与优化和模拟密切相关的用例类型。

It's really about readiness planning right now and preparing an organization to understand the implications, understanding what skill sets would be required, what type of cybersecurity protocols would need to be in place to make it viable in an enterprise context, and begin to think about the types of use cases that would be very germane around optimization and simulation.

克里斯蒂亚诺，非常感谢你接受《财富》500 强巨头与颠覆者播客的采访。你坐在 AI 宇宙的中心。我不确定人们是否意识到高通在他们使用的几乎所有设备中嵌入了多少。你能给我们一些规模和范围吗？你们在《财富》500 强中排名第 117 位，大约有 50 亿台设备由高通芯片驱动。

Cristiano, thank you so much for sitting with us for the Fortune 500 Titans and Disruptors podcast. You are sitting in the center of the universe of AI. I'm not sure that people realize quite how much Qualcomm is embedded in pretty much everything that they use from a device standpoint. Can you maybe just give us some size and scale to it? You're 117th on the Fortune 500, about 5 billion devices are powered by Qualcomm chips.

而且还在增加。

And counting.

而且还在增加。那么，请给我们介绍一下你们在这里所做事情的规模。

And counting. So, kind of give us some scale to what you're doing here.

是的。顺便说一句，感谢你抽出时间。很高兴在这里与你交谈。看，高通是一家非常独特的公司。在人们听说骁龙之前，我们常说高通可能是没人知道的最大的公司。我认为当我们开始时，就是创造现在每个人都在使用的基础技术。对于无线，我们参与了每一代无线技术。但当你看看无线革命发生了什么，我认为你现在把我们的智能手机视为最不可分割的设备，我们的技术基本上被移动行业的规模推动到许多不同的地方。我认为今天，高通不仅在智能手机领域，我认为我们进入了 PC 领域。我们进入了可穿戴领域，那些未来的个人 AI 设备。我们在汽车领域开展非常大规模的业务，在一些人的汽车中，并且正在进入工业领域，一直到未来的机器人和数据中心。我认为我们非常高兴。我们有一个非常大规模的半导体业务。我们大多数人是在公司的移动业务中成长起来的，但我们开始看到技术变得与许多其他行业相关，我认为这就是我们现在正在做的事情。

Yeah. By the way, thank you for taking the time. Very happy to be here talking to you. Look, Qualcomm's a very unique company. We used to say before people got to hear about Snapdragon, we used to say Qualcomm's probably the biggest company nobody knows about. And I think when we started, it was about creating the fundamental technologies that everybody uses now. I think for wireless, we have been into every generation of wireless. But as you look at what happened with the wireless revolution, and I think you look at our smartphone right now as most inseparable device, our technology basically propelled by the scale of the mobile industry it to go in many different places. And I think in today, I think Qualcomm not only is in the smartphones, I think we enter into the PC space. We are into the wearable space, those future personal AI devices. We're going very large-scale business on automotive, in some people's cars, and it's going to industry and industrial, and all the way to what they're going to be the future robotics and data center. And I think we're very happy. We have a very large-scale semiconductor business. I think most of us grew up in the company within mobile, but we started to see the technology became relevant to so many other industries, and I think that's what we're doing right now.

是的，我认为你们有点像我们每天使用的技术的大脑，而且越来越如此，无论是我们驾驶的汽车，还是我们使用的手机，或者我们使用的电脑，你们都在那里，嵌入在设备中。我们刚刚出版了新一期的《财富》杂志，封面是 AI 临界点。感觉我们正处于新的 AI 浪潮中，也许它还没有真正达到消费者峰值，人们仍然在想，这是什么，我如何使用它，它意味着什么？但当然，从 AI 能力的角度来看，感觉我们正处于一条垂直上升的直线。你同意吗？发生了什么，也许你可以给我们一些大局观，过去几个月发生了什么使它如此？

Yeah, I think of you all as sort of like the brains of the tech that we use every day, and increasingly so, whether it's the cars we drive, or the phones that we use, or the computers we're on, you all are there, embedded in the devices. We just had a new issue of Fortune come out, and our cover is the AI tipping point. It feels like we're very much in a new wave of AI, where maybe it hasn't really hit peak consumer yet, and people are still kind of wondering on the whole, what is this, and how do I use it, and what's it mean? But certainly from the standpoint of the capabilities of AI, it feels like we're at this just straight vertical line up. Would you agree with that? And what's happened, maybe you could give us like some big picture of what has happened the last few months that has made it so?

100%。我认为我们一直在为这一刻做准备，我们看待这个问题的方式与其他一些公司略有不同。你知道，我们制造了很多人们每天使用的设备，设备是人们与 AI 互动的方式。AI 的某些方面将会改变，你知道，我们实际上如何思考计算。我认为我们看待这个问题的方式是，我们所做的一切都将处理某种形式的 AI，或者连接到云端来做大量的 AI。这就是我们如何走向计算的过渡，我们不再担心操作系统，我们不再担心应用程序。我认为导致这个临界点的原因是，在 2026 年，我们一直说这将是智能体之年。现在你有智能体，可以让 AI 有用来做许多许多不同的事情。我将把这一点分解为不同方面，当你思考 AI 的演变时。我想我们都开始听说 ChatGPT，有一个聊天框，你去问一个问题。但然后你也有这个令人难以置信的发展，AI 如何成为下一个更高层次的语言，你可以用来写代码。但然后还有其他方面。

100%. I think we have been preparing for this point, and we look at this a little bit different than some of the other companies. You know, we build a lot of devices that people use every day, and devices how people are going to interact with AI. And there are aspects of AI that are going to change, you know, how we actually think about computing in general. I think the way we look into this is everything we do, it's going to be processing some form of AI, or connecting to the cloud to do a lot of AI. This is how we're going to go to this transition of computing that we don't worry about OSs, we don't worry about applications. And I think what is causing this tipping point, it's that in this year 2026 we've been saying that it's going to be the year of agents. It's now you have agents that can, you know, make AI useful to do many many different things. And I'm going to break this down into different aspects when you think about the evolution of AI. I think we all started heard about ChatGPT, there's a chat box, you go and ask a question. But then you also have this incredible development about how AI just become the next higher level language that you can use to do to write code. But then there other aspects.

一个方面是，我们从一开始就在说，现在因为大语言模型、因为大型视觉模型，AI 能像我们理解世界的方式理解世界，用自然语言与我们交流，这就在人类和计算机之间创造了一个新的用户界面。现在，一旦你把所有这些整合到一个智能体中，你可以告诉智能体你想要什么，这将从根本上改变我们与这些设备的交互方式。随之而来的是，我认为人们开始看到巨大的规模。我认为人们刚刚开始理解现在有各种智能体。智能体会进入计算机为你做事，会进入云端为你做事。我认为这就是我们开始在消费领域和企业领域的日常事务中看到大规模扩展的方式。我们对此感到兴奋，因为它将创造一个新设备的巨大周期，这些设备将是智能的、聪明的，并将我们与那些智能体连接起来。

One aspect is, and we've been saying this since the very beginning, which is now because of large language models, because of large visual models, the AI understands the world the way we understand the world, communicates with us the way we communicate with natural language, and that creates a new user interface between humans and computers. And now once you have all of those things put together in an agent that you can tell the agent what you want, that's going to fundamentally change how we interact with those devices. And with that, I think guys start to get a lot of scale. I think people are just starting to understand there are agents for everything now. There are agents going to go into the computer and do things for you, it's going to go to the cloud and do things for you. And I think that's how we're going to start to see massive amount of scale into everyday things that we do on the consumer space, on the enterprise space. And we're excited about that because it's going to create a big cycle of new devices that are going to be intelligent, they're going to be smart, and they'll be connecting us with those agents.

AI 驱动的设备通常需要极快的网络才能充分发挥作用。这是高通和 Amon 花了几十年构建的东西。高通技术推动了 3G、4G 和 5G 蜂窝网络的推出。Amon 于 1995 年首次加入高通担任工程师，指导了高通的全球 5G 推广。如今，Amon 正着眼于 6G，他相信 6G 将提供全息图、协作机器人和无人驾驶汽车等应用所需的效率和性能。你在 MWC（世界移动通信大会）上关于我们如何从 2G 发展到现在的 6G 以及你的生态系统的对话让我印象深刻。我想知道你是否能带我们回顾一下。你在高通很长时间了，30 年。你实际上是一位“回旋镖”CEO。你曾在这里，离开，又回来，一步步晋升。但请带我了解技术上的转变，以及 6G 对你我这样的人意味着什么。

AI-powered devices often demand lightning-fast networks to operate fully. Something that Qualcomm and Amon have spent decades building. Qualcomm technology powered the launch of the 3G, 4G, and 5G cellular networks. Amon, who first joined Qualcomm as an engineer in 1995, directed Qualcomm's global 5G rollout. Today, Amon is eyeing 6G, which he believes will deliver the efficiency and performance required by applications like holograms, collaborative robots, and driverless cars. I was really struck by your MWC, your World Mobile Congress conversation about how we've gotten from 2G to now 6G and ecosystem of you. I was wondering if you could kind of walk us through that. You've been at Qualcomm a long time, 30 years. You're actually a boomerang CEO. You were here, you left, you came back, you've risen through the ranks. But walk me through that transition in tech and what 6G means for people like you and me.

每一代偶数代的无线技术都是巨大的。所以，2G 是巨大的，4G 是巨大的，那么 6G 仅仅因为是偶数代也会是巨大的。但我认为除此之外，6G 可能是我们在无线领域看到的最大的转变之一，而且它将远远超出我们对无线只是连接的想象。它还将涉及 AI 如何成为网络的一部分。整个行业会感觉有点不同。我在 MWC 上说过，我是故意想引起争议的。如果你记得电信是如何开始的，对吧？你有一个拨号音，然后你打电话给别人，对吧？突然你看看今天的电信，已经非常非常不同了。你有一个非常高容量的数据网络，你可以流媒体电视、做数据、按需计算，做一堆不同的服务，远远超出打电话。我认为这种转变也将在 6G 时发生。

Every even number generation of wireless is huge. So, 2G was huge, 4G was huge, then 6G just being an even number is going to be huge. But I think beyond that, I think 6G is one of the probably biggest transitions we're going to see in wireless and it's going to be way beyond, I think, how we think about wireless just connectivity. But it's going to be also how AI is going to be part of the networks. And it's going to feel a little bit different for the entire sector. I said at MWC, and I wanted to be provocative on purpose. If you remember how telecom started, right? You have a dial tone and you call somebody, right? And all of a sudden you look at telecom today, it's very, very different. You have a very high capacity data network where you stream television, you do data, you do computation on demand, you do a bunch of different services way beyond calling somebody. I think that type of transition is going to be also what's going to happen when you go to 6G.

我们经常谈论智能体。我们现在看到的一件事是，随着这些智能体和这种新型的 AI 设备，我们将看到我们称之为个人 AI 设备的设备。例如眼镜，这很自然，因为如果 AI 理解我们说的话、我们听到的、我们看到的，眼镜非常接近我们的感官，就像靠近你的眼睛、耳朵、嘴巴，你转头，你看到东西。所有这些信息对于智能体为你做事将是非常重要的上下文。所以，6G 的一个特点当然是，我需要有一个网络，我所看到的一切都可以以非常高的性能和速度流式传输到云端。所以，我们所有人都将成为行走的摄像头，对吧？这种“看到我所看到的”概念就是 6G 要做的。所以，6G 的一个消费者能关联到的特点是，这个无线电对我有什么好处？它将是一个非常快的上行链路。如果你想想 5G 发生了什么，5G 使高清视频流式传输到你的手机、你的笔记本电脑成为可能。现在，你将把信息流式传输到云端，这对于智能体和模型来说将是非常重要的上下文。这是连接方面。但 6G 的大局是因为 RF 信号可以被视为物理 AI。

So, we talk about agents a lot. One of the things that we are seeing now with those agents and this new kind of classes of AI devices is we are going to see devices we call them personal AI devices. Glasses, for example, it's very natural because if the AI understands what we say, what we hear, what we see, glasses is very close to our senses, like close to your eyes, your ears, your mouth, you turn your head, you see things. And all of this information is going to be very important context for agents to do things for you. So, one of the features of 6G, of course, is I need to have a network that everything that I see can get streamed to the cloud at a very high performance and high speed. So, all of us are going to be walking cameras, right? And this concept of see what I see is what 6G is going to do. So, one of the features of 6G that consumers relate to, what is this radio going to do for me? It's going to be a very fast uplink. If you think about what happened with 5G, 5G enabled streaming of high definition video to your phone, to your laptop. Now, you're going to stream information up to the cloud, which is going to be very important context for agents and for models. That's the connectivity side. But the big picture of 6G is because RF signals can be looked at as physical AI.

什么是 RF 信号？

And what's an RF signal?

它就像无线电信号，射频从塔、基站到你的手机，这就是你通过空中传输数据的方式。所有这些信号，都是电磁波，我们将把这些东西视为物理 AI。它只是传感器数据，你可以将 AI 应用于网络来理解所有这些事情。所以，如果我举个例子，当我看着我的汽车业务，我们有辅助驾驶和自动驾驶系统，我们有一堆传感器的输入。车上有摄像头，但也有雷达。就像雷达，你知道，发送信号，我会得到反射回来，然后它映射你周围的一切。有时当你看到一些高级驾驶系统的演示时，你会在屏幕上看到雷达能检测到的所有不同车辆。所以，想想我们每个人连接到 6G 网络，我们发送和接收的无线电。当 AI 处理所有这些数据时，这就像一个大规模的雷达。所以，6G 要做的另一件事，不仅在你的社区，不仅在城市、州，而是在整个国家层面，将绘制世界的数字孪生，我认为。这也变得非常有趣。当然，在今天的环境中，如果我这么说，每个人都会理解。比如你可以做无人机检测。所有移动的东西都会被跟踪。你有一个大规模的雷达。你可以管理整个未来的空中经济。我们看地图，你会看到这里拥堵吗？是绿色？黄色？红色？有了 6G，你将能够映射每一辆车、每一辆公交车、每一辆自行车、每一个行人，所有移动的东西。道路是什么？你周围的一切。你将能够使用 AI 来细化和检测不同的物体。你可以获得整个世界的数字孪生，随着我们继续演进我们对计算的思考方式，这将对智能体来说是非常非常重要的数据。所以，6G 是一个巨大的转变。我对此非常兴奋。这也是像高通这样的公司非常适合我们一直在做的事情的地方，我们如何多元化。我们可以提出一个从设备到网络再到数据中心的端到端故事。

It's like a radio signal, radio frequency that go from the tower, the base station to your phone, so that's how you transmit data over the air. All of those signals, which are electromagnetic waves, we're going to look at those things as physical AI. It's just sensor data and you can apply AI to the network to make sense of all of these things. So, if I give you an example, when I look at my automotive business and we have assisted driving and autonomous systems and we have input from a bunch of sensors. There's cameras on the car, but it also has radars. Like a radar, you know, send the signal, I'll get a reflection back and then he maps everything around you. Sometimes when you look at some demonstration of some advanced driving system, you see the screen all of the different cars that the radar can detect. So, think about every single one of us connecting to the 6G network, the radio that we transmit and we receive. When the AI processes all this data, this is like a radar at scale. So, another thing that 6G is going to do, not only at your neighborhood, not only at the city, at the state, but at the entire country level, will map the digital twin, I think, of the world. And that also becomes very interesting. Of course, in today's environment, if I say that, everybody will understand it. Like you can do drone detection. Everything that is moving is going to be tracked. You have a radar at scale. You can manage the entire future aerial economy. The things that we also look at maps and you see is there congestion here? Is it green? Is it yellow? Is it red? With 6G, you'll be able to map every car, every bus, every bicycle, every pedestrian, everything that's moving. What are the roads? Everything around you. You'll be able to use AI to refine and detect different objects. You can acquire a digital twin of the entire world and that's going to be very, very important data for agents as we continue to evolve how we think about computing. So, 6G is a big transformation. I'm very excited about it. And it's also where a company like Qualcomm is perfect for everything we have been doing, how we diversify. We can come up with an end-to-end story from the device to the network and to the data center.

那里有太多要解包的了，我想深入探讨很多内容。

There is so much to unpack there and I want to get to a lot of it.

那么，对于普通大众来说，将收集比以往更多的数据、更个性化的数据，并可用于与我们设备建立个人关系。但这有太多应用，我们可以达到非常精细的理解水平。但我想谈谈我们将与设备建立的关系，以及我们设备的未来形态可能是什么样。听起来我们正在从技术响应我们需求的关系，转向因为所有这些额外数据而预测我们需求并感知世界的关系。对吗？

So, for just the average Joe out there, it's more data, more personalized data than ever before will be collected and will be usable to create a personal relationship with our devices. But there are so many applications for that, that we can get to a really granular level of understanding everything out there. But I wanted to talk about that relationship that we're going to have to our devices, and then also what the future format could look like for what our devices will be. It sounds like we're moving from a relationship with our tech that is responsive to what we want, to anticipating what we want and sensing in the world because of all this extra data. Is that correct?

是的，这个说法很好。我认为你总结时提到的一点很重要。当你思考我们与设备的关系时，举个例子：我认为我们在计算领域体验的第一款个人设备是 PC。就像第一款设备，我们开始在 PC 上做很多事情。但后来手机出现了，发生的事情是人们并没有放弃 PC。你仍然拥有它。你仍然使用它。它非常有用。但某些工作负载确实从 PC 转移到了手机。我举个例子，因为现在你随时都带着电脑。例如，当电子商务开始时，人们开始在 PC 上进行大量电子商务。现在世界上大多数人会在手机上做电子商务。所以，但现在，快进到我们开始看到的情况。你看到所有那些不同的公司都在构建我们所谓的个人 AI 设备。你看到了很多 AI 公司。有一些秘密的形态我不能告诉你，但我认为我们几乎与所有公司都有合作。

Yeah, that's a good way to put it. And I think one of the things you said as you summarize it, there's an important point. When you think about what our relationship with devices is, an example: I think the first personal device we all got to experience in computing was the PC. Like the first device, and we started doing a lot of things on the PC. But then the phone arrived, and what happened is people didn't abandon the PC. You still have it. You still use it. It's incredibly useful. But certain workloads did shift from the PC to your phone. I'll give you an example, because now you have the computer with you all the time. As an example, when e-commerce started, people started to do a lot of e-commerce on your PC. Most of the world right now will do e-commerce on a mobile phone. So, but now, fast forward to what we're starting to see right now. You see all those different companies building what we call personal AI devices. You saw a lot of the AI companies. There are some secret form factors that I cannot tell you about, but I think we're working with pretty much all of them.

OpenAI、Meta？

OpenAI, Meta?

所有公司。而且人们会穿戴不同的东西。眼镜是最容易理解的，但还会有更多。它们会是珠宝、别针、吊坠，你穿戴的东西，可以连接到智能体。然后进行对话。如果 AI 理解我们写的东西，能读我们读的一切，看我们看的一切，那么用例类型会更加个人化，需要更多上下文，并且需要以稍微不同的方式发生。因为如果不是这样，你只需拿出手机，今天就能做。那么，我如何描述呢，我想，对你来说。比如，你走在路上，你有一副眼镜，你会说：“我真的很喜欢这个。我想买这个。亚马逊上多少钱？”它会说：“哦，是这个。亚马逊上就是这个价格。”然后你会说：“你能渲染一下我戴上这个的样子吗？”然后它会被渲染出来。所以，这将是一种不同的低摩擦体验，工作负载将开始转移，就像从 PC 转移到手机一样，某些事情你将通过智能体来做。我们甚至经常说这个例子，它显示了上下文的重要性。你走在路上，智能体会说：“我注意到你现在有 10 分钟。你能和我谈谈吗？你的日历上有一个冲突。我想问问你的选择。”很有趣的是，这些东西也会跨设备交互，比如，一个会议弹出来，智能体说：“你和医生预约有冲突。你想让我打电话给医生重新安排吗？”然后，它真的打了电话说：“我代表这个人打电话。我想更改预约。你下周什么时候有空？”所以，这些将是不同类型的用例，随着它们开始以低摩擦开发，我们将看到它们开始与其他类型的设备交互。思考这个问题的方式，这就是我们一直在谈论的生态系统，以你为例。行业正在发生巨大转变。我们来自一个以智能手机为中心的世界，我们为此感到非常自豪，因为我们在其中占了很大一部分。智能手机是你数字世界的中心。然后发生的是，一切都在智能手机周围。就像你拿起智能手机，你要用它做一切。如果你有可穿戴设备，可穿戴设备的工作只是扩展手机的功能。就像人们有时会买与手机同品牌的可穿戴设备，因为它可以给你通知，发送传感器数据。那是最好的。现在，当你使用智能体中的 UI 时，中心不再是你数字生活的中心。它是智能体。智能体将在你的手机、PC 和不同设备上显现。你会看到很多关于跨设备体验的讨论。当你有了这种转变，智能体需要理解上下文才能主动。你有今天训练过的模型，基于我们创造并放在互联网上的所有数据。所以，你进去，你看书和社交帖子，然后训练模型。想想明天，如果我们都带着摄像头看到我们所看到的，数据量是巨大的。将 dwarf 今天训练模型的数据，这就是 AI 将如何进化，并且将非常个性化地针对你。

All of them. And you have different things that people wear. Glasses is the easiest one to understand, but they're going to be more. They're going to be jewelry, pins, pendants, things that you wear, and it can connect to an agent. And now have this conversation. If the AI understands what we write, can read everything we read, see everything we see, the type of use cases are going to be a lot more personal, needs to have a lot more context, and needs to happen a little different. Because if it not, you just pull up your phone and you do it today. So, how do I describe that, I think, for you. Like, for example, you'll be walking around and you have a glass and you're going to say, "I really like this. I'd like to buy this. How much is it on Amazon?" And it's going to say, "Oh, it's this. This is how much it is on Amazon." And they're going to say, "Can you render how I'm going to look with this?" And it's going to get rendered. And so, it's going to be a different kind of low-friction experience, and workloads are going to start to shift in the same way they shift from your PC to phone, certain things you're going to do with an agent. We even have this example that we often say, which shows the importance of context. You're going to be walking around and the agent's going to say, "I just noticed you have 10 minutes right now. Can you talk to me? I have a conflict on your calendar. I'd like to ask you for options." Like, it's very interesting how also those things going to interact across devices, which is, you know, a meeting pop up and the agent said, "You have a conflict with a doctor appointment. Do you want me to call the doctor for you and reschedule?" And then, you know, it actually called and it said, "I'm calling on behalf of this person. I would like to change the appointment. What's your availability next week?" So, those are going to be different type of use cases, and as they started to get developed with low friction, we're going to see they'll start interacting with them with other types of devices. And the way to think about this, this is the way that we've been talking about the ecosystem of you as an example. There's a big shift in the industry. We are coming from and we're very proud of it because we had a big piece of that in the world that is a smartphone-centric. The smartphone is the center of your digital world. And then what happens is everything is around that smartphone. That's like you pick your smartphone and you're going to do everything with it. If you have a wearable device, the job of the wearable device is just to extend the functionality of your phone. Like people sometimes buy the wearable from the same brand of the phone because it's kind of give you notifications, you send sensor data. That's best. Now, the center when you use UI in an agent, the center of your digital life is no longer the phone. It's the agent. And the agent will manifest itself on your phone, on your PC, and across different devices. You're going to see a lot of discussions about experience across devices. And when you have this shift, the agent needs to understand context to be proactive. You have models that have been trained today on all of this data that we created and put it on the internet. So, you go in, you look at the books and social posts and in your train models. Think about tomorrow if all of us walking around with cameras to see what we see, that amount of data is massive. Will dwarf the data that's been training models today, and that's how AI is going to evolve and it's going to be very personalized to you.

高通的技术自最早型号以来就嵌入在智能手机中。该公司在 2000 年代初与苹果和 LG 等行业领导者达成了早期协议。如今，智能手机仍占高通高利润芯片销售的约 75%。但 AI 的快速进步正在将移动技术领域扩展到智能手机之外。Amon 并没有远离手机，而是积极投入，开发新的工具和系统，连接并增强人们已经使用的设备。你处于一个可以看到微软在构建什么、OpenAI 在构建什么、Meta 在构建什么的位置。我们已经谈了一些关于眼镜的事情，但你认为未来哪种形态会真正胜出？

Qualcomm technology has been embedded in smartphones since the earliest models. The company locked down early deals with industry leaders like Apple and LG in the early 2000s. And smartphones today are still about 75% of Qualcomm's high-margin chip sales. But rapid advancements in AI are now expanding the mobile technology landscape beyond the smartphone. Rather than moving away from the phone, Amon is leaning in, developing new tools and systems that connect and enhance the devices people already use. You sit in a position where you see what Microsoft is building, what OpenAI is building, what Meta is building. We've talked some about the glasses, but what future form do you think really will be the one that wins?

看，会有很多，对吧？有趣的是，那些智能体必须一直和你在一起。所以，这就是为什么我们看到可穿戴设备的整个概念正在转变为个人 AI 设备。而且它将是你穿着舒适的东西。你现在有了时尚和技术的混合，人们穿戴的 AI。

Look, you're going to have many, right? And the interesting thing about those agents is they have to be with you all the time. So, that's why we saw the whole concept of wearable is turning into a personal AI device. And it's going to be things that you are comfortable wearing. You have the mix now of fashion and technology, and you have AI that people wear.

而且我能在商店里买到，比如 2027 年、2028 年？

And available for me to buy in stores like 2027, 2028?

我想是的。你会在今年年底开始看到一些。

I think so. You're going to start to see some of those towards the end of this year.

太棒了。

Amazing.

是的。

Yes.

太棒了。所以，智能手机一直是主要设备。

Awesome. So, smartphones have been the primary device.

你觉得智能手机还会在多久内保持主要设备的地位？

How much longer do you think smartphones will be the primary device for?

我认为变化已经在发生了。如果我必须做个预测——而这种预测超级难做，所以我说我有 50% 的概率会猜对——但我要说，今年会是智能体之年，你会开始看到越来越多人们佩戴的形态。等到了 27、28 年，你会开始看到工作负载的转移。智能手机不会消失，但我是这样描述的：你掏出手机，你得掏出手机、打开手机、解锁手机，然后做手机该做的事，对吧？然而，手机并不适合让你指着东西，或者拿起手机跟手机说话。所以当你有了那些其他设备，某些事情——不是所有事情，而是某些事情——对你来说会更直觉，直接做就行了。比如，假设你戴着一副带摄像头的智能眼镜，你拿到一张餐厅账单，你看着账单，上面有个二维码，你说：“付这个，完成后通知我。”就这样，搞定了。所以我认为在 27、28 年内，我们会看到大量工作负载转移到那些设备上。

I think it's already in the process of a change. If I have to make a prediction — and it's super hard to make those predictions, so I'll say I'll give a 50% chance that I'll be correct — but I'll say this year will be the year of agents, and you'll start to see more and more form factors of things that people wear. When you start to get to '27, '28, you're going to start to see workload shift. The smartphone's not going to go anywhere, but here's how I describe this. If you pull up your phone, you have to pull up your phone, open your phone, unlock your phone, and you're going to do phone things, right? However, the phone is not natural for you to be pointing at things, or picking up your phone and talking to your phone. So as you have those other devices, certain things — not everything, but certain things — are going to be more intuitive for you to just do it. Like, for example, let's say you wear smart glasses that have a camera, and you get a restaurant bill, and you look at the bill, and it has a QR code, and you say, "Pay this, and just notify me once completed." And that's it, it's done. So I think what we're going to see within '27, '28 is a lot of workloads going to shift to those devices.

你觉得到那时那些设备会上市吗？我们已经看到智能眼镜了。

And will those devices be on market by then, you think? We see the smart glasses already.

它们已经开始进入市场了。我认为 27、28 年它们会规模化。我们常谈的一点是，那些设备现在只有几千万台。我认为在未来 5 年内，它们很可能达到几亿台，然后到十亿台。这一切都取决于智能体的成熟度。智能体替你做事的能力成熟了，一切就会变得非常自然。我相信我们所有人都会拥有不同的设备。这还会带来另一个有趣的动态。

They're starting to get on market. I think '27, '28 they get scale. One of the things we talk about is those devices right now are in the tens of millions. I think within the next 5 years, it's very possible they're going to go to hundreds of millions and going to get to a billion. And it's all about the maturity of the agents. The maturity of agents doing things for you, and then it's going to become very, very natural. And I believe that's how all of us are going to have different devices. And it creates another interesting dynamic.

但你觉得智能体会有一个主要设备吗？比如会有像吊坠或者眼镜这样的东西吗？听起来你相当看好眼镜。

But do you think there'll be a primary device for agents? Like will there be like the pendant or the glasses? It sounds like you're pretty bullish on glasses.

眼镜。我非常看好眼镜。原因是，你看，我自己就戴眼镜。但我认为人类对眼镜非常适应。眼镜非常自然。你转头，摄像头就能看到你眼睛看到的东西。它离你的耳朵很近，离你的嘴很近。你要读什么东西，摄像头就能读。所以我认为眼镜是主要形态。但让我说一件重要的事。你买 PC，它是消费电子设备。你买手机，它是消费电子设备。你有消费电子品牌。但当你想到你佩戴的东西，因为它贴近你、贴近你的感官，而智能体需要非常低摩擦、直觉——那是我们的时尚设备。所以现在你把时尚和技术混在一起了。这非常有趣。比如，你看看一些眼镜公司，它们可以变成科技公司，它们的估值倍数会扩张。但你买的仍然是一件时尚设备。你相信一家消费电子公司会做一款眼镜的六种颜色，然后所有人都戴那款吗？还是人们会挑他们想要的品牌。我认为接下来会发生的是，你会看到越来越多我们佩戴的东西变智能，会出现一批新的玩家。有一件事我可以精确地告诉你：每一代新的无线技术——就拿移动通信举例——玩家会变，行业会变。所以我认为我们会再次看到这种情况。

On glasses. I'm very bullish on glasses. And the reason is because, look, I wear glasses. But I think humans are very comfortable with glasses. And glasses are very natural. You turn your head, that's where the camera is going to see what your eyes are going to be seeing. It's very close to your ear. It's very close to your mouth. You're going to read something, the camera can read it. So glasses, I think, is the primary form factor. But let me tell something important about this. When you buy a PC, it's a consumer electronic device. You buy a phone, it's a consumer electronic device. You have consumer electronic brands. But when you think about things that you wear, because it's close to you, close to your senses, and agents need to be very low friction, intuitive — that's our fashion devices. So now you have the mix of fashion and technology. It's very interesting. For example, if you look at some of the glasses companies, they can become a technology company, their multiples going to expand. But you're still going to be buying a fashion device. Would you believe that a consumer electronic company will do one glass in six colors, and everybody's going to wear that? Or people are going to pick the brand that they want. And I think what's going to happen is you're going to see more things that we wear becoming smart, and it's going to be a new set of players. One thing I can tell you with precision: every new generation of wireless — just think about mobile as an example — the players change, the industry changes. So I think we're going to see that again.

所以你认为未来的 AI 设备会像苹果或谷歌那样吗？

So you all think there'll be like an Apple or a Google of the future AI device?

在个人 AI 设备上，会更横向，集中度会更低，会更加碎片化，因为不是每个人都穿同样的衣服，不是每个人都戴同样的眼镜。所以我认为会有很多不同的公司。关键在于控制点。过去控制点是操作系统和应用商店；现在变成了你使用的智能体。而且不会是一个智能体统治一切，而是不同的智能体。会有不同的智能体供你选择使用。我认为有趣的是，你开始看到像 OpenClaw 这样的东西，还有一堆东西被安装——

In personal AI devices, more horizontal, and it's going to be less concentration, it's going to be a lot more fragmentation, because not everybody wears the same clothes, not everybody wears the same glasses. So I think you're going to have a number of different companies. And the key thing is going to be the control point. They used to be the OS and the app store; it becomes the agents that you use. And it's not going to be like one agent that rules them all, it's going to be different agents. There are going to be different agents that you choose to use. And I think the interesting thing is you started to see things like with OpenClaw and a bunch of things getting installed for —

OpenClaw，给不知道的人解释一下，那是一个重大的智能体式 AI 时刻，真的席卷了互联网。它发布后能做非常惊人的事情。不过它也带来了很多网络安全问题。比如人们的数字钱包被联网之类的。

OpenClaw, for anyone who doesn't know, was like a big agentic AI moment that really took over the internet. It was released and it could do just incredible things. Although there were a lot of cybersecurity issues that came with it. Like people's digital wallets were coming online and things.

我举这个例子的原因是，它并不只限于那些个人 AI 设备；举个例子，连你的手机都会变。在上次财报电话会上我们谈过这个——没人注意。但我们说，中国的字节跳动推出了一款智能手机。那款手机有一个像 OpenClaw 一样的智能体。基本上是一回事。你跟手机说话，或者给手机发短信，说“帮我做这个”，它就会进入你的手机、进入你的应用，打开一个应用，关掉另一个应用，开始替你做事。所以我认为我们会看到行业的控制点在变化。不再是操作系统和应用商店。而是你会选择哪些智能体或 claw？会有多个。它会在你现有的设备上替你做事。然后会有新类别的设备，你搭配你选择的智能体来使用。它们会在你日常生活中替你做事。我认为事情就是这样演变的，AI 就是这样规模化的。也许这就是为什么我回到我们开始的地方。那就是临界点。

The reason I brought this example is because it's just not unique to those personal AI devices; even your phone's going to change, as an example. In the last earnings call, we talked about this — nobody paid attention to it. But we said that ByteDance in China launched a smartphone. And the smartphone has had an agent like OpenClaw. And basically it's the same thing. You talk to the phone or you text your phone and you say do this for me, and it goes to your phone and goes to your apps and starts one app, closes another app, and starts doing things for you. So I think we're going to see the control point of the industry changing. It's not about the OS and the App Store. It's going to be what are the agents or the claw that you're going to select? It's going to be multiple of them. It's going to do things for you on your existing devices. And then there are going to be new classes of devices that you're going to use with your agent of choice. They're just going to be doing things for you as you go about your day. And I think that's how this thing is going to pan out and how AI is going to get scale. Maybe that's why I go back to how we started. That's the tipping point.

随着 AI 无处不在，安蒙正在重新定位高通以迎接这一时刻。公司已投资于快速增长的汽车 AI 市场，预计未来 5 年将增长超过 65%。高通汽车业务在 2026 年第一季度创造了创纪录的 11 亿美元营收，连续第二个季度超过 10 亿美元。2025 年，高通收购了半导体和无线连接公司 AlphaWave，这是进军数据中心业务的关键一步。公司还投资于智能家居设备、机器人和边缘计算。

As AI becomes ubiquitous, Amon is repositioning Qualcomm to meet the moment. The company has invested in the quickly growing automotive AI market, which is projected to grow by more than 65% over the next 5 years. Qualcomm's automotive sector generated a record $1.1 billion in revenue in the first quarter of 2026, marking its second consecutive quarter surpassing $1 billion. In 2025, Qualcomm acquired semiconductor and wireless connectivity company AlphaWave, a key step into the data center business. The company has also invested in smart home devices, robotics, and edge computing.

我可以跟你聊未来聊一整天，但我也很想了解你是如何驾驭高通（Qualcomm）的。过去 5 年你担任 CEO，而你在公司待了差不多 30 年。你花了很大力气让业务多元化，智能手机和其中的骁龙芯片仍然是你的最大业务，但它已不是你唯一的业务线。我很好奇，在你的领导下，你是如何真正把重点放在让高通的营收线多元化上的，又是如何引导团队去执行的，因为现在汽车对你来说是个大领域，每个季度都有可观的营收进来，还有物联网、PC，而这些在 5 年前都还没有现在的规模。那么，你是如何围绕这个多元化使命来引导团队的？

I mean, I could talk future with you all day long, but I do also want to understand how you're navigating Qualcomm. In the last 5 years you've been the CEO, you've been with the company mostly for the last 30 years. You've gone to lengths to diversify the business, where smartphones and your Snapdragon chip within smartphones is still the top business for you, but it's not the only business line you have. I'm curious how you have, under your leadership, really put an emphasis on diversifying revenue lines for Qualcomm, and how you've oriented your team to execute on it, because now automotive is a big space for you, meaningful revenue coming in every quarter, Internet of Things, PCs, and those things were not to the scale of what they were 5 years ago. So, how have you oriented the team around this diversification mission?

这对我们来说是个重大而重要的议题。我认为它一直是我们的优先事项。我当 CEO 后的头号优先事项是：高通的技术与如此多的行业相关，我们有能力建立领导地位。我们怎么做，而且是同时做所有这些事，这并不容易，因为历史上很多公司在尝试做新事情、发展新核心能力时并不都那么成功。我觉得好的一点是，我始终可以依靠公司一项非常独特的资产。当然，我这么说可能很偏心。我也是以工程师身份加入高通的。我认为我们拥有不可思议的技术人才，这大概是我们所做一切的基础。我们意识到我们拥有一个可能独一无二的技术组合。大多数人想到高通就想到移动，但我们在无线通信领域拥有每一项技术——不只是蜂窝，我们在 Wi-Fi、蓝牙和定位上都是第一——而且我们拥有各种形态的计算。我们自己做 CPU、GPU、神经处理单元和图像信号处理器。所以我把公司组织成能够利用并扩展我们的技术路线图，去服务所有不同行业的需求和要求。比如，想想我们的 GPU 从渲染屏幕开始，当你进入汽车，你要同时渲染 12 块不同的屏幕，你必须扩展那种能力，我们必须在做的每件事里构建安全性。所以我们打造了一台工程机器，把我们的技术扩展到其他行业。我们建立了不同的业务部门，真正专注于为行业需求打造领先平台。我们必须建立许多不同的软件团队。有趣的是，我们实际上是在同样的运营费用水平下做到这一点的。我们从移动业务，走到今天的移动、个人 AI 业务、PC 业务，再到机器人、工业和汽车，一直到数据中心，而且全部并行推进。这并不容易。我们在外面看起来不错，内部就像高压锅。但我认为公司的文化——公司有不断重塑自己的历史。如果你看移动行业的公司，每一代无线技术都伴随着一大片公司墓地，而我们仍然在这里。我们挺过了这一切，我认为公司这种重塑自己的文化，这种好奇心去学习新事物、创新、做新事情，让我们能够建立结构并并行执行。我认为我们作为一家公司可能被误解了。人们总是在追逐闪光的东西，想要即时满足。当我看着投资者，我们一直在走一段旅程。让业务多元化、增长非移动业务需要时间。我想当我们走到今年上半年末——我们的投资者日已经定好日期——我们会说明我们在数据中心做什么，那让我们非常兴奋。但我非常相信高通。我认为高通的独特之处在于，我们可能是少数几家既能做低于 1 毫瓦的芯片，又能做数据中心 2,000 瓦芯片的半导体公司之一，而且全部并行。这正是我继续执行这一战略、多元化并增长公司的动力。

It's a big, important topic for us. I think it has been our priority. I think my number one priority when I became CEO: Qualcomm technologies are relevant to so many industries. We have the ability to build a leadership position. How do we do this, and we do all of it at the same time, which is not easy, because a lot of companies historically are not all very successful when they try to go do new things, develop new core competencies. I think the good thing is I could always rely on a very unique asset of the company. I think, obviously very partial. I also started at Qualcomm as an engineer. I think we have incredible technical talent, and I think that was probably the foundation to what we did. We realized that we had a probably unique technology portfolio. I think most people think of Qualcomm like they think about mobile, but we have every single technology in wireless communication — not only cellular, we're number one in Wi-Fi and Bluetooth and position location — but also we have every form of compute. We do our own CPUs or GPUs or neural processing units or image signal processors. So I organized the company in a way that we could leverage and scale our technology road map to serve the needs and requirements for all different industries. For example, if you think about our GPU from rendering screens, when you go to a car, you have to render 12 different simultaneous screens, you have to scale that capability, we have to build safety across everything that we do. So we built an engineering machine to scale our technology to other industries. We built different business units that will be really focused on building leading platforms for industry requirements. We have to build a lot of different software teams. And what is interesting about this is we actually did this with the same level of operating expenses. We went from being in the mobile business to now being in the mobile, the personal AI business, the PC business, into robotics and industrial and automotive, all the way to the data center, and doing all of this in parallel. It was not easy. I think we put a good face on the outside; inside it's like a pressure cooker. But I think the culture of the company — the company has a history of always reinventing itself. If you look at companies in the mobile industry, with every generation of wireless there was a big cemetery of companies, and we're still here. We survived all of it, and I think this culture of the company to reinvent itself, to be able to have the curiosity to learn new things, innovate, do new things, enabled us to build the structure and execute it in parallel. I think we're probably misunderstood as a company. I think people are always chasing the shiny object. People want instant gratification. When I look at investors, we have been on a journey. It takes time to diversify the business, growing the non-mobile business. I think as we get to the end of this first half of the year — we're going to have our investor day, the date has been set — we're going to say what we're doing in data center, which we're very excited about. But I'm a big believer in Qualcomm. I think what's unique about Qualcomm is we're probably one of the few semiconductor companies that can do a sub-1 mW chip and a 2,000 W chip for a data center within that range, all of it in parallel. And that's kind of the motivation I have to keep executing on this strategy and diversifying, growing the company.

你作为 CEO 几乎把全部身家押在想法上、并让团队朝那个方向努力的能力很棒。而且这也需要时间。5 年前你 75% 的业务是智能手机业务——今天是多少？

Your ability to almost bet the farm on ideas as a CEO and get the team oriented in that direction is great. And it also takes time. You had 75% of your business being a smartphone business 5 years ago — what is it today?

看，我认为我们的目标是到 2029 年做到大约 50/50。我们现在正朝着到 2029 年约 220 亿美元完全非移动业务的方向执行，而这还不包括一些新的押注，比如数据中心。尤其是——你提到押注一个想法。不只是押注一个想法，还要应对大量质疑，比如我们说要去汽车领域，就有人说，你们对汽车一无所知。你们要收购恩智浦（NXP）——那没成。你们不可能成功。看看今天：我们大概是汽车行业先进芯片的最大供应商。同样，我们进入 PC，所有人都说没有 x86 就没人能进 PC，我们说，好吧，我不这么认为。我们就继续执行。这个市场会变。现在人们理解了移动和 PC 的融合。苹果刚发布了基于移动芯片的 Neo。我们一直相信这一点。所以我认为现在轮到我们的工业和机器人业务。这就是信念。我认为这就是对我们技术、对打造领先平台能力的信心，以及我们能在市场上竞争并获胜的事实。

Look, I think our goal is to get to about 50/50 in 2029. I think we've now been executing to about $22 billion of completely non-mobile business by '29, and that doesn't include some of the new bets such as the data centers, for example. Especially — I think you mentioned something about betting on an idea. Not only betting on an idea, also dealing with a lot of skepticism, like when we said we're going to go to automotive, and it's sort of like, you guys don't know anything about automotive. You were going to buy NXP — that didn't go through. There's no way you're going to succeed. And look at today: we're probably the largest provider of advanced silicon for the automotive industry. Same thing, we entered into the PC, everybody said nobody can go into PC if you don't have x86, and we said, well, I don't think so. We're just going to keep executing. This market is going to change. There's now people understand convergence between mobile and PC. It's just Apple just launched the Neo that is based on silicon for mobile. We always believed that. So I think that's now going to be our industrial and robotics business. So that's the conviction. I think that's the confidence in our technology and ability to build a leading platform and the fact that we can compete and win in the marketplace.

所以我想谈谈两条新业务线，有些是全新的，但数据中心——我记得你去年秋天做了一个重大宣布，说你要进入这个领域，我相信今年晚些时候还会有更多。科技公司的大量资本开支都投向了数据中心。我们知道它能推动 AI 的未来，但它也带来问题。这里的能源状况阻碍了规模扩张。彭博最近有一篇报道说，一半的数据中心计划因材料问题之类的原因停滞或被取消。然后还有一个更大的共识：我们是不是在过度建设？这会不会让那些投入大量资金的公司陷入麻烦？你能不能给我一个大图景，讲讲我们在这一轮 AI 数据中心热潮中处于什么位置，并帮我拆解一下，因为这里面内容很多。

So I want to talk about two of the new lines of business, and some are brand new, but data centers — I think you really made a big announcement last fall that you're going into the space, and I'm sure more to come later this year. A lot of the capex spending from tech companies has been going into the data centers. We know it can fuel the future of AI, but it also comes with issues. The energy situation here prevents the scale. There's a report in Bloomberg recently that half of data center plans are kind of stalled or being canceled because of material issues and things like that. And then there's just a bigger consensus of, like, are we overbuilding here? Could this run companies into problems who are pouring lots of money in? Can you give me a big picture of where we are in this AI data boom and unpack that for me, because there's a lot here.

我们一直在谈的、AI 接下来会发生什么——它会改变一切算力，它会处理大量数据，数据会越来越多。我认为对 AI 算力的需求，尤其是推理——因为你训练一个模型，然后要把它投入生产，那就是推理——会持续增长。我认为对算力的需求会很高。我怎么描述这件事呢，不管人们讨论有没有泡沫，最简单的描述方式是这样——也许这个例子不太好，因为我不认为情况完全一样，但这是我唯一能举的例子。让我们回到 2000 年，那时有互联网泡沫，对吧？那时人们在说互联网会变成什么样。那个时候——我现在可以告诉你，26 年之后——互联网比人们想象的大得多。不管他们当时怎么想，跟实际发生的事相比都太小了。它不是一年之内发生的，但它确实发生了。我认为从长远看，AI 很可能仍然被低估了。算力和数据的量会继续增长，而且不只在云端增长——所有设备都会处理 AI。试着想象我们谈的这个未来：各种不同的设备、智能体等等。我认为这还会继续下去。我们可以争论曲线的斜率会不会变，是不是现在所有人都在为赢而战。我喜欢回到 2000 年互联网那个例子。最开始，人们大概会说 MapQuest 会成为地图。

Everything we've been talking about, what's going to happen with AI — it's going to change every compute, it's going to process a lot of data, it's going to have more and more data. I think the demand for AI computing, and especially inferencing — because you train a model, but then you want to put it into production, that's inferencing — that will continue to increase. I think the demand for compute is going to be high. And the way I would describe this, whether people talk about whether there's a bubble or not a bubble, here's the simple way to describe it — and maybe a bad example, because I don't think it's exactly like that, but it's the only example I can provide. Let's go back to the year 2000, when there was the dot-com bubble, okay? When people were saying what the internet is going to be. At that time — I would tell you right now, 26 years later — the internet is way bigger than people thought. Whatever they thought, it was small versus what exactly happened. It didn't happen all in one year, but it did happen. I think in the long run, AI is probably still underestimated. The amount of compute and the amount of data is going to continue to increase, and it's going to increase not only on the cloud — all the devices are going to be processing AI. Just try to imagine this future we're talking about: all the different devices and agents and all of that. I think that's going to continue to go. We can have an argument about whether the slope of the curve is going to change, whether everybody is now playing to win right now. I like to go back to this internet example in the year 2000. In the very beginning, probably people said MapQuest is going to be the map.

对。

Right.

也许并不是——或者 Orkut、MySpace 会成为社交网络。也许是，也许不是。我想我们都看到了后来发生了什么。

And maybe it isn't — or Orkut, and MySpace are going to be the social networks. Maybe it is, maybe it isn't. I think we saw what happened.

在新市场里，那就是先发劣势。

A first-mover disadvantage there in a new market.

我认为接下来会发生的是，现在还为时过早。所有人都在为赢而战，最终会有——也许会有少数几个赢家。也许更多。我们不知道。

I think what's going to happen is it's too early. Everybody's playing to win, and there's going to be — maybe there's going to be a handful of winners. Maybe there's more. We don't know.

你可以说——

You could argue —

你知道，未来几年，增长的斜率会有不同的变化，但增长会非常高，而且会持续非常高。短期内还有别的问题。这是算力的增长，这是能源的可获得性，这就是现实。我认为芯片进展非常快，能源基础设施项目则不是。

You know, as we go to the next few years, you're going to have different changes in the slope of the growth, but the growth is going to be very high and will continue to be very high. On the short term, you have other issues. This is the growth of compute. This is the energy availability, and that is the reality. I think silicon moves very fast. Energy infrastructure projects don't.

对。尤其是在美国。

Right. Especially here in the US.

而且我其实认为，在很多情况下这会是一个全球现象。所有这些都创造了机会。让我举个例子。手机和数据中心没有关系，但我要举一个手机的例子，因为它某种程度上说明了我们如何看待数据中心的下一步。手机是一个非常非常难的工程问题，因为如果你看今天的手机——一路回溯到你的功能机，对吧？你的手机有惊人的算力。你用手机做的事情多得多。但它仍然能放进口袋。它不能发热。你会把它贴在脸上。而电池没有太大变化。你拥有的能量就是那么多，而且必须撑一整天。如果撑不了一整天，它就没用。我必须把很高的算力密度塞进很小的空间。我不能用液冷。我也没有无限的电力。我不能插到墙上。所以我必须做一件叫“解耦计算”的事。我必须为每个任务设计专用计算。我举个例子——你大概还记得，我们都有 iTunes、iPod 和 PC 的时候，MP3 刚开始的时候，很多 MP3 解码是由 CPU 完成的。在手机里不是这样。CPU 太耗电。我们必须有一个专用加速器专门做 MP3 解码。你拍照的时候，我们有一个专用加速器专门做 JPEG 编码，举个例子。所以当我看数据中心正在发生的事，我喜欢做这个类比。这是算力需求，这是能源。所以就是说，好，你必须设计一种不同的架构，去匹配能源的可获得性。所以我认为现在会出现这个新趋势：数据中心的不同架构会走向能效优先，而我们相信我们在这里有角色可扮演。比如，当我们开始谈后 GPU 架构时，人们说：“你们根本不懂这个。你们不知道自己在说什么。GPU 是万能的。它是数据中心的解决方案。”后来，Nvidia 和 Groq 出现，人们说：“嗯，也许不同的事情需要不同的架构。”而这正是我们在做的。我们基本上在构建一个方案，从 CPU 视角和推理视角看会更节能，而且是为 AI 规模化、公司必须竞争、总拥有成本变得重要的时代设计的，它在算力和内存上会有不同的架构。我对此很乐观。我想我们会在 6 月底左右向世界公布那是什么。那将是下一个使命。我个人在这上面花了很多时间，很兴奋。我认为那也许是高通可以发挥作用的领域。

And I actually think in many cases it's going to be like a global phenomenon. All of this creates an opportunity. Let me give you an example. Phone has nothing to do with data center, but I'm going to give you a phone example because that kind of informs how we're thinking about what we're going to be doing in the data center. The phone is a very, very challenging engineering problem, because if you look at your phone today — go back all the way to your feature phone, right? Your phone has an incredible amount of computing power. There's a lot more things you do with your phone. But still fits in your pocket. It cannot get hot. You're going to touch your face. And the battery didn't change much. The energy that you have is the energy that you have, and it has to last all day. If it doesn't last all day, it's not useful. I have to pack a lot of compute density in a very small space. I cannot have liquid cooling. And I don't have unlimited power. I cannot plug anything to the wall. So therefore I had to go do something called disaggregated computing. I have to design this specialized computing for every task. The way I'll give you an example — I think you'll probably remember when all of us had iTunes and iPods and the PC, when MP3 started, a lot of the MP3 decode would be done by the CPU. Not in the phone. CPU burns too much power. We have to have a dedicated accelerator just to do MP3 decode. When you take a photo, we have a dedicated accelerator just to do JPEG encode, as an example. So when I look at what's happening at the data center, I like to draw this parallel. This is the demand for compute. This is the energy. So it's like, okay, you have to design a different architecture that is going to map to what the energy availability is. So I think there's going to be now this new trend of different architectures for the data center that are going to be energy efficient, and that's where we believe we have a role to play. For example, when we start talking about post-GPU architecture, people said, "You guys don't know anything about this. You don't know what you're talking about. The GPU is the do-all. It's the solution for data center." Then, Nvidia with Groq, people said, "Well, maybe there are different architectures for different things." And that's exactly what we're doing. We're basically building a solution that, from a CPU perspective and an inference perspective, is going to be more energy efficient, and it's going to be designed for when AI gets to scale and companies are going to have to compete — total cost of ownership matters — and it's going to have a different architecture for compute and memory. And I'm optimistic about it. I think we're going to tell the world what that is towards the end of June. And that's going to be the next mission. I'm personally spending a lot of time on this, and I'm excited. I think that's where maybe Qualcomm has a role to play.

再问一个关于你接下来方向的问题。我想听听你对机器人技术的看法。有很多预测说双足机器人——街上会有十亿台，做各种各样的事。工业机器人领域才刚刚真正开始起量。看起来那会是我们真正最先看到它们的地方，但家里像《杰森一家》里那种机器人感觉还比较远。你能谈谈机器人技术吗？

One more on where you're heading. I want to get your perspective on robotics. There's a lot of predictions about bipedal robots — there's going to be a billion of them on the streets doing all sorts of things. It's just starting to really hit in industrial robotics. It seems like that'll be the first place we'll really see them, but the Judy Jetson's robot in your house feels farther off. Can you just give me some on the robotics?

你大概从这场对话里注意到了，我喜欢做类比，然后看我们能从过去学到什么、应用到未来。我来做一个机器人技术和汽车之间的类比。对我们来说，我们在汽车领域成功的原因，是你不能把一台服务器放进汽车后备箱。所以我们需要做大量计算。你需要一个非常节能的方案。我认为这就是为什么我们很自然地进入机器人领域。机器人是一个边缘 AI 问题，就像汽车是一个边缘 AI 问题。然后我的类比来自我们思考 Robotaxi 和辅助驾驶的时候。我们进入汽车领域时，除了为数字座舱提供芯片、为 ADAS 和自动驾驶提供处理器，我们还开始构建辅助驾驶的软件栈。

As you probably noticed from this conversation, I like drawing parallels and then looking at things that we can learn from the past and apply in the future. I'll do a parallel between robotics and automotive. For us, the reason we became successful in automotive is because you cannot put a server in the trunk of a car. So we needed to do a lot of computing. You needed to have a very energy efficient [solution]. And I think that's why it was very natural for us to go to robotics. Robotics is an edge AI problem, like a car is an edge AI problem. And then my parallel comes when we think about robo-taxi and assisted driving. When we go to automotive, in addition to providing silicon for the digital cockpit and processors for ADAS and autonomy, we also started to build a stack for assisted driving.

你会发现，所有人都在考虑无人驾驶出租车。我们终将实现它。但无人驾驶出租车需要很长时间的训练。你可以把一套系统从零训练到 95%，但从 95% 到 99.999% 的安全水平，以至于可以移除方向盘、你只需坐着等待，这需要时间、需要里程、需要大量训练。但辅助驾驶的机会，当你敢于在需要时接管方向盘，那是巨大的。你可以在每辆车上做到这一点。这就是我对机器人技术的看法。它将从工业开始，从机器人能做的任务开始。你可以让机器人完美地执行某一项任务。你可以通过视频训练机器人。你可以通过我们称之为远程操作的方式训练，机器人会模仿。它会在那项任务上非常出色。然后你下载另一个任务，执行另一个任务。机器人将与你同行，在我们家里为你做一切，但这需要时间，因为每个房子都不同，如何导航等等。所以我做了这样的类比。我认为机器人在许多工业环境中都有巨大机会。比如像在杂货店或超市晚上补货这样简单的事情。因此，我们看到客户对我们的芯片有大量兴趣和活动。许多都类似于我们在汽车领域使用的配方。这将是一个巨大的机会，然后最终训练和成熟度将达到无人驾驶出租车的水平，我们将拥有像《杰森一家》那样的通用机器人。我认为机会很大。我认为物理 AI 使这些成为可能。作为高通，我们喜欢这一点，因为它像汽车一样，是边缘 AI。AI 需要在机器人内部。

And what you realize is there was everybody thinking about a robo-taxi. And we're going to get there. But a robo-taxi takes a long time for you to train. You can train a stack from zero to 95%, but for you to go from 95 to 99.999 so it's safe to the point that you remove the steering wheel of the car, you just sit and wait, that requires time, requires mileage, requires a lot of training. But the opportunity for assisted driving, when you dare to pick up the steering wheel if you need it, that's massive. You can do that in every car. That's how I think about robotics. It's going to start with industrial. It's going to start with tasks that a robot can do. You can perfect a robot to do one particular task. You can train the robot on video. You can train by doing that we call teleoperations and the robot will imitate. It will be very good at that task. And then you download another task, you do another task. The robot's going to walk with you, do everything for you in our house, and that's going to take time, you know, because every house is different, how you navigate, and all of those things. So, I drew that parallel. I think there's a massive opportunity for robots in a lot of industrial settings. Like something simple as restocking a shelf at night in a grocery store or supermarket. So, we're seeing a lot of activities from our customers interest in our chips. Many of it, it's kind of the same recipe we're using in automotive. That's going to be a big opportunity and then eventually the level of training and maturity will be like the robot taxi example that we're going to have general purpose robots like the Jetsons. I think the opportunity is big. I think, you know, physical AI enable those things. We like that as Qualcomm because it like a car, that's an edge AI. The AI needs to be in the robot.

而不是云端。

Versus cloud.

云端会有一些东西，但机器人需要实时完成一切，这确实是边缘 AI 的完美案例。而且机器人有不同的智能层次。我们称之为系统零、系统一、系统二。系统零就像，你知道，机器人应该抓住某物然后逃脱，然后再抓住它。它必须超快、非常低延迟。系统一，你告诉机器人拿起桌子上的这个东西。然后摄像头会看到、识别这是什么，它拿起来。系统二，推理并在机器人加云端中做事。这是一个巨大的机会，我们认为它将非常重要，特别是当你考虑专门的工业用例时。而且会有各种尺寸和形态的机器人。

There's going to be things in the cloud, but the robot needs to do everything real time and that's really the robot is a perfect case of an edge AI. And the robot has different levels, I think, of intelligence. So, we call them like system zero, system one, system two. System zero is like, you know, you get the robot is supposed to grab something and escape and he go and grab it again. He has to be super fast, very low latency. System one, you tell the robot pick up this in this table. Then the camera will see, recognize this what it is, it pick it up. System two, reasoning and do things in the robot plus the cloud. It's a great opportunity and I think we believe it's going to be significant, especially when you think about dedicated industrial use cases. And there's going to be robots in all sizes and form factors as well.

安蒙是一位技术乐观主义者。自 2021 年掌舵以来，安蒙帮助将科幻技术变为现实。高通正在开创免手驾驶、人形机器人和旨在主动完成任务的人工智能工具。该公司位于圣地亚哥的总部就是这些项目实践的活实验室。但在美国，对人工智能的怀疑仍然很高。皮尤研究中心 2025 年的一项调查发现，一半的美国公众对人工智能进入日常生活更担忧而非兴奋。安蒙和高通希望弥合这一差距，将非凡的技术变成日常工具。

Amon is a techno-optimist. Since taking the reins in 2021, Amon has helped turn the tech of science fiction into reality. Qualcomm is pioneering hands-free driving, humanoid robots, and AI tools designed to proactively finish tasks before they're even asked. The company's San Diego headquarters is a living lab of these projects in action. But skepticism around AI remains high in the US. A 2025 Pew Research Center survey found that half of the American public are more concerned than excited about AI entering their daily lives. Amon and Qualcomm want to bridge that gap, turning extraordinary tech into everyday tools.

在一个我们已经感受到一些技术反噬的世界里，有一些小运动，人们说：“哦，扔掉我的手机，我的智能手机。我只想再用翻盖手机。”或者孩子们抵制 Facebook，他们在做其他事情。我们是不是离它太近了？就像你和我热爱技术，你戴着眼镜，你知道，你想要大量数据，但你认为公众和消费者真的会愿意分享他们生活的一切和数据吗？

In a world where we already are feeling some tech lash, there's like small movements where people are saying, "Oh, throw out my phone, my smartphone. I just want to be on a flip phone again." Or kids are resisting Facebook and they're like, you know, doing other things. Are we just too close to it? Like you and I love technology and you wear glasses and you know, you want lots of data, but do you think actually the public and the consumer will want to be sharing everything about their lives and data like that?

这个问题很难回答，但我想用两个不同的数据点来回答。一个数据点是，如果它有用，如果它有帮助，用户就会拥抱它。我想我们已经看到，我们一直在谈论隐私问题，但如果你看看过去几十年发生的事情，我认为越来越多的用户，尤其是消费者，正在注册不同的平台。如果它有用，如果它消除了摩擦，对他们来说是更好的做事方式。我还认为智能体将消除目前技术中存在的大量杂乱。答案的另一部分实际上更重要。这将进一步区分谁应该是数据的保管者，谁是值得信赖的公司，谁不是值得信赖的公司，因为当你想到我们所有人，例如，成为行走的摄像头，这是现在比我们今天拥有的能力高出显著水平，这就是为什么我们看到对 6G 的浓厚兴趣。它经常与主权 AI 和其他一些事情的讨论联系在一起，因为它变得像关键基础设施，不仅因为能够检测所有移动、飞行的东西，还因为来自我们所有人的数据量，你将在世界上走动做事并向智能体发送数据。所以，这将是一个不同的世界，但这有点像我们在过去十年中看到的情况。

It's very difficult to answer this question, but I think I will answer this with two different data points. One data point is if it's useful, if it's helpful, users are going to embrace it. I think we have seen we've been talking about this issue of privacy, but if you just look at what happened over the past few decades, I think more and more our users, especially the consumers are signing up for different platforms. If it's useful, if it removes friction, it's a better way for them to do things. I also think agents will remove a lot of the clutter that exists with technology right now. The other part of the answer is actually more important. This is going to further separate who should be the custodians of the data, who are the trusted companies, who are not the trusted companies because when you think about all of us, for example, being walking cameras, this is now a significant level of capability above and beyond what we have today and that's why we see a lot of interest, for example, in 6G. It gets often associated with conversation about sovereign AI and a bunch of other things because it becomes like a critical infrastructure, not only because of the ability to detect everything that is moving, is flying, but also because of the amount of data that is from all of us, which you're going to be walking around and doing things in the world and sending data to the agents. So, it's going to be a different world, but it's kind of what we have seen over the past decade.

这只是一个演变。越来越多的数据将存储在云端，消费者对此更加成熟，并且会有很多关于如何处理这些数据的监管。

It is just going to be an evolution. More and more data is going to be on the cloud and consumers are more mature about it and there's going to be a lot of regulation about you know, what to do with this data.

所以，这是一种有趣的思考方式，关于谁可能成为赢家。我认为就像人工智能世界的许多事情一样，归根结底是你信任谁、你信任什么？如果最终任何人都能使用这些强大的大语言模型和这些智能体创建这些设备，那么可能归结为你宁愿让你的数据由苹果保管还是由 Meta 保管？我不知道，这些可能是消费者必须做出的权衡。

So, that's an interesting way of kind of thinking about who the winners of this could be. I think like it's so much of the AI world, it comes down to like who can you trust and what do you trust? And if anyone can create these devices eventually using these powerful LLMs and these agents, then it probably will come down to like would you rather have Apple with your data or would you rather have Meta with your data? I don't know, those are probably the tradeoffs consumers will have to make.

消费者始终是更复杂的讨论，但让我们以企业为例。这正是你现在开始看到的。企业有很多用于编码的人工智能应用。你有开源，你有不同公司提供的服务。只需看看公司如何处理他们的电子邮件，例如，他们的电子邮件，所有内部数据。那些公司信任哪些云公司为他们提供服务？所以，我相信你会看到很多这样的事情，并实际上为许多企业公司创造商业机会。回到你刚才说的评论。

The consumer's always a more complicated discussion, but let's just talk about enterprises as an example. That's exactly where you started to see right now. You have a lot of AI use for coding for enterprise. You have open source, you have, you know, different companies provide the service. Just look about how companies dealt with their email, for example, their email, all of their internal data. Who are the cloud companies that those companies have trusted to provide the service to them? So, I believe you're going to see a lot of those things and create actually business opportunities for a lot of the enterprise companies. Go back to the comment you just said.

Open Claw 刚推出时，人们会问：这东西安全吗？但你会看到，世界上一些大公司会推出面向企业的 Open Claw 版本，然后你会说：好，我愿意信任这家公司，把我的数据交给它。所以我认为这就是它的演进方式。我们已经看到了面向手机的 Claw、面向 PC 的 Claw。它还会进入你的汽车。它将无处不在。

Open Claw, when it started, people said, is this thing safe? But you're going to see that there are going to be Open Claw versions of some of the major companies in the world for enterprise, and then you're going to say, okay, I would trust this company with my data. So I think that's how this is going to evolve. We've seen our Claw for phones, Claw for PCs. It's going to be in your car. It's going to be everywhere.

我能感觉到你的技术乐观主义简直要溢出来了。你对你正在构建的东西如此兴奋，这是应该的，你确实在推动我们所有人都将生活其中的 AI 未来。我想听听你的看法：我们对这件事的兴奋程度和普通人的兴奋程度之间存在脱节。人们看到像 OpenAI 刚刚提出了一个 13 点提案，把当下比作新政时代、进步时代的历史时刻，认为我们需要新的社会结构、新的技术结构、新的一切。工作可能不一样了，甚至某种程度上可能不存在了。你怎么面对这一点？我们如何让你的乐观主义在一个消费者似乎对这样的未来有很多恐惧的世界里具有传染性？

So I can feel the techno-optimism just oozing off of you. You're so excited about what you're building, and as you should be, and you're really powering this AI future that we're all going to live in. I want to get your perspective on there's a disconnect between our level of excitement for this and the average person's excitement for this. People are seeing things like OpenAI just came out with a 13-point proposal equating the moment to the New Deal era, progressive era of history where we need a new societal structure, new tech structure, new everything. Jobs might not be the same or even exist to some extent. How do you level with that? Like how do we get your optimism to be contagious in a world where it seems like there's a lot for consumers to fear with this future?

首先，我们真的、真的在思考如何构建高效计算，让这项技术能够赋能人们。我认为我们至少有一个过往记录，那就是智能手机。每一项新技术，你都可以滥用。会有一些缺点，但你要看总体。我认为智能手机带来的是连接所有人，在许多国家用信息赋能人们。所以我们一直看到，人们实际上是如何变得具备数字能力、第一次体验互联网的，靠的就是智能手机。我认为 AI 也有这种赋能人们的能力。我不是那种认为 AI 会超越人类的人。我不这么认为。也许因为我们做的一些芯片就是人类使用的芯片。但我想给你我个人的反应。我不知道这是否正确。这只是我个人的反应。我 90 年代初从工程学院毕业，那是 92 年。在 90 年代初，你去办公室，那里有一台传真机，你收到一堆传真，我记得看着那些传真，你得再打一条消息发回去，然后电子邮件开始出现。接着互联网出现了。你想想，从没有互联网、没有电子邮件之类的东西，到有了互联网，那是多大的转变。它改变了我们的工作方式。那些是一堆不同的工具，从根本上非常不同。我也以同样的方式看待软件程序员。我们开始用汇编语言给计算机编程，那是实际的机器语言。突然间你有了像 C 这样的高级语言，可以用更高的层次来写，而现在你有了 AI 替你写。所以我把这些东西看作非常强大的工具，它可能被滥用，但它带来的变化可能和互联网到来时一样大，而我们挺过来了。所以在这件事上，我更像一个乐观主义者，而不是悲观主义者。

First of all, we really, really think about how can we build efficient computing that enables this technology in a way that it can empower people. And I think we have at least we have a track record, which is the smartphone. Everything, every new technology, you can misuse. You're going to have some drawbacks, but you look in aggregate. I think what the smartphone enabled is connected everyone, empower people with information in many countries. So we see that all the time how people actually became digitally capable and actually experienced the internet for the first time, it was with the smartphone. And I think AI has this capability to empower people. I am not one of those that think AI is going to be better than humans. I don't think so. And maybe because some of the chips that we do are the chips that the humans use. But I'll give you my personal reaction to this. I don't know if it's right. It's just my personal reaction. I graduated in engineering school in the early '90s. It was '92. You still have in the early '90s, you go to the office and there's a fax machine and you get a bunch of faxes and I remember look at the faxes and you have to type another message and send it and then email started to happen. And then the internet happened. And you think about it, how big of a transition was that when we didn't have the internet, we didn't have things like email and then we had the internet. It changed how we do work. Those are a bunch of different tools fundamentally very different. I also look the same way when you think about a software programmer. We start computers with programming assembly language like step is the actual machine language. All of a sudden you have a higher level language like C that you can write in a higher and now you have the AI that writes for you. So I look at those things as very powerful tools and it can be misused, but it's going to be probably as big of a change as when the internet arrived and we survived that. So I'm more of an optimist than a pessimist on this.

好吧，我希望你是对的。那听起来是个很棒的未来，非常感谢你分享你的见解。

Well, I hope you're right. That sounds like a great future and thank you so much for sharing your insights.

没问题。很高兴和你进行这次对话。

No problem. Very happy to have this conversation with you.

我也是。

Me, too.