Alphabet Inc. (NASDAQ:GOOG) Q4 2025 Earnings Conference Call February 4, 2024 4:30 PM ET
Operator 操作员
Welcome, everyone. Thank you for standing by for the Alphabet Fourth Quarter and Fiscal Year 2024 Earnings Conference Call. At this time, all participants are in a listen-only mode. After the speaker presentation, there will be a question-and-answer session. [Operator Instructions]
欢迎大家。感谢各位耐心等待参加 Alphabet 2024 年第四季度及财年收益电话会议。当前,所有参会者均处于仅收听模式。在演讲结束后,将进行问答环节。[操作员指示]
I would now like to hand the conference over to your speaker today, Jim Friedland, Senior Director of Investor Relations. Please go ahead.
现在,我想把会议交给今天的发言人,投资者关系高级总监吉姆·弗里德兰。请开始。
Jim Friedland
Thank you. Good afternoon, everyone, and welcome to Alphabet's fourth quarter 2024 earnings conference call. With us today are Sundar Pichai; Philipp Schindler; and Anat Ashkenazi.
谢谢大家。下午好,欢迎参加 Alphabet 2024 年第四季度财报电话会议。今天与我们一起的有桑达尔·皮查伊、菲利普·辛德勒和阿纳特·阿什克纳齐。
Now I'll quickly cover the Safe Harbor. Some of the statements that we make today regarding our business, operations and financial performance may be considered forward-looking. Such statements are based on current expectations and assumptions that are subject to a number of risks and uncertainties. Actual results could differ materially. Please refer to our Forms 10-K and 10-Q, including the risk factors. We undertake no obligation to update any forward-looking statement.
现在我将快速介绍安全港原则。我们今天关于业务、运营和财务表现的某些陈述可能被视为前瞻性声明。此类声明基于当前的预期和假设,并受到多种风险和不确定因素的影响。实际结果可能存在重大差异。请参阅我们的 10-K 和 10-Q 表格,包括风险因素。我们不承担更新任何前瞻性声明的义务。
During this call, we will present both GAAP and non-GAAP financial measures. A reconciliation of non-GAAP to GAAP measures is included in today's earnings press release, which is distributed and available to the public through our Investor Relations website located at abc.xyz/investor. Our comments will be on year-over-year comparisons unless we state otherwise.
在本次电话会议中,我们将提供 GAAP 和非 GAAP 财务指标。非 GAAP 与 GAAP 指标的对账信息包含在今天发布的收益新闻稿中,该新闻稿已通过我们的投资者关系网站 abc.xyz/investor 向公众发布并可供查阅。除非另有说明,我们的评论将基于年度同比比较。
And now I'll turn the call over to Sundar.
现在我将电话交给桑达尔。
Sundar Pichai
Thanks, Jim, and hello, everyone. We delivered another strong quarter in Q4, driven by our leadership in AI and our unique full stack. We're making dramatic progress across compute, model capabilities, and in driving efficiencies. We're rapidly shipping product improvements, and seeing terrific momentum with consumer and developer usage. And we're pushing the next frontiers, from AI agents, reasoning and deep research, to state-of-the-art video, quantum computing and more.
谢谢,Jim,大家好。我们在第四季度交出了一份强劲的成绩单,这得益于我们在人工智能领域的领导地位和独特的全栈技术。我们在计算、模型能力和效率提升方面取得了显著进展。我们正在快速推出产品改进,并在消费者和开发者使用方面取得了巨大动能。同时,我们正在开拓新的前沿领域,从人工智能代理、推理和深度研究,到先进的视频技术、量子计算等。
The company is in a great rhythm and cadence, building, testing, and launching products faster than ever before. This is translating into product usage, revenue growth, and results. In Search, AI overviews are now available in more than 100 countries. They continue to drive higher satisfaction in Search usage. Meanwhile, circle to search is now available on over 200 million android devices.
公司正处于良好的节奏和步调,比以往更快地构建、测试和推出产品。这转化为产品使用、收入增长和成果。在搜索方面,AI 概览现已在 100 多个国家推出,并持续提升搜索使用的满意度。同时,圈选搜索现已覆盖超过 2 亿台安卓设备。
In Cloud and YouTube, we said at the beginning of 2024 that we expected to exit the year at a combined annual revenue run rate of over $100 billion. We met that goal and ended the year at a run rate of $110 billion. We are set up well for continued growth. So today I'll provide an update on our AI progress and how it's improving our core consumer products. Then I'll touch on Cloud, YouTube, Platforms and Devices, and Waymo.
在云计算和 YouTube 方面,我们在 2024 年初表示,预计全年合并年收入运行率将超过 1000 亿美元。我们达成了这一目标,并以 1100 亿美元的运行率结束了这一年。我们已为持续增长做好了充分准备。因此,今天我将更新我们的 AI 进展,以及它如何改进我们的核心消费产品。然后,我会谈及云计算、YouTube、平台与设备以及 Waymo。
Let's start with AI. Last quarter, I outlined the three areas of our differentiated full stack approach to AI innovation. Our leading AI infrastructure, our world-class research, including models and tooling, and our products and platforms that bring these innovations to people at scale.
让我们从人工智能开始。上个季度,我概述了我们差异化的全栈人工智能创新方法的三个领域。我们的领先人工智能基础设施,我们世界一流的研究,包括模型和工具,以及将这些创新大规模带给人们的产品和平台。
First, AI Infrastructure. Our sophisticated global network of cloud regions and data centers provides a powerful foundation for us and our customers, directly driving revenue. We have a unique advantage, because we develop every component of our technology stack, including hardware, compilers, models, and products. This approach allows us to drive efficiencies at every level, from training and serving, to developer productivity.
首先,AI 基础设施。我们复杂的全球云区域和数据中心网络为我们和客户提供了强大的基础,直接推动收入增长。我们拥有独特的优势,因为我们自主开发技术栈的每个组件,包括硬件、编译器、模型和产品。这种方法使我们能够在各个层面提升效率,从训练和服务到开发者的生产力。
In 2024, we broke ground on 11 new cloud regions and data center campuses in places like South Carolina, Indiana, Missouri, and around the world. We also announced plans for seven new subsea cable projects, strengthening global connectivity. Our leading infrastructure is also among the world's most efficient.
在 2024 年,我们在南卡罗来纳州、印第安纳州、密苏里州及全球各地启动了 11 个新的云区域和数据中心园区的建设。我们还宣布了七个新的海底电缆项目计划,以加强全球连接。我们的领先基础设施也是全球最高效的之一。
Google data centers deliver nearly 4x more computing power per unit of electricity compared to just 5 years ago. These efficiencies, coupled with the scalability, cost and performance we offer, are why organizations increasingly choose Google cloud's platform. In fact, today, cloud customers consume more than 8x the compute capacity for training and inferencing compared to 18 months ago. We'll continue to invest in our cloud business to ensure we can address the increase in customer demand.
谷歌数据中心每单位电力提供的计算能力比五年前提高了近四倍。这些效率,再加上我们提供的可扩展性、成本和性能,是组织越来越选择谷歌云平台的原因。事实上,如今云客户用于训练和推理的计算能力消耗比 18 个月前增加了 8 倍以上。我们将继续投资于云业务,以确保能够满足客户需求的增长。
Second, world class research including models. In December, we unveiled Gemini 2.0, our most capable AI model yet, built for the agentic era. We launched an experimental version of Gemini 2.0 flash, our workhorse model with low latency and enhanced performance. Flash has already rolled out to the Gemini app, and tomorrow we are making 2.0 flash generally available for developers and customers, along with other model updates. So stay tuned.
其次,世界级的研究包括模型。去年 12 月,我们发布了 Gemini 2.0,这是我们迄今为止最强大的 AI 模型,专为智能代理时代打造。我们推出了 Gemini 2.0 Flash 的实验版本,这是我们的主力模型,具有低延迟和增强性能。Flash 已经在 Gemini 应用程序中推出,明天我们将正式向开发者和客户提供 2.0 Flash,并发布其他模型更新。敬请关注。

真不一定能最终胜出,钱不等于能量,别的不说,苹果很强,Safari就很一般,远不如Chrome,浏览器是非常核心的产品。
Late last year, we also debuted our experimental Gemini 2.0 Flash Thinking Model. The progress-to-scale thinking has been super fast, and the reviews so far have been extremely positive. We are working on even better thinking models and look forward to sharing those with the developer community soon. Gemini 2.0 advances in multi modality and made of tool use enable us to build new agents that bring us closer to our vision of a universal assistant.
去年年底,我们还首次推出了实验性的 Gemini 2.0 Flash 思维模型。进展速度非常快,迄今为止的反馈也极为正面。我们正在研发更先进的思维模型,并期待尽快与开发者社区分享。Gemini 2.0 在多模态和工具使用方面的进步,使我们能够构建新的智能体,从而更接近我们对通用助手的愿景。
One early example is Deep Research. It uses agentic capabilities to explore complex topics on your behalf, and give key findings, along with sources. It launched in Gemini Advanced in December, and is rolling out to android users all over the world. We are seeing great product momentum with our consumer Gemini app, which debuted on IOS last November.
一个早期的例子是 Deep Research。它利用代理能力代表你探索复杂主题,并提供关键发现及其来源。它于去年 12 月在 Gemini Advanced 上推出,并正在向全球的 Android 用户推广。我们的消费者 Gemini 应用在去年 11 月于 iOS 上首次亮相,我们正看到强劲的产品发展势头。
And we have opened up trusted tester access to a handful of research prototypes, including Project Mariner, which can understand and reason across information on a browser screen to complete tasks and Project Astra. We expect to bring features from both to the Gemini app later this year.
我们已向少数研究原型开放了受信任的测试者访问权限,其中包括 Project Mariner,它可以理解和推理浏览器屏幕上的信息以完成任务,以及 Project Astra。我们预计今年晚些时候将在 Gemini 应用中引入这两个项目的功能。
We're also excited by the progress of our video and image generation models. Veo 2, our state-of-the-art video generation model, and Imagen 3, our highest quality text-to-image model. These generative media models, as well as Gemini, consistently top industry leader boards and score top marks across industry benchmarks. That's why more than 4.4 million developers are using our Gemini models today, double the number from just 6 months ago.
我们对视频和图像生成模型的进展也感到兴奋。Veo 2 是我们最先进的视频生成模型,Imagen 3 是我们最高质量的文本生成图像模型。这些生成式媒体模型以及 Gemini 一直在行业排行榜上名列前茅,并在各类行业基准测试中获得最高评分。这就是为什么如今已有超过 440 万名开发者在使用我们的 Gemini 模型,这一数字是 6 个月前的两倍。
And we continue to drive research breakthroughs in quantum computing. At the end of last year, we announced Willow, our new state-of-the-art quantum computing chip that can reduce errors exponentially as we scale up using more qubits. Willow is an important step in our journey to build a useful quantum computer with practical applications. This technology holds so much promise, which is why there was real excitement around this breakthrough.
我们继续推动量子计算研究的突破。去年年底,我们发布了 Willow,这是一款最先进的量子计算芯片,能够在使用更多量子比特扩展时成倍减少错误。Willow 是我们打造具有实际应用价值的量子计算机旅程中的重要一步。这项技术充满潜力,因此这一突破引发了真正的兴奋。
Third, our Products and Platforms put AI into the hands of billions of people around the world. We have seven Products and Platforms with over 2 billion users and all are using Gemini. That includes Search, where Gemini is powering our AI overviews. People use Search more with AI overviews and usage growth increases over time as people learn that they can ask new types of questions. This behavior is even more pronounced with younger users who really appreciate the speed and efficiency of this new format.
第三,我们的产品和平台将 AI 带入全球数十亿人的手中。我们有七款用户超过 20 亿的产品和平台,所有这些都在使用 Gemini。其中包括搜索,Gemini 正在为我们的 AI 概览提供支持。人们在 AI 概览的帮助下使用搜索的频率更高,并且随着时间的推移,使用量会随着人们发现可以提出新类型的问题而增长。这种行为在年轻用户中尤为明显,他们非常欣赏这种新格式的速度和效率。

搜索是一个重要的入口,社交软件是一个更重要的入口。
We also are pleased to see how Circle to Search is driving additional Search use and opening up even more types of questions. This feature is also popular among younger users. Those who have tried Circle to Search before now use it to start more than 10% of their searches. As AI continues to expand the universe of queries that people can ask, 2025 is going to be one of the biggest years for Search innovation yet.
我们也很高兴看到“圈选搜索”如何推动额外的搜索使用,并开启更多类型的问题。这个功能在年轻用户中也很受欢迎。那些之前尝试过“圈选搜索”的用户,现在使用它来发起超过 10% 的搜索。随着人工智能不断扩展人们可以提出的问题范围,2025 年将成为搜索创新最重要的一年之一。
Now, let me turn to key highlights from the quarter across Cloud, YouTube, Platforms and Devices, and Waymo. First, Google Cloud. Our AI-powered cloud offerings enabled us to win customers such as Mercedes-Benz, Mercado Libre and Servier. In 2024, the number of first-time commitments more than doubled, compared to 2023.
现在,让我转向本季度在云计算、YouTube、平台与设备以及 Waymo 方面的关键亮点。首先是 Google Cloud。我们的 AI 驱动云产品使我们赢得了梅赛德斯-奔驰、Mercado Libre 和 Servier 等客户。2024 年,首次承诺的数量相比 2023 年增长了两倍以上。
We also deepened customer relationships. Last year, we closed several strategic deals over $1 billion, and the number of deals over $250 million doubled from the prior year. Our partners are further accelerating our growth, with customers purchasing billions of dollars of solutions through our Cloud marketplace.
我们还加深了与客户的关系。去年,我们达成了数笔超过 10 亿美元的战略交易,且超过 2.5 亿美元的交易数量是前一年的两倍。我们的合作伙伴进一步加速了我们的增长,客户通过我们的云市场购买了数十亿美元的解决方案。
We continue to see strong growth across our broad portfolio of AI-powered Cloud solutions. It begins with our AI Hypercomputer, which delivers leading performance and cost, across both GPUs and TPUs. These advantages help Citadel with modeling markets and training, and enabled Wayfair to modernize its platform, improving performance and scalability by nearly 25%.
我们在广泛的 AI 驱动云解决方案组合中持续实现强劲增长。这一切始于我们的 AI 超级计算机,它在 GPU 和 TPU 方面均提供领先的性能和成本优势。这些优势帮助 Citadel 进行市场建模和训练,并使 Wayfair 现代化其平台,将性能和可扩展性提升近 25%。
In Q4, we saw strong uptake of Trillium, our sixth-generation TPU, which delivers 4x better training performance and 3x greater inference throughput compared to the previous generation. We also continue our strong relationship with NVIDIA. We recently delivered their H200-based platforms to customers and just last week, we were the first to announce a customer running on the highly-anticipated Blackwell platform.
在第四季度,我们看到第六代 TPU Trillium 的强劲增长,其训练性能比上一代提升了 4 倍,推理吞吐量提高了 3 倍。我们还继续与 NVIDIA 保持紧密合作。最近,我们向客户交付了基于 H200 的平台,而就在上周,我们率先宣布有客户在备受期待的 Blackwell 平台上运行。
Our AI developer platform, Vertex AI, saw a 5x increase in customers year-over-year, with brands like Mondelez International and WPP building new applications and benefitting from our 200+ foundation models. Vertex usage increased 20x during 2024, with particularly strong developer adoption of Gemini Flash, Gemini 2.0, Imagen 3, and most recently, Veo.
我们的 AI 开发者平台 Vertex AI 的客户数量同比增长了 5 倍,Mondelez International 和 WPP 等品牌正在构建新应用,并受益于我们 200 多种基础模型。2024 年 Vertex 的使用量增长了 20 倍,尤其是 Gemini Flash、Gemini 2.0、Imagen 3 以及最新的 Veo 在开发者中得到了广泛采用。
We are also seeing strong growth in our AI-powered databases, data analytics, and cybersecurity platforms. Customers including Radisson Hotels are now using Gemini to Search and analyze multi-modal data from across multiple Clouds. Our AI-powered Threat Intelligence and Security Operations products help customers, including Vodafone and AstraZeneca, identify, protect and defend against threats.
我们还看到我们的 AI 驱动数据库、数据分析和网络安全平台的强劲增长。包括丽笙酒店在内的客户现在使用 Gemini 来搜索和分析跨多个云的多模态数据。我们的 AI 驱动威胁情报和安全运营产品帮助客户(包括沃达丰和阿斯利康)识别、保护和防御威胁。
Our growing portfolio of AI applications is also seeing strong customer adoption. In Q4, we introduced Google Agentspace, which helps enterprises synthesize data with Google-quality Search, create Gemini-powered agents, and automate transactions for employees. In addition, we recently gave all Google Workspace Business and Enterprise customers access to all of our powerful Gemini AI capabilities to help boost their productivity.
我们不断增长的人工智能应用组合也正在受到客户的广泛采用。在第四季度,我们推出了 Google Agentspace,帮助企业利用 Google 级别的搜索来整合数据,创建由 Gemini 驱动的智能代理,并为员工自动化交易。此外,我们最近向所有 Google Workspace 商业版和企业版客户开放了我们强大的 Gemini AI 功能,以帮助提升他们的工作效率。

除社交软件以外,Google在软件领域的优势是非常大的,甚至YouTube可以补一点社交属性。
Moving to YouTube. Nielsen data shows YouTube continues to be number one in streaming watchtime in the U.S., with our share of streaming now at a record high. On election day alone, over 45 million viewers across the U.S. watched election-related content on YouTube. Our early investment in podcasts is paying off. We integrated podcasts into the core YouTube experience, particularly with video.
转向 YouTube。尼尔森数据显示,YouTube 继续在美国的流媒体观看时长中排名第一,我们的流媒体份额现已创下历史新高。仅在选举日当天,全美就有超过 4500 万观众在 YouTube 上观看与选举相关的内容。我们对播客的早期投资正在取得回报。我们已将播客整合到核心 YouTube 体验中,尤其是与视频结合。
We are now the most frequently used service for consuming podcasts in the U.S., according to a recent Edison report. This success reflects our long-term approach of investing in emerging trends, from mobile to the living room. We now have over 250,000 creators in the YouTube Shopping affiliate program in the U.S. and Korea alone. We expanded YouTube Shopping at the end of last year to three additional countries, allowing even more creators to share their favorite products with fans and grow their businesses.
根据最近的 Edison 报告,我们现在是美国最常用的播客消费服务。这一成功反映了我们长期投资新兴趋势的策略,从移动端到客厅。目前,仅在美国和韩国,我们的 YouTube Shopping 联盟计划已有超过 25 万名创作者。去年年底,我们将 YouTube Shopping 扩展到另外三个国家,使更多创作者能够与粉丝分享他们最喜爱的产品并发展业务。
Philipp will talk more about YouTube performance later in the call. Next, Platforms and Devices. Google One's performance has been outstanding, and is one of our fastest growing subscription products in terms of subscribers and revenue growth. Last month, we announced the first beta of Android 16, plus new Android updates, including a deeper Gemini integration coming to the new Samsung Galaxy S25 series. We also recently announced Android XR, the first Android platform built for the Gemini era. Created with Samsung and Qualcomm, Android XR is designed to power an ecosystem of next-generation extended reality devices, like headsets and glasses.
Philipp 稍后将在通话中更多讨论 YouTube 的表现。接下来是平台和设备。Google One 的表现非常出色,是我们在订阅用户和收入增长方面增长最快的订阅产品之一。上个月,我们发布了 Android 16 的首个测试版,以及新的 Android 更新,其中包括即将应用于全新三星 Galaxy S25 系列的更深度 Gemini 集成。我们还最近宣布了 Android XR,这是首个为 Gemini 时代打造的 Android 平台。Android XR 由三星和高通共同打造,旨在为下一代扩展现实设备(如头戴式设备和智能眼镜)提供支持的生态系统。
Finally, a few words on Waymo which made tremendous progress last year, safely serving more than 4 million passenger trips. It is now averaging over 150,000 trips each week, and growing. Looking ahead, Waymo will be expanding its network and operations partnerships to open up new markets, including Austin and Atlanta this year and Miami next year. And in the coming weeks, Waymo One vehicles will arrive in Tokyo for their first international road trip.
最后,谈谈 Waymo,去年取得了巨大进展,安全提供了超过 400 万次乘客出行服务。现在每周平均行程超过 15 万次,并且还在增长。展望未来,Waymo 将扩展其网络和运营合作伙伴关系,以开拓新市场,包括今年进入奥斯汀和亚特兰大,明年进入迈阿密。此外,在未来几周内,Waymo One 车辆将抵达东京,开启其首次国际道路行程。
We're also developing the sixth-generation Waymo Driver, which will significantly lower hardware costs. I want to thank our employees around the world for another great quarter. 2025 is going to be exciting and we’re all ready for it.
我们还在开发第六代 Waymo Driver,这将大幅降低硬件成本。我要感谢我们全球的员工,又一个出色的季度。2025 年将会令人兴奋,我们都已做好准备。
Philipp, I will hand it over to you.
Philipp,我将把它交给你。
Philipp Schindler
Thanks, Sundar, and hello, everyone. I will quickly cover performance for the quarter, then frame the rest of my remarks around the progress we are delivering across Search, Ads, YouTube and Partnerships, highlighting the impact AI is having on our business and our customers.
谢谢,Sundar,大家好。我将快速回顾本季度的业绩,然后围绕我们在搜索、广告、YouTube 和合作伙伴关系方面取得的进展展开其余发言,重点介绍人工智能对我们的业务和客户的影响。
Google Services revenues were $84 billion for the quarter, up 10%, driven primarily by 11% year-on-year growth in advertising revenues. Strong growth in Search and YouTube advertising was partially offset by year-over-year decline in network revenues. In terms of vertical performance, the 13% increase in Search and other revenues was led by financial services followed by retail. The 14% growth in YouTube advertising revenues was driven by strong spend on U.S. election advertising, with combined spend from both parties almost doubling from what we saw in the 2020 elections.
谷歌服务季度收入为 840 亿美元,同比增长 10%,主要受广告收入同比增长 11%的推动。搜索和 YouTube 广告的强劲增长部分被网络收入的同比下降所抵消。从垂直行业表现来看,搜索及其他收入增长 13%,其中金融服务行业增长领先,其次是零售行业。YouTube 广告收入增长 14%,主要受美国大选广告支出的强劲推动,双方的总支出几乎是 2020 年大选时的两倍。
Now, in Q4, we saw continued strong growth in revenues from Search. We had lots of exciting updates in December, and we're rapidly integrating our AI innovation into our consumer experiences. We've already started testing Gemini 2.0 in AI overviews and plan to roll it out more broadly later in the year. In Search, we're seeing people increasingly ask entirely new questions using their voice, camera, or in ways that were not possible before, like with Circle to Search.
现在,在第四季度,我们看到搜索业务的收入持续强劲增长。我们在 12 月推出了许多激动人心的更新,并正在快速将我们的 AI 创新融入用户体验中。我们已经开始在 AI 概览中测试 Gemini 2.0,并计划在今年晚些时候更广泛地推广。在搜索方面,我们发现人们越来越多地使用语音、摄像头或其他以前无法实现的方式(如 Circle to Search)来提出全新的问题。
We're making these benefits available to more consumers. Google is already present in over half of journeys where a new brand, product, or retailer are discovered. By offering new ways for people to Search, we're expanding commercial opportunities for our advertisers. Shoppers can now take a photo of a product and, using lens, quickly find information about the product, reviews, similar products, and where they can get it for a great price. Lens is used for over 20 billion visual search queries every month, and the majority of these searches are incremental.
我们正在让更多消费者享受这些福利。Google 已经出现在超过一半的新品牌、产品或零售商被发现的旅程中。通过为人们提供新的搜索方式,我们正在为广告商拓展商业机会。购物者现在可以拍摄产品照片,并使用 Lens 快速查找有关该产品的信息、评论、类似产品以及以优惠价格购买的渠道。Lens 每月用于超过 200 亿次视觉搜索查询,其中大多数搜索是增量的。
Retail was particularly strong this holiday season, especially on Black Friday and Cyber Monday, which each generated over $1 billion in ad revenue. Interestingly, despite the U.S. holiday shopping season being the shortest since 2019, retail sales began much earlier, in October, causing the season to extend longer than anticipated.
今年假日季零售表现尤为强劲,尤其是在黑色星期五和网络星期一,这两天的广告收入均超过 10 亿美元。有趣的是,尽管美国的假日购物季是自 2019 年以来最短的一次,但零售销售从 10 月就早早开始,使得购物季比预期更长。
People shop more than a billion times a day across Google. Last quarter, we introduced a reinvented Google shopping experience, rebuilt from the ground up with AI. This December saw roughly 13% more daily active users on Google shopping in the U.S. compared to the same period in 2023. Closing out on Search with travel, and sharing another interesting trend where we saw spend expand to travel Tuesday. This contributed to 20% year-on-year revenue growth for travel advertisers across Cyber Monday and Travel Tuesday.
人们每天在 Google 上购物超过十亿次。上个季度,我们推出了全新的 Google 购物体验,从零开始用 AI 重新构建。今年 12 月,美国 Google 购物的日活跃用户数比 2023 年同期增长约 13%。在搜索方面,以旅行收尾,并分享另一个有趣的趋势,即我们看到支出扩展到了旅行星期二。这推动了网络星期一和旅行星期二期间旅行广告商的收入同比增长 20%。

搜索引擎本身是一个购物的网站。
Moving to Ads. We continue investing in AI capabilities across media buying, creative and measurement. As I said before, we believe that AI will revolutionize every part of the marketing value chain, and over the past quarter, we've seen how our customers are increasingly focusing on optimizing their use of AI. As an example, Petco used DemandGen campaigns across targeting, creative generation, and bidding to find new pet parent audiences across YouTube. They achieved a 275% higher return on ad spend and a 74% higher click through rate than their social benchmarks.
转向广告。我们继续在媒体购买、创意和测量方面投资于 AI 能力。正如我之前所说,我们相信 AI 将彻底改变营销价值链的每个环节,而在过去一个季度中,我们看到客户越来越专注于优化他们对 AI 的使用。例如,Petco 在 YouTube 上使用 DemandGen 活动来优化定位、创意生成和竞价,以寻找新的宠物主人受众。他们的广告支出回报率提高了 275%,点击率比社交媒体基准高出 74%。
On media buying, we made YouTube select creator takeovers generally available in the U.S. and will be expanding to more markets this year. Creators know their audience the best and creator takeovers help businesses connect with consumers through authentic and relevant content.
在媒体购买方面,我们在美国普遍推出了 YouTube 精选创作者接管,并将在今年扩展到更多市场。创作者最了解他们的受众,创作者接管帮助企业通过真实且相关的内容与消费者建立联系。
Looking at Creative, we introduced new controls and made reporting easier in PMAX, helping customers better understand and reinvest into their best performing assets. Using asset generation in PMAX. Event Ticket Center achieved a 5x increase in production of creative assets, saving time and effort. They also increased convergence by 300% compared to the previous period when they used manual assets.
从创意角度来看,我们在 PMAX 中引入了新的控制措施,并简化了报告,使客户能够更好地理解并重新投资于其表现最佳的资产。使用 PMAX 的资产生成功能,Event Ticket Center 的创意资产产出提高了 5 倍,节省了时间和精力。他们还相比之前使用手动资产的时期,提高了 300%的转化率。
And finally, Measurement. Last week, we made Meridian, our marketing mix model, generally available for customers, helping more business reinvest into creative and media buying strategies that they know work. Based on a Nielsen meta-analysis of marketing mix models, on average, Google AI-powered video campaigns on YouTube deliver 17% higher return on advertising spend than manual campaigns.
最后,衡量。上周,我们正式向客户推出了 Meridian 营销组合模型,帮助更多企业将资金重新投入到他们已验证有效的创意和媒介购买策略中。根据尼尔森对营销组合模型的元分析,平均而言,YouTube 上由 Google AI 驱动的视频广告活动比手动广告活动的广告支出回报率高 17%。
Turning to YouTube. We saw robust revenue growth backed by continued growth and watch time across ad supported and premium experiences. Our focus here remains on building a streaming platform that enables creators to thrive and unlock the full potential of AI. Expanding on our state-of-the-art video generation model, we announced Veo 2, which creates incredibly high quality video in a wide range of subjects and styles. It's been inspiring to see how people are experimenting with it. We'll make it available to creators on YouTube in the coming months.
转向 YouTube。我们看到强劲的收入增长,这得益于广告支持和高级体验的持续增长以及观看时间的增加。我们的重点仍然是打造一个流媒体平台,使创作者能够蓬勃发展并释放 AI 的全部潜力。在我们最先进的视频生成模型的基础上,我们发布了 Veo 2,它能够以各种主题和风格创作出极高质量的视频。看到人们如何尝试使用它,令人倍感鼓舞。我们将在未来几个月内向 YouTube 创作者提供该工具。
We continue to invest in helping YouTube creators work with brands. All advertisers globally can now promote YouTube creator videos and ad campaigns across all AI-powered campaign types in Google Ads. And creators can tag partners in their brand videos. Sephora used DemandGen's Shorts-Only channel to boost traffic and brand searches for the Holiday Gift Guide campaign and leverage creator collaborations to find the best gift. This drove an 82% relative uplift in searches for Sephora Holiday.
我们持续投资,帮助 YouTube 创作者与品牌合作。现在,全球所有广告商都可以在 Google Ads 的所有 AI 驱动的广告系列类型中推广 YouTube 创作者的视频和广告活动。创作者还可以在品牌视频中标记合作伙伴。Sephora 利用 DemandGen 的 Shorts-Only 频道提升了假日礼物指南活动的流量和品牌搜索,并通过创作者合作来寻找最佳礼物。这推动了 Sephora 假日搜索量相对增长 82%。
Shorts continues its ascent and is closing the gap with long form. In 2024, the monetization rate of Shorts relative to in-stream viewing increased by more than 30 percentage points in the U.S., and we expect to make additional progress in 2025. We're making it easier for advertisers to benefit from Shorts on all screens. We're particularly excited by its success on connected TV, which now makes up 15% of shorts viewing in the U.S.
短视频持续增长,并逐步缩小与长视频的差距。2024 年,短视频相对于插播观看的变现率在美国提高了 30 多个百分点,我们预计将在 2025 年取得更多进展。我们正在让广告商更容易在所有屏幕上受益于短视频。我们尤其对其在联网电视上的成功感到兴奋,目前短视频在美国的联网电视观看占比已达 15%。
Using a combination of ad formats, Louis Vuitton reached their overall objectives on both long-form and short-form content. Their shorts exceeded luxury goods benchmark for average view duration by 89% for equivalent video lengths, while their long-form content exceeded the benchmark by over 15%, with strong engagement from Gen Z and Millennials.
通过结合多种广告格式,路易威登在长短视频内容上都达成了整体目标。他们的短视频在相同时长下的平均观看时长比奢侈品行业基准高出 89%,而长视频则超出基准 15% 以上,并且在 Z 世代和千禧一代中获得了强烈的互动。
Looking into the living room, we continue to be number one in streaming watch time in the U.S for nearly two years, according to Nielsen, and our share of streaming is at a record high. Viewers globally streamed over 1 billion hours of YouTube content daily on their TVs in 2024. YouTube makes multi-year investments to tap into shifting consumer behavior. The current surge in living room viewership directly reflects years of work to build the right products and partnerships.
根据尼尔森的数据,放眼客厅,我们在美国的流媒体观看时长已连续近两年保持第一,我们的流媒体份额也创下历史新高。2024 年,全球观众每天在电视上观看超过 10 亿小时的 YouTube 内容。YouTube 进行多年投资,以把握消费者行为的变化。目前客厅观看量的激增,直接反映了多年来在打造合适的产品和合作伙伴关系方面的努力。
Creators are now prioritizing high-quality viewing experiences that truly shine on TV screens, inspiring even more viewers to tune in. In fact, the number of creators making majority of revenue from TV is up over 30% year-on-year. We have also invested in podcasts, where popular shows like Club Shay Shay and Lex Friedman are increasingly a visual format. YouTube creators and viewers are embracing this.
创作者现在优先考虑在电视屏幕上真正出彩的高质量观看体验,激励更多观众收看。事实上,从电视获得大部分收入的创作者数量同比增长了 30%以上。我们还投资于播客,像 Club Shay Shay 和 Lex Friedman 这样的热门节目正越来越多地采用视频格式。YouTube 创作者和观众都在积极接受这一变化。
In 2024, people watched over 400 million hours of podcasts each month on living room devices alone. YouTube is now the most popular service for podcast listening in the U.S., according to Edison. As always, let me wrap with the strong momentum we're seeing in partnerships, where the breadth of what Google has to offer is increasingly being recognized.
在 2024 年,人们每月仅在客厅设备上就观看了超过 4 亿小时的播客。根据 Edison 的数据,YouTube 现在是美国最受欢迎的播客收听服务。像往常一样,让我以我们在合作伙伴关系中看到的强劲势头来总结,谷歌所能提供的广度正越来越受到认可。
Sundar mentioned our deepening partnership with Samsung. Another expanding partnership is with Citi, who is modernizing its technology infrastructure with Google Cloud to transform employee and customer experiences. Using Google Cloud, it will improve its digital products, streamline employee workflows, and use advanced high-performance computing to enable millions of daily computations. This partnership also fuels Citi's generative AI initiatives across customer service, document summarization, and search to reduce manual processing.
Sundar 提到了我们与三星日益加深的合作伙伴关系。另一个正在扩展的合作伙伴是花旗银行,该公司正在利用 Google Cloud 现代化其技术基础设施,以改善员工和客户体验。借助 Google Cloud,花旗银行将优化其数字产品,简化员工工作流程,并利用先进的高性能计算支持每天数百万次计算。此次合作还推动了花旗银行在客户服务、文档摘要和搜索等方面的生成式 AI 计划,以减少人工处理。
With that, allow me a moment to thank Googlers everywhere for their extraordinary commitment and to our customers and partners for their continued trust. Anat, over to you.
借此机会,请允许我向全球的谷歌员工表示感谢,感谢他们非凡的奉献精神,同时也感谢我们的客户和合作伙伴一直以来的信任。Anat,交给你了。
Anat Ashkenazi
Thank you, Philipp. We're pleased with the continued momentum we're seeing across the business as Alphabet revenue for 2024 reached $350 billion, up 14% on a reported basis and 15% in constant currency versus 2023. My comments will focus on year-over-year comparisons for the fourth quarter, unless I state otherwise.
谢谢你,Philipp。我们对整个业务持续增长的势头感到满意,2024 年 Alphabet 的收入达到 3500 亿美元,按报告基准同比增长 14%,按不变汇率计算同比增长 15%。除非另有说明,我的评论将集中在第四季度的年度同比比较。
I will start with the results at the Alphabet level and we'll then cover our segment results. I'll end with some commentary and expectations over the first quarter and full year 2025. We had another strong quarter in Q4 with robust momentum across the business. Consolidated revenue of $96.5 billion, increased by 12% in both reported and constant currency. Search remained the largest contributor to revenue growth, followed by Cloud.
我将从 Alphabet 层面的业绩开始,然后介绍我们的各个业务部门的业绩。最后,我会对第一季度和 2025 全年做一些评论和展望。我们在第四季度再次取得了强劲的业绩,整个业务保持了强劲的增长势头。合并收入达到 965 亿美元,同比增长 12%,无论是按报告口径还是按固定汇率计算均如此。搜索仍然是收入增长的最大贡献者,其次是云业务。
Total cost of revenue was $40.6 billion, up 8%. Tech was $14.8 billion, up 6%. We continue to see a revenue mix shift with Google Search growing at double-digit levels, while network revenues, which have a much higher tech rate, declined. Other cost of revenue was $25.8 billion, up 9%, with the increase primarily driven by content acquisitions costs, primarily for YouTube, followed by depreciation, due to increasing investments in our technical infrastructure.
总收入成本为 406 亿美元,增长 8%。技术成本为 148 亿美元,增长 6%。我们继续看到收入结构的变化,Google 搜索以两位数增长,而技术成本率更高的网络收入有所下降。其他收入成本为 258 亿美元,增长 9%,主要由内容采购成本(主要是 YouTube)推动,其次是由于我们在技术基础设施上的投资增加导致的折旧。
Growth in content acquisition and depreciation were partially offset by our year-over-year decline in hardware costs due to the shift in timing of our made-by-Google launches to the third quarter 2024 compared to the fourth quarter of 2023.
内容获取和折旧的增长部分被我们的硬件成本同比下降所抵消,这一下降是由于我们的 Google 自研产品发布时间从 2023 年第四季度调整至 2024 年第三季度。
In terms of total expenses, the year-over-year comparisons reflect $1.2 billion in exit charges that we took in the fourth quarter of 2023 in connection with actions to optimize our global office space. As previously disclosed, those charges were allocated across the expense lines in other costs of revenue and OpEx based on associated headcount.
就总支出而言,同比比较反映了我们在 2023 年第四季度因优化全球办公空间而产生的 12 亿美元退出费用。正如先前披露的那样,这些费用根据相关员工人数分配到其他收入成本和运营支出(OpEx)的各项支出类别中。
Total operating expenses decreased 1% to $24.9 billion. R&D investments increased by 8%, primarily driven by increase in compensation and depreciation expenses, partially offset by the impact of charges for office [ph] space optimization in the fourth quarter of 2023.
总运营费用下降 1%至 249 亿美元。研发投资增长 8%,主要受薪酬和折旧费用增加的推动,部分被 2023 年第四季度办公空间优化费用的影响所抵消。
Sales and marketing expenses decreased 5%, primarily reflecting the optimization charges last year, as well as declines in compensation and in ads and promotion expenses due to the timing shift of the Pixel launch from Q4 to Q3. G&A expenses declined by 15%, reflecting a shift of timing in our charitable contributions, as well as the optimization charges last year.
销售和营销费用下降了 5%,主要反映了去年的优化成本,以及由于 Pixel 发布从第四季度提前到第三季度而导致的薪酬、广告和促销费用的下降。一般和行政费用下降了 15%,反映了我们慈善捐赠时间的调整,以及去年的优化成本。
Operating income increased 31%, the score [ph] to $31 billion, and operating margin increased to 32%, representing 4.6 points of margin expansion. Net income increased 28% to $26.5 billion, and earnings per share increased 31% to $2.15. We delivered free cash flow of $24.8 billion in the fourth quarter and $72.8 billion for the full year 2024. We ended the quarter with $96 billion in cash and marketable securities.
营业收入增长 31%,达到 310 亿美元,营业利润率提高至 32%,利润率扩张 4.6 个百分点。净收入增长 28%至 265 亿美元,每股收益增长 31%至 2.15 美元。我们在第四季度实现了 248 亿美元的自由现金流,全年自由现金流为 728 亿美元。我们以 960 亿美元的现金和可交易证券结束本季度。
Turning to segment results. Google Service revenues increased 10% to $84.1 billion, reflecting the strong momentum across Google Search and YouTube ads. Google Search and other advertising revenues increased by 13% to $54 billion. The robust performance of Search was once again broad-based across verticals, led by the financial service vertical due to strength in insurance, followed by retail.
转向分部门业绩。Google 服务收入增长 10% 至 841 亿美元,反映了 Google 搜索和 YouTube 广告的强劲势头。Google 搜索及其他广告收入增长 13% 至 540 亿美元。搜索的强劲表现再次在各个行业广泛展开,其中金融服务行业因保险业务的强劲增长而领先,其次是零售行业。
YouTube advertising revenue increased 14% to $10.5 billion, driven by brand, followed by direct response advertising. Network advertising revenue of $8 billion, were down 4%. In the fourth quarter, the year-over-year comparison in all of our advertising revenue lines was impacted by the increase in strength in advertising revenue in Q4 2023, in part from APAC-based retailers.
YouTube 广告收入增长 14%,达到 105 亿美元,主要由品牌广告推动,其次是直接响应广告。网络广告收入为 80 亿美元,下降了 4%。在第四季度,我们所有广告收入类别的同比比较受到 2023 年第四季度广告收入增长的影响,部分原因是亚太地区零售商的贡献。
Subscription platforms and device revenues increase 8% to $11.6 billion, primarily reflecting growth in subscription revenues, partially offset by the shift in timing of the launch of our made by Google devices to the third quarter, compared with the fourth quarter in 2023. We continue to have significant growth in our subscription products, primarily due to increase in the number of paid subscribers across YouTube TV, YouTube Music Premium, and Google One.
订阅平台和设备收入增长 8%至 116 亿美元,主要反映了订阅收入的增长,部分被我们自研 Google 设备的发布时间从 2023 年第四季度推迟至第三季度所抵消。我们的订阅产品仍保持显著增长,主要由于 YouTube TV、YouTube Music Premium 和 Google One 的付费订阅用户数量增加。
With regards to Platform, we saw a slight increase in the growth rate in Play, primarily due to a strong increase in the number of buyers. Google's service operating income increased 23% to $32.8 billion, and operating margin increased from 35% to 39%, representing a meaningful margin expansion.
关于平台,我们看到 Play 的增长率略有上升,主要是由于买家数量的大幅增加。谷歌的服务运营收入增长 23% 至 328 亿美元,运营利润率从 35% 提高到 39%,实现了显著的利润率扩张。
Turning to the Google Cloud segment, which continued to deliver very strong results this quarter. Revenue increased by 30% to $12 billion in the fourth quarter, reflecting growth in GCP across core GCP products, AI infrastructure, and generative AI solutions. Once again, GCP grew at a rate that was much higher than cloud overall. Healthy Google Workspace growth was primarily driven by increase in average revenue per seat.
转向谷歌云业务,本季度继续取得非常强劲的业绩。第四季度收入增长 30%,达到 120 亿美元,反映了核心 GCP 产品、AI 基础设施和生成式 AI 解决方案的增长。GCP 的增长率再次远高于整体云业务。健康的 Google Workspace 增长主要由每席平均收入的增加推动。
Google Cloud operating income increased to $2.1 billion, and operating margin increased from 9.4% to 17.5%. We're pleased with the work the cloud team is doing to deliver valuable solutions to the customer and generate revenue growth, as well as its continued focus on driving efficiencies across the cloud business.
Google Cloud 的营业收入增长至 21 亿美元,营业利润率从 9.4% 提高至 17.5%。我们对云团队所做的工作感到满意,他们致力于为客户提供有价值的解决方案并推动收入增长,同时持续关注提升云业务的运营效率。
As for Other Bets, for the fourth quarter, revenue were $400 million, and the operating loss was $1.2 billion. The year-over-year decline in revenue and increase in operating loss primarily reflect the milestone payment in the fourth quarter of 2023 for one of the Other Bets.
至于 Other Bets,第四季度收入为 4 亿美元,运营亏损为 12 亿美元。收入的同比下降和运营亏损的增加主要反映了 2023 年第四季度针对其中一个 Other Bets 的里程碑付款。
Turning to Alphabet level activities, the largest component of this line is our investments in AI research and development activities which support all of Alphabet. As a reminder, Alphabet level activities have included nearly all severance charges from reductions in workforce and office space charges. In the fourth quarter of 2024, the biggest factor in year-over-year comparison is the $1.2 billion in charges in the fourth quarter of 2023, almost entirely in connection with office space optimization.
转向 Alphabet 级别的活动,该类别中最大的一部分是我们在人工智能研究和开发活动方面的投资,这些活动支持整个 Alphabet。提醒一下,Alphabet 级别的活动包括几乎所有因裁员和办公空间调整而产生的遣散费用。在 2024 年第四季度,年度同比比较的最大因素是 2023 年第四季度的 12 亿美元费用,几乎全部与办公空间优化相关。
With respect to CapEx, our reported CapEx in the fourth quarter was $14 billion, primarily reflecting investments in our technical infrastructure, with the largest component being investment in servers, followed by data centers, to support the growth of our business across Google Services, Google Cloud, and Google DeepMind. In Q4, we returned value to shareholders in the form of $15 billion in share purchases and $2.4 billion in dividend payments. Overall, we returned a total of nearly $70 billion to shareholders in 2024.
关于资本支出,我们在第四季度报告的资本支出为 140 亿美元,主要反映了对技术基础设施的投资,其中最大部分是对服务器的投资,其次是数据中心,以支持我们的业务在 Google 服务、Google Cloud 和 Google DeepMind 领域的增长。 在第四季度,我们通过 150 亿美元的股票回购和 24 亿美元的股息支付向股东回馈了价值。 总体而言,我们在 2024 年向股东回馈了近 700 亿美元。
Turning to 2025, I would like to provide some commentary on several factors that will impact our business performance in both the first quarter and the full year 2025. First in terms of revenue, I'll highlight two items that will have meaningful impact on Q1 revenue across the company.
展望 2025 年,我想就几个因素提供一些评论,这些因素将影响我们在 2025 年第一季度和全年的业务表现。首先,在收入方面,我将重点介绍两项对公司第一季度收入将产生重大影响的事项。
First in terms of revenues, I'll highlight two items that will have meaningful impact on Q1 revenues across the company. The first is the impact of foreign exchange rates. At the current spot rates, we expect a larger headwind to our revenues from the strengthening of the U.S dollar relative to key currencies in Q1 versus Q4 2024.
首先,在收入方面,我将重点介绍两项对公司第一季度收入产生重大影响的因素。第一项是外汇汇率的影响。按照当前的即期汇率,我们预计由于美元相对于主要货币在 2024 年第一季度相比第四季度的走强,我们的收入将面临更大的逆风。
Second is the impact of leap year. We expect a headwind from having one less day of revenue in Q1 2025 compared with leap year in the first quarter of 2024. As for our segments, Google Services, advertising revenue in 2025 will be impacted by lapping the strength we experience in the financial service vertical throughout 2024. And in Cloud, given the revenues are correlated with the timing of deployment of new capacity, we could see variability in cloud revenue growth rates depending on when new capacity comes online during 2025.
其次是闰年的影响。我们预计 2025 年第一季度的收入将比 2024 年第一季度的闰年少一天,因此会面临一定的逆风。至于我们的各个业务板块,Google 服务和广告收入在 2025 年将受到 2024 年全年金融服务行业强劲表现的基数效应影响。而在云业务方面,由于收入与新容量的部署时间相关,我们可能会看到云收入增长率的波动,这取决于 2025 年新容量上线的时间。
Moving to investments, starting with our expectation for CapEx for the full year 2025. As we mentioned on the Q3 call, as we expand our AI efforts, we expect to increase our investments in capital expenditure for technical infrastructure, primarily for servers, followed by data centers and networking. We expect to invest approximately $75 billion in CapEx in 2025 with approximately $16 billion to $18 billion of debt in the first quarter. The expected total investment level may fluctuate from quarter-to-quarter, primarily due to timing of deliveries and construction schedules.
转向投资,首先是我们对 2025 年全年资本支出的预期。正如我们在第三季度电话会议上提到的,随着我们扩展人工智能工作,我们预计将增加在技术基础设施方面的资本支出投资,主要用于服务器,其次是数据中心和网络。我们预计在 2025 年资本支出约为 750 亿美元,其中第一季度约有 160 亿至 180 亿美元的债务。预计的总投资水平可能会因交付时间和施工进度的不同而在各季度间波动。
In terms of expenses, first, the increase in our investment in CapEx over the past few years will increase pressure on the P&L, primarily in the form of higher depreciation. In 2024, we saw 28% year-over-year growth in depreciation as we put more technical infrastructure assets into service. Given the increase in CapEx investments over the past few years, we expect the growth rate and depreciation to accelerate in 2025.
在支出方面,首先,过去几年来我们对资本支出的投资增加,将对损益表带来更大压力,主要表现为更高的折旧费用。2024 年,随着更多技术基础设施资产投入使用,我们的折旧费用同比增长 28%。鉴于过去几年资本支出投资的增加,我们预计 2025 年的增长率和折旧将加速。
Second, we expect some headcount growth in 2025 in key investment areas such as AI and cloud. As you just heard from Sundar, we're delivering products and solutions to customer at a rapid pace, building, testing, and launching products faster than ever before. And as I mentioned on the Q3 call, we're doing that while also focusing on driving further efficiencies in how we operate the business.
其次,我们预计 2025 年在 AI 和云等关键投资领域会有一定的人员增长。正如你刚刚从 Sundar 那里听到的,我们正在以快速的节奏向客户交付产品和解决方案,比以往任何时候都更快地构建、测试和推出产品。正如我在第三季度电话会议上提到的,我们在做到这一点的同时,也专注于推动业务运营的进一步效率提升。
Before we take questions, I'd like to recap the financial results for the year. For the full year 2024, revenue grew by 14% or by $43 billion, reaching $350 billion. Google Services and Google Cloud each continue to see double-digit revenue growth coupled with margin extension. YouTube and cloud revenues combined, ended the year at $110 billion annual run rate. And in 2024, we generate total income of $112 billion, an increase of 33% from 2023.
在回答问题之前,我想总结一下今年的财务业绩。2024 年全年,收入增长了 14%或 430 亿美元,达到 3500 亿美元。Google 服务和 Google Cloud 均继续实现两位数的收入增长,同时利润率也有所提升。YouTube 和云业务的总收入在年底达到了 1100 亿美元的年度运行率。2024 年,我们的总收入达到 1120 亿美元,比 2023 年增长 33%。

强大的盈利能力,已经是M7中最强。
We're pleased with the momentum we're seeing in AI innovation and monetization. We've been using AI to improve the performance of our ads business for well over a decade, and Cloud is generating billions in annual revenue from AI infrastructure and generative AI solutions. We're also excited about the potential to bring new experiences to users that will provide additional opportunities for monetization. And I look forward to sharing more in our progress throughout the year.
我们对人工智能创新和变现的势头感到满意。十多年来,我们一直在利用人工智能提升广告业务的表现,而云计算正在通过人工智能基础设施和生成式人工智能解决方案创造数十亿美元的年收入。我们也对为用户带来新体验的潜力感到兴奋,这将提供更多变现机会。我期待在今年分享更多我们的进展。
Sundar, Philipp, and I will now take your questions.
Sundar、Philipp 和我现在将回答你们的问题。
Question-and-Answer Session
问答环节
Operator 操作员
[Operator Instructions] And our first question comes from Brian Nowak with Morgan Stanley. Your line is now open.
[操作员说明] 我们的第一个问题来自摩根士丹利的布莱恩·诺瓦克。您的线路已打开。
Brian Nowak
Thanks for taking my questions. I have two, one for Sundar, one for Anat. Sundar, maybe kind of step back on Search, there's a -- it seems like there's a lot of advances to come with Gen AI and agentic possibilities with Search. Can you just sort of walk us through your big picture vision over the next few years of how you think about your Search product continue to evolve to stay at the top of the funnel and drive more engagement and monetization for your users and advertisers.
谢谢回答我的问题。我有两个问题,一个给 Sundar,一个给 Anat。Sundar,也许可以从搜索的角度回顾一下——似乎生成式人工智能和代理能力在搜索方面有很多进展。你能否为我们概述一下你对搜索产品在未来几年如何持续发展,以保持在用户获取的顶端,并推动更多用户和广告商的参与和变现的宏观愿景?
And then the second one, Anat, I think about 90 days ago, you talked about sort of further efficiencies and areas of simplification on the OpEx space. Can you just sort of walk us through some examples of where you see the potential for further efficiencies to the OpEx space, excluding the DNA step-ups that we have to come in 2025. Thanks.
然后是第二个问题,Anat,我想大约 90 天前,你谈到了在运营支出领域进一步提高效率和简化的方向。你能否举一些例子,说明你认为在运营支出领域还有哪些潜在的提升效率的空间,不包括我们在 2025 年必须面对的 DNA 增长?谢谢。
Sundar Pichai
Thanks, Brian. On Search, obviously, we view, I mean, this has been a long continual journey. AI overviews has been the next step. It's playing out positively, as we have indicated, the metrics look great, and we are obviously trading on that experience, bringing better and better models, expanding to the number of queries where it works and so on. But there's a lot more to come. I think we'll continue bringing AI in more powerful ways, in multi-modal ways.
谢谢,布莱恩。显然,在搜索方面,我们认为这是一段漫长而持续的旅程。AI 概览是下一个步骤。正如我们所指出的,它的表现是积极的,指标看起来很好,我们显然在这一经验的基础上不断优化,带来更好的模型,扩大其适用的查询数量等。但未来还有更多发展。我认为我们将继续以更强大的方式、更多模态的方式引入 AI。

AI可能不是替代而是新增,两种方式的区别还是挺大,AI提问的复杂度远高于搜索,AI取代的可能不是搜索而是咨询、教育,等等。
Things like what we've done with Lens Circle to Search, you can imagine the future with Project Astra. You can also imagine areas like we have done in Gemini Deep Research, possibilities where you are really dramatically expanding the types of use cases for which Search can work, things which don't always get answered instantaneously but can take some time to answer. Those are all areas of explorations and you will see us, putting newer experiences in front of users through the course of 2025. And so I do feel the opportunity space with AI, there's a lot of unlock ahead.
像我们在 Lens Circle to Search 中所做的那样,你可以想象 Project Astra 带来的未来。你还可以想象类似我们在 Gemini Deep Research 中所做的领域,探索搜索可以发挥作用的更多可能性,比如那些不一定能立即得到答案、但需要一些时间来解答的情况。这些都是我们正在探索的领域,你会看到我们在 2025 年陆续向用户推出新的体验。因此,我确实认为 AI 带来的机会空间仍然有很大的潜力可挖掘。
Anat Ashkenazi
Thanks. And for the question with regards to where do we see or where do I see leverage moving forward and some of the comments I've made on the previous call. I certainly see opportunities for further productivity and efficiency, and this is one of our priority areas. And we're going to do that so that we can make sure we continue to invest in areas such as AI and cloud where we see potential for continued growth.
谢谢。关于杠杆率未来的变化以及我在上一次电话会议上提到的一些评论,我确实看到进一步提高生产力和效率的机会,这是我们的重点领域之一。我们将这样做,以确保我们能够继续投资于人工智能和云计算等领域,因为我们认为这些领域具有持续增长的潜力。
I'll remain focused on areas that I've mentioned before, which include the technical infrastructure, so the $75 billion in CapEx I mentioned for this year, the majority of that is going to go towards our technical infrastructure, which includes servers and data centers. So ensuring we do that in the most efficient way is critical.
我将继续专注于我之前提到的领域,其中包括技术基础设施。因此,我提到的今年 750 亿美元的资本支出,其中大部分将用于我们的技术基础设施,包括服务器和数据中心。因此,确保我们以最高效的方式执行这一点至关重要。
Second is managing headcount growth, and we're going to be investing in areas of growth, such as AI and cloud, but looking across the organization and moderating that growth will be important. Optimizing the real estate footprint is one of the areas I've mentioned. We're continuing to focus on that. As well as looking at how we simplify the organization, we've previously mentioned bringing like areas together. Sundar talked about bringing some of the AI research teams together so that we can operate with increased speed, but also how we operate within the organization.
其次是管理员工增长,我们将投资于人工智能和云计算等增长领域,但审视整个组织并适度控制增长将是重要的。优化房地产布局是我提到的一个方面,我们将继续关注这一点。同时,我们也在考虑如何简化组织结构,之前我们提到将相似的领域整合在一起。Sundar 讨论了将一些人工智能研究团队整合在一起,以便我们能够以更快的速度运作,同时也优化我们在组织内部的运作方式。
Using our own AI tools to how we run the business, whether it's the code that Sundar mentioned on the previous call, writing code with AI, or even running some of our key processes using AI tools. So we're looking at all that. It's going to -- it's -- this is not a one-quarter type of effort. It's going to continue throughout the year, and we're going to continue to focus on that so that we can support the growth in other areas.
使用我们自己的 AI 工具来运营业务,无论是 Sundar 在上一次通话中提到的代码、使用 AI 编写代码,还是甚至使用 AI 工具运行我们的一些关键流程。所以我们正在全面考虑这些。这不是一个季度就能完成的工作。这将贯穿全年,我们将继续专注于此,以支持其他领域的增长。
Operator 操作员
Our next question comes from Doug Anmuth with JPMorgan. Your line is now open.
我们的下一个问题来自摩根大通的 Doug Anmuth。您的线路已打开。
Doug Anmuth
Thanks for taking the questions, one for Philipp and one for Anat. Philipp, can you just talk more about the expanded rollout of ads on AI overviews and perhaps what additional things you may have learned in 4Q? And I guess in particular, just curious if you rolled out to a higher percentage of commercial queries and is it still fair to say that you're monetizing nearly on par with existing search?
感谢您回答这些问题,一个给 Philipp,一个给 Anat。Philipp,您能否谈谈关于 AI 概览中广告扩展推广的更多信息,以及您在第四季度可能学到的其他内容?我特别好奇,您是否将其推广到更高比例的商业查询,并且是否仍然可以说您的变现几乎与现有搜索持平?
And then Anat, just on the -- on cloud growth, a little bit of decel 3Q to 4Q, but it sounded like you also suggested that your capacity constrained in the fourth quarter. I just wanted to push on that a little bit more. How -- is that accurate and is it fair to say that revenue growth could have been higher with much more capacity? Thanks.
然后,Anat,就云增长而言,从第三季度到第四季度略有放缓,但听起来你也提到第四季度的产能受限。我只是想再深入探讨一下。这是否准确?是否可以说如果产能更大,收入增长本可以更高?谢谢。
Philipp Schindler
So, on your first question, first of all, AI overviews, which is really nice to see, continue to drive higher satisfaction and search usage. So, that's really good. And as you know, we recently launched the ads within AI overviews on mobile in the U.S., which builds on our previous rollout of ads above and below. And as I talked about before, for the AI overviews overall, we actually see monetization at approximately the same rate, which I think really gives us a strong base on which we can innovate even more.
所以,关于你的第一个问题,首先,AI 概览的表现非常不错,持续推动更高的用户满意度和搜索使用率。这是一个好现象。正如你所知,我们最近在美国的移动端 AI 概览中推出了广告,这是在之前广告上下展示的基础上进一步扩展的。正如我之前所说,整体来看,AI 概览的变现率实际上大致相同,我认为这为我们提供了一个坚实的基础,让我们能够进一步创新。

AI是个底层的工具,对很多传统业务都有提升效率的作用,不仅仅是搜索,也包括手机、社交软件,几乎覆盖所有的领域。
Anat Ashkenazi
On the cloud questions, so first I'm excited that we ended the quarter at $12 billion and a 30% year-over-year growth, very impressive growth. And as I've mentioned, the prepared remark, GCP grew at a much higher rate than overall cloud. Two items to think about from a deceleration perspective. The first is we are lapping a very strong quarter of [indiscernible] deployment in Q4 of 2023.
关于云业务的问题,首先,我很高兴我们本季度以 120 亿美元的业绩收官,并实现了 30%的同比增长,增长非常令人印象深刻。正如我在准备好的发言中提到的,GCP 的增长率远高于整体云业务。从增速放缓的角度来看,有两个因素需要考虑。首先,我们正在与 2023 年第四季度一个非常强劲的[indiscernible]部署进行对比。
The second is the one you've alluded to. We do see and have been seeing very strong demand for AI products in the fourth quarter in 2024. We exited the year with more demand than we had available capacity. So we are in a tight supply demand situation, working very hard to bring more capacity online. As I mentioned, we've increased investment in CapEx in 2024, continue to increase in 2025, and will bring more capacity throughout the year.
第二个是你提到的那个。我们确实看到并一直看到 2024 年第四季度对 AI 产品的强劲需求。我们在年底时的需求超过了可用产能。因此,我们处于供需紧张的局面,正在努力增加产能上线。正如我所提到的,我们在 2024 年增加了资本支出投资,并将在 2025 年继续增加,并将在全年带来更多产能。
Doug Anmuth
Thank you both. 谢谢你们俩。
Operator 操作员
Your next question comes from Eric Sheridan with Goldman Sachs. Your line is now open.
您的下一个问题来自高盛的埃里克·谢里丹。您的线路已打开。
Eric Sheridan
Thank you for taking the question. I'll just ask one. Sundar, with the news that came out of China a little over 2 weeks ago, I think investors have been asking a lot of questions about the long-term cost curve for AI as AI moves from the infrastructure layer to the application layer or from training to inference and maybe even custom silicon becomes more dominant across the theme. I would love to get your perspective on your take on that news a couple of weeks ago and what it might mean for Alphabet longer term. Thank you.
谢谢你接受这个问题。我只问一个。Sundar,大约两周前来自中国的消息出来后,我认为投资者一直在询问很多关于人工智能长期成本曲线的问题,特别是当人工智能从基础设施层转向应用层,或者从训练转向推理,甚至定制芯片在这一趋势中变得更加主导。我想听听你对几周前那条新闻的看法,以及它对 Alphabet 长期发展的可能影响。谢谢。
Sundar Pichai
Thanks, Eric. Look, I think there's been a lot of observations on DeepSeek. First of all, I think a tremendous team. I think they've done very, very good work. Look, I think for us it's always been obvious, over time, there's frontier model development, but you can drive a lot of efficiency to serve these models really, really well.
谢谢,Eric。听着,我认为关于 DeepSeek 有很多观察。首先,我认为他们是一个出色的团队,我觉得他们做得非常非常好。听着,我认为对我们来说,这一直很明显,随着时间的推移,有前沿模型的开发,但你可以提高很多效率来很好地服务这些模型。
And if you look at one of the areas in which the Gemini model shine is the Pareto frontier of cost, performance, and latency and if you look at all three attributes, I think we lead this Pareto frontier. And I would say both are 2.0 flash models or 2.0 flash thinking models. They are some of the most efficient models out there, including comparing to DeepSeek's V3 and R1. And I think a lot of it is our strength of the full stack development into an optimization, our obsession with cost per query, all of that, I think, sets us up well for the workloads ahead, both to serve billions of users across our products and on the cloud side.
如果你看看 Gemini 模型在某些领域的优势,其中一个亮点就是成本、性能和延迟的帕累托前沿。如果你同时考虑这三个属性,我认为我们在这个帕累托前沿处于领先地位。我会说这两者都是 2.0 闪存模型或 2.0 闪存思维模型。它们是目前最有效的模型之一,包括与 DeepSeek 的 V3 和 R1 相比。我认为这在很大程度上归功于我们在全栈开发和优化方面的优势,以及我们对每次查询成本的执着追求。所有这些都使我们在未来的工作负载中处于有利位置,无论是为我们产品的数十亿用户提供服务,还是在云端提供支持。
Couple of things I would say are, if you look at the trajectory over the past 3 years, the proportion of the spend towards inference compared to training has been increasing, which is good because obviously inference is to support businesses with good ROIC. And so I think that trend is good. I think the reasoning models, if anything, accelerates that trend because it's obviously scaling upon inference dimension as well.
我想说的几件事是,如果你看看过去三年的发展轨迹,相较于训练,用于推理的支出比例一直在增加,这是一件好事,因为显然推理是为了支持具有良好投资回报率的业务。因此,我认为这一趋势是好的。我认为推理模型如果有什么作用的话,就是加速了这一趋势,因为它显然也在推理维度上扩展。
And so I think, look, I think part of the reason we are so excited about the AI opportunity is we know we can drive extraordinary use cases because the cost of actually using it is going to keep coming down, which will make more use cases feasible. And that's the opportunity space. It's as big as it comes. And that's why you're seeing us invest to meet that moment.
所以我认为,听着,我认为我们对人工智能机遇如此兴奋的部分原因是,我们知道自己可以推动非凡的用例,因为实际使用它的成本将持续下降,这将使更多用例成为可能。而这就是机遇空间,它的规模巨大。这就是为什么你会看到我们正在投资,以迎接这一时刻。
Eric Sheridan
Thank you. 谢谢你。
Operator 操作员
Your next question comes from Michael Nathanson with MoffettNathanson. Your line is now open.
您的下一个问题来自 MoffettNathanson 的 Michael Nathanson。您的线路已打开。
Michael Nathanson
Thank you. I have two, one for Philipp and one for Anat. Philipp, question for you is we're starting to see more AI tools on eCommerce sites and something like research with AI recommendations is on Google Shopping. Can you talk about how that product and other AI tools are impacting shopping behavior? I've had that impact in modernization.
谢谢。我有两个问题,一个给 Philipp,一个给 Anat。Philipp,你的问题是,我们开始在电商网站上看到更多的 AI 工具,比如 Google Shopping 上的 AI 推荐研究。你能谈谈该产品和其他 AI 工具如何影响购物行为吗?我在现代化方面也感受到了这种影响。
And then I guess the $75 billion question Anat is, how do you think about long-term capital intensity for this business? It sounds like there's a bit of constraint in terms of getting things built, but how do you think about the modeling of capital intensity going forward, and what are the things that you're looking forward to in terms of whether or not this is the right level of spend? Thank you.
那么,我想 750 亿美元的问题是,Anat,你如何看待这项业务的长期资本强度?听起来在建设方面存在一些限制,但你如何考虑未来资本强度的建模,以及你关注哪些因素来判断这是否是合适的支出水平?谢谢。
Philipp Schindler
Look, excellent question. We've been using our advances in AI to make search for products on Google even easier, obviously. And in Q4, we actually introduced quite a transformed Google Shopping experience, which we rebuilt from the ground up with AI. And people shop more than a billion times a day across Google. Last quarter, we introduced this fully reinvented Google Shopping experience. In December, we saw roughly, I mentioned, as 13% more daily active users in Google Shopping in the U.S compared to the same period last year. So that's a good development here.
看,这是个很好的问题。显然,我们一直在利用人工智能的进步让 Google 上的产品搜索变得更加便捷。在第四季度,我们实际上推出了一个全新改造的 Google 购物体验,我们从零开始用 AI 重新构建了它。每天有超过十亿次的购物行为发生在 Google 上。上个季度,我们推出了这个全新改造的 Google 购物体验。12 月,我们在美国的 Google 购物日活跃用户数比去年同期增长了大约 13%。所以这是一个不错的进展。

搜索引擎的购物体验和推荐引擎的不一样,是两种不同的体验。
And the new Google Shopping experiences, specifically to your question, uses AI to really intelligently show the most relevant products, helping to speed up and simplify your research. You get an AI-generated brief with top things to consider for your search, plus maybe products that meet your needs. So shoppers very often want low prices. So the new page not only includes like deal finding tools like price comparison, price insights, price tracking throughout, but there's also a new and dedicated personalized deals page where you can browse deals for you, and all this is really built on the backbone of AI. So we think this is a very interesting opportunity.
新的 Google 购物体验,具体回答你的问题,利用 AI 智能地展示最相关的产品,帮助加快和简化你的研究。你会得到一个 AI 生成的摘要,其中包含搜索时需要考虑的关键要素,以及可能符合你需求的产品。购物者通常希望获得低价。因此,新页面不仅包含价格比较、价格洞察、价格追踪等优惠查找工具,还有一个全新的个性化优惠页面,你可以浏览专属优惠,而这一切都建立在 AI 的基础上。因此,我们认为这是一个非常有趣的机会。
Anat Ashkenazi
And on the question of capital expenditure, and I think you may have two questions in there, one is just the capital intensity and then how would we think about return on that invested capital. So on the first one, certainly we are looking ahead, but we are managing very responsibly with a very rigorous even internal governance process looking at how do we allocate the capacity and what would we need to support the customer demand externally, but also across the Google, the Alphabet business.
关于资本支出的问题,我认为你可能有两个问题,一个是资本强度,另一个是我们如何看待已投资资本的回报。首先,当然我们在展望未来,但我们也在非常负责任地管理这一过程,采用严格的内部治理流程来审视如何分配产能,以及我们需要什么来支持外部的客户需求,同时也包括整个 Google 和 Alphabet 业务的需求。
And as you've seen in the comment I've just made on cloud, we do have demand that exceeds our available capacity, so we'll be working hard to address that and make sure we bring more capacity online. We do have the benefit of having a very broad business, and we can repurpose capacity, whether it's through Google Services or Google Cloud to support, as I said, whether it's Search or GDM or Google Cloud Customers, we can do that in a more efficient manner.
正如您在我刚刚对云的评论中所看到的,我们的需求确实超出了可用容量,因此我们将努力解决这一问题,并确保增加更多的容量上线。我们拥有非常广泛的业务,这使我们能够重新分配容量,无论是通过 Google 服务还是 Google Cloud 来支持,无论是搜索、GDM 还是 Google Cloud 客户,我们都可以以更高效的方式实现这一点。
We also look at every investment that we make to ensure that we're doing it in the most cost-efficient way to optimize our data center. As you know, our strategy is mostly to rely on our own self-design and build data centers. So they're industry-leading in terms of both cost and power efficiency at scale. We have our own customized TPUs. They're customized for our own workload. So, they do deliver outstanding superior performance and CapEx efficiency. So, we're going to be looking at all that when we make decisions as to how we're going to progress capital investments throughout the coming years.
我们还会审视我们所做的每一项投资,以确保我们以最具成本效益的方式优化数据中心。正如你所知,我们的战略主要依赖于自主设计和建设数据中心。因此,在成本和大规模的能效方面,它们都处于行业领先地位。我们拥有自主定制的 TPU,这些 TPU 是针对我们的工作负载量身定制的,因此它们能够提供卓越的性能和资本支出效率。因此,在决定未来几年如何推进资本投资时,我们会综合考虑所有这些因素。
Michael Nathanson
Thank you. 谢谢你。
Operator 操作员
Your next question comes from Mark Shmulik with Bernstein. Your line is now open.
您的下一个问题来自 Bernstein 的 Mark Shmulik。您的线路已打开。
Mark Shmulik
Yes, thanks for taking the question. Sundar, I just wanted to follow-up to Brian's question from earlier. With your own Project Mariner efforts and a competitor's recent launch, it seems there's suddenly really strong momentum on AI consumer agents and kind of catching up to that old Google duplex vision. I think when you look a few years ahead, where do you see consumer agents going and really, what does it mean to Google Search outside of Lens? Is there room for both to flourish or will they eventually be in conflict with each other? Thank you.
是的,谢谢你接受我的提问。Sundar,我只是想跟进一下之前 Brian 的问题。随着你们自己的 Project Mariner 计划以及竞争对手最近的发布,AI 消费者代理似乎突然获得了强劲的动力,并且在某种程度上追赶上了旧版 Google Duplex 的愿景。我想当你展望未来几年时,你认为消费者代理会走向何方?除此之外,这对 Google 搜索意味着什么?Lens 之外是否有空间让两者共同发展,还是它们最终会相互冲突?谢谢。
Sundar Pichai
Look, I think first of all, we are definitely seeing a lot of progress in the underlying capabilities of these models. Gemini 2.0 was definitely built with the view of enabling more agentic use cases. And so, I actually, and we are definitely seeing progress inside, and I think we'll be able to do more agentic experiences for our users. Look, I actually think all of this expands the opportunity space. I think historically we've had information use cases, but now you can have, you can act on your information needs in a much deeper way. It's always been our vision when we have talked about Google Assistant, et cetera. So I think the opportunity space expands.
听着,我认为首先,我们确实在这些模型的基本能力上看到了很多进展。Gemini 2.0 的构建确实是为了支持更多自主代理的用例。因此,我实际上认为,我们确实在内部看到了进展,我认为我们将能够为用户提供更多自主代理的体验。听着,我实际上认为所有这些都扩展了机会空间。我认为从历史上看,我们主要有信息使用的用例,但现在你可以以更深入的方式满足你的信息需求。这一直是我们的愿景,当我们谈论 Google Assistant 等时也是如此。所以我认为机会空间正在扩大。
I think there's plenty of, it feels very far from a zero-sum game. There's plenty of room, I think, for many new types of use cases to flourish. And I think for us, we have a clear sense of additional use cases we can start to tackle for our users in Google Search. And all the early work with AI overview shows that users will react positively to that. So, I'm pretty excited about what's ahead.
我认为有很多空间,这感觉与零和游戏相去甚远。我认为有足够的空间让许多新类型的用例蓬勃发展。而且我认为对于我们来说,我们清楚地知道可以在 Google 搜索中开始为用户解决的额外用例。所有关于 AI 概览的早期工作表明,用户对此会有积极的反应。所以,我对未来感到非常兴奋。
Operator 操作员
Your next question comes from Ross Sandler with Barclays. Your line is now open.
您的下一个问题来自巴克莱的罗斯·桑德勒。您的线路已打开。
Ross Sandler
Great. One on infrastructure and then one on the guidance on revenue. So Sundar, if we look at the inference cost per 1 million tokens and not the API pricing we see for Gemini versus something like GPT-4, but the raw cost of generating inference tokens on your TPU stack. How much more efficient do you think you guys are in terms of generating 1 million tokens compared to inference costs running on your cloud peers? Do you see this as an advantage as everything shifts to inference?
很好。一个关于基础设施,然后一个关于收入指导的问题。那么,Sundar,如果我们看每百万个 token 的推理成本,而不是我们在 API 定价中看到的 Gemini 与 GPT-4 之类的比较,而是你们 TPU 体系上生成推理 token 的原始成本。你认为在生成一百万个 token 方面,相比于运行在你们云端同行上的推理成本,你们的效率提高了多少?随着一切转向推理,你认为这是一个优势吗?
And then Anat, you call out lapping the financial services category strength in 2025 as a bit of a problem for Search. Could you just quantify a little bit of that? Is that kind of the same as when you guys talk about lapping the Asia outbound advertiser channel? Any numbers you could put around that headwind? Thank you very much.
然后,Anat,你提到在 2025 年金融服务类别的增长放缓对搜索业务来说是个小问题。你能稍微量化一下吗?这是否类似于你们之前提到的亚洲出境广告商渠道的增长放缓?有没有任何数据可以说明这一不利因素?非常感谢。
Sundar Pichai
Ross, like, look, the whole TPU project started, V1 was effectively an inferencing chip, so we've always, part of the reason we have taken the end-to-end stack approach is that, so that we can definitely drive a strong differentiation in end-to-end optimizing, and not only on a cost, but on a latency basis and a performance basis. We have the Pareto frontier we mentioned, and I think our full-stack approach and our TPU efforts all play -- give a meaningful advantage, and we plan, you already see that. I know you asked about the cost, but it's effectively captured.
Ross,听着,整个 TPU 项目的启动,V1 实际上是一个推理芯片,所以我们一直以来——我们采用端到端堆栈方法的部分原因是为了能够在端到端优化方面实现强有力的差异化,不仅在成本上,而且在延迟和性能方面也是如此。我们提到了帕累托前沿,我认为我们的全栈方法和 TPU 的努力都带来了有意义的优势,而我们计划——你已经看到了这一点。我知道你问的是成本问题,但这实际上已经涵盖了。
When we price outside, we pass on the differentiation. It's partly why we've been able to bring forward flash models at very attractive value props, which is what is driving both developer growth. We've doubled our developers to 4.4 million in just about 6 months. Vertex usage is up 20x last year. And so all of that is a direct result of that approach and so we will continue doing that.
当我们在外部定价时,我们传递了差异化。这部分解释了为什么我们能够以极具吸引力的价值主张推出闪存模型,这正是推动开发者增长的原因。我们在大约 6 个月内将开发者数量翻倍至 440 万。Vertex 的使用量去年增长了 20 倍。所有这些都是该方法的直接结果,因此我们将继续这样做。
Anat Ashkenazi
And on the question regarding my comment on lapping the strength in financial services, this is primarily related to the structural changes with regards to insurance, it is more specifically within financial services, it was the insurance segment and we saw that continue, but it was a one-time kind of a step up and then we saw it throughout the year. I am not going to give any specific numbers as to what we expect to see in 2025, but I am pleased with the fact that we are seeing and continue to see strength across really all verticals including retail and exiting the year in a position of strength. If anything, I would highlight as you think about the year, the comments I have made about the impact of FX, as well as the fact that we have one less day of revenue in Q1.
关于我对金融服务行业增长的评论,这主要与保险相关的结构性变化有关,更具体地说,是金融服务中的保险板块,我们看到这一趋势持续,但它是一次性的提升,随后贯穿全年。我不会提供关于 2025 年预期的具体数字,但我很高兴看到并继续看到所有行业的强劲表现,包括零售,并且全年以强势收尾。如果要强调一点,我会提到汇率影响的影响,以及第一季度收入减少一天的事实。
Operator 操作员
Your next question comes from Justin Post with Bank of America. Your line is now open.
您的下一个问题来自美国银行的贾斯廷·波斯特。您的线路已打开。
Justin Post
Great. Thank you. A couple for Philipp. You mentioned higher search usage with overviews. Just wondering how you're feeling about overall Search usage. Is it accelerating as you integrate more AI? I know there's a lot of traffic growth at competitors with AI, but just wondering if you're seeing a real increase in total volumes of information gathering.
很好,谢谢。给 Philipp 两个问题。你提到带有概览的搜索使用率更高。只是想知道你对整体搜索使用情况的看法。随着你们整合更多的 AI,搜索使用是否在加速?我知道竞争对手在 AI 方面的流量增长很大,但只是想知道你们是否看到信息收集总量的真正增长。
And then second, on YouTube, thinking about kind of maybe a move from more professional content to user-generated content, what is that doing for your usage, and how do you think about the margin impact from that? Thank you.
其次,在 YouTube 上,考虑从更专业的内容转向用户生成内容,这对你的使用情况有什么影响?你如何看待由此带来的利润率影响?谢谢。
Philipp Schindler
Maybe on -- Justin on Search usage overall, our metrics are healthy. We are continuing to see growth in Search on a year-on-year basis in terms of overall usage. Of course, within that, AI overviews has seen stronger growth, particularly across all segments of users, including in younger users. So it's being well received. But overall, I think through this AI moment, I think Search is continuing to perform well. And as I said earlier, we have a lot more innovations to come this year. I think the product will evolve even more. And I think as you make Search, as you give, as you make it more easy for people to interact and ask follow-up questions, et cetera, I think we have an opportunity to drive further growth.
也许关于搜索的整体使用情况,贾斯汀可以回答,我们的指标是健康的。我们继续看到搜索在整体使用方面按年增长。当然,在此之中,AI 概览的增长更为强劲,尤其是在所有用户群体中,包括年轻用户。因此,它受到了良好的欢迎。但总体而言,我认为在这个 AI 时代,搜索仍然表现良好。正如我之前所说,今年我们还有更多创新即将推出。我认为产品将进一步发展。我认为,当你让搜索变得更容易让人们互动并提出后续问题等时,我们有机会推动进一步增长。
Sundar Pichai
And some color on your YouTube question. Look, YouTube ads overall had a very healthy growth in Q4 driven by brand followed by direct response. The U.S. election advertising led brand revenue growth, and we saw nearly double the spending from 2020. I mentioned that we also had a strong contribution from finance, retail, and the tech verticals. So overall, strong operating metrics. Watch time growth remains robust, particularly in key monetization opportunity areas, such as Shorts and Living Room, just to set the stage here one more time.
关于你的 YouTube 问题,补充一些信息。整体来看,YouTube 广告在第四季度实现了非常健康的增长,品牌广告增长领先,其次是直接回应广告。美国大选广告推动了品牌收入增长,我们看到支出几乎是 2020 年的两倍。我提到,我们还在金融、零售和科技行业垂直领域取得了强劲的贡献。总体而言,运营指标表现强劲。观看时长增长依然稳健,特别是在关键的变现机会领域,如 Shorts 和客厅设备,再次为大家奠定背景。
On your question specifically on the UGC side, look, we have a very strong position with creators and we've always said creators are at the center of YouTube success here. They're the number one most important thing we care about and this strong position really gives us a lot of confidence here. Today we have more than 3 million channels that are actually in the YouTube Partner Program, which is really an incredible program. So we're very confident with the position we have and where this goes.
针对你关于 UGC 方面的问题,看看,我们在创作者方面拥有非常强大的地位,我们一直说创作者是 YouTube 成功的核心。他们是我们最关心的首要因素,而这一强势地位确实让我们充满信心。如今,我们在 YouTube 合作伙伴计划中拥有超过 300 万个频道,这确实是一个了不起的计划。因此,我们对自己的地位以及未来的发展充满信心。
Justin Post
Thank you. 谢谢你。
Operator 操作员
And our last question comes from Ken Gawrelski with Wells Fargo. Your line is now open.
我们最后一个问题来自肯·加雷尔斯基,来自富国银行。您的线路已打开。
Ken Gawrelski
Thank you very much. Two, please. I want to focus on Gemini and the consumer agent side. Sundar, there was some press reports that suggested you have an ambitious goal on growing the usage by the end of '25. Two questions on that, please. First, what’s -- how should we think about the approach that you're going to employ to achieve this goal? Is it more aggressive marketing for Gemini as a standalone product, or is it tighter integration into existing experiences, whether it be search mail, maps, et cetera.
非常感谢。两位,请。我想专注于 Gemini 和消费者代理方面。Sundar,有一些新闻报道称,你们有一个雄心勃勃的目标,希望在‘25 年底前增加使用量。关于这个有两个问题。首先,我们应该如何看待你们为实现这一目标所采用的方法?是更积极地将 Gemini 作为独立产品进行营销,还是更紧密地集成到现有体验中,比如搜索、邮件、地图等?
And then the second one is, how should we think about the future monetization opportunity of Gemini? Today, it's really a premium subscription offering or a free offering. Over time, do you see an ad component and anything you can share on that, please? Thank you.
然后第二个问题是,我们应该如何看待 Gemini 未来的变现机会?目前,它主要是一个高级订阅服务或免费服务。随着时间的推移,您认为会有广告组件吗?对此您能分享些什么吗?谢谢。
Sundar Pichai
First of all, we've had strong momentum for Gemini on the app side, particularly through the second half of 2024. And some of it is we've made it more easily accessible. We've brought it to, for example, with a dedicated app on iOS, which has been super positively received and definitely getting a lot of traction there. So definitely driving organic growth by putting the product out. We just last week rolled out with our 2.0 series of models. So 2.0 Flash, I mean, I think that's one of the most capable models you can access at the free tier. So that's definitely contributing as well. And so we are rapidly trading.
首先,在应用端,Gemini 的发展势头强劲,尤其是在 2024 年下半年。其中一部分原因是我们让它更易于获取。例如,我们在 iOS 上推出了专门的应用程序,受到了极大的好评,并且确实在该平台上获得了大量关注。因此,通过推出产品,我们正在推动有机增长。就在上周,我们推出了 2.0 系列模型。2.0 Flash——我认为这是免费层级中最强大的模型之一,这无疑也有所贡献。因此,我们正在快速迭代。
We've had a couple of key innovations there. Gemini Live, I think, has been definitely a hit with users, as well as for advanced users, Gemini Deep Research. So I think a combination of innovation, continually trading on the product and making it better, is driving a lot of usage. And we'll have a lot more to come as we go this year. And we obviously have a partnership with Samsung, so there are other things which will contribute to it as well.
我们在那里的关键创新有几个。我认为 Gemini Live 绝对受到用户的欢迎,而对于高级用户来说,Gemini Deep Research 也是如此。因此,我认为创新的结合、不断改进产品并使其更好,正在推动大量使用。今年我们还会有更多内容推出。显然,我们与三星有合作伙伴关系,因此还有其他因素会对此有所贡献。
On the monetization side, obviously for now we are focused on a free tier and subscriptions, but obviously as you've seen in Google over time, we always want to lead with user experience, We do have very good ideas for native ad concepts, but you will see us lead with the user experience. But I do think we're always committed to making the products work and reach billions of users at scale. And advertising has been a great aspect of that strategy. And so just like you've seen with YouTube, we'll give people options over time. But for this year, I think you will see us be focused on the subscription direction.
在盈利方面,显然目前我们专注于免费层和订阅,但正如你在 Google 发展过程中所看到的,我们始终希望以用户体验为先。我们确实对原生广告概念有很好的想法,但你会看到我们仍然以用户体验为主。不过,我认为我们始终致力于让产品发挥作用,并在大规模上触及数十亿用户。而广告一直是这一战略的重要组成部分。因此,就像你在 YouTube 上看到的那样,我们会随着时间的推移为用户提供选择。但对于今年来说,我认为你会看到我们专注于订阅方向。
Ken Gawrelski
Thank you. 谢谢你。
Operator 操作员
Thank you and that concludes our question-and-answer session for today. I would like to turn the conference back over to Jim Friedland for any further remarks.
谢谢大家,今天的问答环节到此结束。我想请 Jim Friedland 先生继续主持会议,并作进一步发言。
Jim Friedland
Thanks everyone for joining us today. We look forward to speaking with you again on our first quarter 2025 call. Thank you and have a good evening.
感谢大家今天加入我们。我们期待在 2025 年第一季度的电话会议上再次与您交流。谢谢,祝您晚安。
Operator 操作员
Thank you everyone. This concludes today's conference call. Thank you for participating. You may now disconnect.
谢谢大家。本次电话会议到此结束。感谢您的参与。您现在可以挂断。