AI投资回报惊喜：沃顿商学院发现75%的企业从AI中获得积极回报

本播客由谷歌赞助。

This podcast is sponsored by Google.

大家好。

Hey, folks.

我是Omar，谷歌DeepMind的产品与设计负责人。

I'm Omar, product and design lead at Google DeepMind.

我们刚在AI Studio推出了全新的可视化编程体验，让你能自由组合AI功能，以前所未有的速度将创意变为现实。

We just launched a revamped vibe coding experience in AI Studio that lets you mix and match AI capabilities to turn your ideas into reality faster than ever.

只需描述你的应用，Gemini就会自动为你配置合适的模型和API接口。

Just describe your app, and Gemini will automatically wire up the right models and APIs for you.

如果需要灵感，点击'手气不错'按钮，我们会帮你启动项目。

And if you need a spark, hit I'm feeling lucky, and we'll help you get started.

访问ai.studio/build创建你的第一个应用。

Head to a i.studio/ build to create your first app.

今天的《AI每日简报》要闻：四分之三的企业已从AI投资中获得正向回报。

Today on the AI Daily Brief, three quarters of enterprises are already seeing a positive ROI from AI investment.

头条速递：Anthropic预计2028年实现盈利，营收目标700亿美元。

And before that in the headlines, Anthropic projects profitability on 70,000,000,000 in revenue by 2028.

《AI每日简报》是聚焦人工智能领域重要新闻与深度讨论的每日播客及视频节目。

The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI.

朋友们，正式内容开始前先插播几条公告。

Alright friends, quick announcements before we dive in.

首先感谢赞助商Gemini、Super Intelligent、Robo、Robots and Pencils以及Blitzy。

First of all, thank you to our sponsors Gemini, Super Intelligent, Robo, Robots and Pencils, and Blitzy.

想获取无广告版本节目，请访问patreon.com/aideallybrief；了解节目赞助、职位机会等信息，请访问aideallybrief.ai。

To get an ad free version of the show, go to patreon.com/aideallybrief, and to learn more about sponsoring the show, or job opportunities, or anything else, visit us at aideallybrief.ai.

在官网上还可查看我们正在进行的AI投资回报基准研究，详情请访问roisurvey.ai。

While you're there, can also learn more about our AI ROI benchmarking study, which is live at roisurvey.ai right now.

欢迎回到AI每日简报头条版，五分钟带您了解所有每日AI新闻。

Welcome back to the AI Daily Brief headlines edition all the daily AI news you need in around five minutes.

Anthropic发布的最新财务预测显示，领先的AI实验室将持续保持强劲增长。

A new financial forecast from Anthropic suggests absolutely unrelenting growth for the leading AI labs.

根据The Information的报道，Anthropic预计到2028年将实现700亿美元营收和170亿美元正向现金流。

According to reporting from the information, Anthropic expects to generate $70,000,000,000 in revenue and have positive cash flows of $17,000,000,000 in 2028.

上个月路透社报道称，Anthropic预计年底ARR（年度经常性收入）将达到90亿美元。

Now last month, Reuters reported that Anthropic was on pace for $9,000,000,000 in ARR by the end of the year.

该报告指出，公司为2026年设定的营收目标在200亿至260亿美元之间。

That report noted a revenue target between 20,000,000,000 and $26,000,000,000 had been set for 2026.

具体来看，Anthropic预计今年API收入将达到38亿美元，是OpenAI最新预测18亿美元的两倍多。

Digging into line items, Anthropic expects API revenue to reach $3,800,000,000 this year, more than doubling the $1,800,000,000 most recently forecast by OpenAI.

Claude Code目前年化收入已达10亿美元，是7月份增速的两倍以上。

Claude Code itself is now generating a billion dollars in annualized revenue, which is more than double its pace from July.

对许多人而言，这份报告充分验证了Anthropic同时推进基础模型、应用层与企业客户直接合作的战略。

Now for many, the report is a big validation for Anthropic's strategy of pursuing foundation models, the application layer, and direct partnerships with enterprise customers all at the same time.

近几个月Anthropic已签署多项大型合作协议，包括德勤和Cognizant数十万席位的全组织部署。

Anthropic has signed a number of large partnerships in recent months including org wide deployments for Deloitte and Cognizant for hundreds of thousands of seats.

另一方面，这些数字表明Anthropic可能即将进行新一轮大规模融资。

On another front, the numbers suggest that Anthropic might have a big fundraising round coming soon.

他们上次融资在9月完成，公司估值1700亿美元。消息人士称下轮融资目标估值将在3000亿至4000亿美元之间。这种增长态势对整个行业而言，无疑是AI处于繁荣期而非泡沫期的积极信号。

Their last fundraising was completed in September and valued the company at $170,000,000,000 Sources suggest the next round would target a valuation between $300 and $400,000,000,000 Now to the extent we can extrapolate this level of growth to the sector in general, it's certainly a positive sign for AI being a boom rather than a bubble.

OpenAI在10月公布的ARR为130亿美元，但Sam Altman最近表示当前数字已远超该数值。

OpenAI reported 13,000,000,000 in ARR in October, but Sam Altman recently said the figure is now well more than that.

暗示2027年实现1000亿美元营收是现实预测，这一时间表被大幅提前。

Hinted that reaching 100,000,000,000 in revenue is a realistic forecast for 2027, shuffling up that timeline substantially.

除了惊人的增长数据，Anthropic的数字还表明其商业模式已找到明确的盈利路径。

Besides stratospheric growth, Anthropic's numbers also suggest that their business model has a clear path forward towards probability.

关于AI难以盈利的担忧一直不绝于耳，但Anthropic披露他们今年毛利率有望达到50%，到2028年将达到77%。

There's been a huge amount of hand wringing about AI being unprofitable, but Anthropic disclosed they're on pace to reach 50% gross profit margin this year and 77% by 2028.

公司预测2027年将首次实现正向自由现金流。

The company is forecasting 2027 is the year they flip into positive free cash flow for the first time.

考虑到这些公司表现与整体股市高度相关，值得注意的是本周AI股票表现动荡，投资者因担忧AI泡沫破裂而撤退。

Now given how tied up the performance of these companies is with the broader stock market, it's worth noting that AI stocks are having a bit of a shaky week as investors pull back on fears of the AI bubble bursting.

周二当天，AI股票领跌导致纳斯达克指数下跌2%，说实话这背后有太多不同诱因。

Back on Tuesday, AI stocks led a 2% fall on the NASDAQ, which frankly had so many different causes.

很难确切判断具体催化剂是什么。

It's very hard to sort out exactly what the catalyst was.

你可以归因于持续的政府停摆、贸易不确定性或实体经济状况恶化推动了这波行情。

You could point to ongoing government shutdown, trade uncertainty or deteriorating conditions in the real economy as driving the move.

高盛CEO大卫·所罗门在周二的活动上表示：'当这些周期来临时，行情可能会持续一段时间，但总有些因素会改变市场情绪，引发回调或改变增长轨迹预期，而我们谁都无法在真正发生前预见到这些变化。'

At an event on Tuesday, Goldman Sachs CEO David Solomon said, When you have these cycles, things can run for a period of time, but there are things that will change sentiment and will create drawdowns or change the perspective on the growth trajectory, and none of us are smart enough to see them until they actually occur.

摩根士丹利CEO泰德·皮克持类似观点，认为我们'应该坦然接受可能出现10%-15%回调的可能性'。

Morgan Stanley CEO Ted Pick had a similar view, commenting that we should quote welcome the possibility that there would be drawdowns.

这些回调并非由某种宏观悬崖效应引发。

10% to 15% drawdowns that are not driven by some sort of macro cliff effect.

尽管存在诸多不同之处，仍有许多人预期会出现宏观悬崖和AI股票全面崩盘，就像互联网泡沫末期那样。

Still many are expecting a macro cliff and a total collapse of AI stocks that mirrors the end of the .com bubble despite all the evidence of all the ways that it's different.

虽然周三AI股票趋于稳定，但风险平衡已明显转变。

And while Wednesday saw a stabilization for AI stocks, the balance of risks has clearly shifted.

景顺集团CEO安德鲁·施洛斯伯格表示：'某种程度上我们更接近调整，而非从这里再上涨10%或20%。'

Andrew Schlossberg, the CEO of Invesco said, There is some point where we will be probably closer to a correction than we are to a 10% or 20% rise up from here.

本周有个有趣的股票异动是Pinterest，投资者似乎对AI炒作失去了耐心。

One of the interesting stock moves this week was Pinterest with investors seemingly running out of patience with AI hype.

该股周三在业绩指引不及预期后暴跌21%。

The stock fell by 21% on Wednesday after guidance came in weaker than expected.

Pinterest也警告称广告支出因关税政策而疲软，但在同一份财报中，CEO比尔·雷迪表示：我们在AI和产品创新上的投资正在获得回报。

Pinterest also warned of softer ad spending linked to tariffs, yet during the same earnings report, CEO Bill Reddy said: Our investments in AI and product innovation are paying off.

我们已成为视觉搜索领域的领导者，并成功将平台转型为服务于6亿消费者的AI购物助手。

We've become a leader in visual search and have effectively turned our platform into an AI powered shopping assistant for 600,000,000 consumers.

延续上周科技巨头财报的主题，投资者似乎不再关心关于AI应用取得成效的声明。

Continuing the theme we saw with Big Tech Earnings last week, investors seem to no longer care about claims that AI adoption is paying off.

他们希望看到真实的投资回报率反映在利润上。

They want to see real ROI falling to the bottom line.

市场情绪低迷的一个原因是迈克尔·伯瑞正在寻找他的下一个大空头目标。

Now one reason for the fallen sentiment is that Michael Burry is looking for his next Big Short.

伯瑞因在金融危机期间做空住房债券获利1亿美元而闻名，后来被克里斯蒂安·贝尔在电影《大空头》中饰演。

Burry famously made $100,000,000 by shorting housing bonds during the financial crisis and was later depicted by Christian Bale in the film The Big Short.

我曾说过，这部电影几乎以一己之力让整整一代交易员宁愿每次都把所有东西称为泡沫，也不愿真正研究基本面。

I've spoken before about how I think that movie itself almost single handedly made it so a generation of traders would rather call everything a bubble every single time than actually try to engage with the fundamentals.

但现实就是如此。

But here we are.

周一晚间，伯瑞透露其对冲基金Scion Asset Management通过价值10亿美元的对Palantir和英伟达的看跌期权做空AI泡沫。

On Monday night, Burry revealed that his hedge fund, Scion Asset Management, is short the AI bubble via a billion dollars in put options on Palantir and Nvidia.

这约占其基金总值的80%，全部押注于这些看空头寸。

That's roughly 80% of the value of his fund concentrated in those bearish bets.

该披露比截止日期提前了两周，显然伯瑞意在引发市场讨论。

The disclosure came two weeks ahead of the deadline, so Burry clearly wanted to spark a narrative.

Palantir CEO亚历克斯·卡普在周三上午接受CNBC采访时煽风点火道：'当我听到空头攻击这家我认为显然是全球最重要的软件公司时，这非常令人愤怒。'

Palantir CEO Alex Karp fanned the discussion in an interview with CNBC on Wednesday morning saying: When I hear short sellers attacking what I believe is clearly the most important software company in the world, it's super triggering.

每次他们做空我们，我们都会三倍努力拿出更好的业绩。

Every time they short us, we are just tripling down on getting better numbers.

某种程度上，就是为了让他们变得更穷。

In part, to make them poorer.

他做空的两家公司正是最赚钱的企业，这非常奇怪。

The two companies he's shorting are the ones making all the money, which is super weird.

认为芯片和本体论是你应该做空的对象简直荒谬至极。

The idea that chips and ontology is what you wanna short is batshit crazy.

黄仁勋也对巴里的持仓提出了质疑。

Jensen Huang also pushed back on Burry's positioning.

他在接受英国天空新闻采访时驳斥了AI泡沫的说法，表示'我们离泡沫还很遥远'。

Speaking with Sky News in The UK, he rejected the idea of an AI bubble saying, We're a long ways away from that.

我确实认为这只是建设阶段的开始，我们正在见证从传统计算方式向人工智能的平台转型。

I really think this is the beginning of the build out, and we're seeing a platform shift from the traditional way of doing computing to artificial intelligence.

当某样东西有利可图时，供应商就会加大生产。

When something is profitable, the suppliers want to make more of it.

这就是AI建设加速的原因——因为AI现在生产力极高、利润丰厚且用户广泛。

That's the reason the AI build out is accelerating because AI is now so productive, so profitable, and used by so many people.

虽然很多人将巴里的交易视为泡沫破裂的确凿信号，但也有人指出他的判断并非总是可靠。

Now while many are latching onto Burry's trade as a surefire sign the bubble is bursting, others noted that he isn't always a reliable signal.

这位做空熊市博主写道：'我尊重巴里，但要记住从他开始做空到崩盘用了两三年时间'。

The short bear posted I respect Burry, however let's remember it took two to three years from the moment he started shorting until the collapse.

值得注意的是，巴里曾在2015、2017、2019、2020、2021和2023年多次预言市场崩盘。

It is also worth noting that Burry called for crashes in 2015, 2017, 2019, 2020, 2021, and 2023.

正如彼得·马卢克十月份指出的，自迈克尔·巴里2023年喊卖出以来，标普500已上涨71%并创下88次新高。

As Peter Malouk pointed out at the October, the S and P five hundred is up 71% and has hit 88 all highs since Michael Burry said sell back in 2023.

还需谨记，巴里并非用自有资金做空市场。

It's also worth keeping in mind that Burry isn't shorting the market with a billion dollars of his own money.

投资者将资金投入巴里的对冲基金，正是看中他做空泡沫的声誉。

Investors put their money with Burry's hedge funds specifically because of his reputation for shorting bubbles.

关键在于，如果他不做空Palantir和英伟达这类公司，反而算是失职。

Point The is, he wouldn't be doing his job if he wasn't shorting the likes of Palantir and Nvidia at some point.

尽管如此，围绕AI的市场正变得越来越有趣。

Still, markets around AI are getting more interesting.

德意志银行正在考虑做空AI股票，以此对冲他们在AI数据中心的风险敞口。

Deutsche Bank is considering shorting AI stocks as a way to hedge their exposure to AI data centers.

这家德国银行已向数据中心项目提供了数十亿美元的贷款，一位高管表示他们在这个主题上押下了重注。

The German bank has extended billions of dollars in loans to data center projects with one executive stating that they've bet big on the theme.

目前，高层正在就银行如何对冲风险敞口展开讨论。

Now, high level conversations are underway on how the bank can hedge their exposure.

可选方案包括使用名为合成风险转移（SRT）的衍生品为部分债务购买违约保护。

Options include buying default protection on some of the debt using derivatives called synthetic risk transfers or SRTs.

据报道，他们也在考虑直接做空一篮子与AI相关的股票。

They're also reportedly looking to simply short a basket of stocks associated with AI.

德意志银行主要向亚马逊、微软和谷歌等超大规模企业提供贷款，但他们正逐渐向下游延伸，也开始为小型新云服务商提供贷款。

Deutsche has primarily lent to hyperscalers like Amazon, Microsoft, and Google, but they're increasingly going down the stack and lending to smaller neoclouds as well.

这在一定程度上反映了AI建设阶段的变化。

Now part of what this is reflective of is a changing of the phase for the AI build out.

迄今为止，AI基础设施主要由超大规模企业提供现金流支持，但建设规模不断扩大将需要债务融资来维持。

Until now, AI infrastructure has largely been cash flowed by the hyperscalers, but the size of the build out increasingly will require debt financing to continue.

黑石集团科技全球主管Tony Kim本周在接受采访时表示：毫无疑问，面对AI所需的数万亿美元资本支出，企业将需要进入债务市场来为这种扩张提供资金。

Speaking with the information this week, BlackRock's global head of tech Tony Kim said: There is no doubt that with the trillions of dollars of capex required for AI, companies will need to tap into debt markets to fund this expansion.

Kim认为这是向前发展的必要条件，并评论道：科技公司将不得不摒弃对杠杆的厌恶。

For his part, Kim believes this is a necessity to move forward, commenting: Tech companies will have to shed their aversion to leverage.

Kim认为，由于这些公司目前杠杆率极低，在这方面还有很大的操作空间。

Now for his part, Kim thinks that because these companies are so unlevered right now, there's a lot of room to run with that.

但这绝对是值得关注的事情。

But this is certainly something to keep an eye on.

我们今天将以两个与Perplexity相关的故事作为结束。

We'll close out today with two stories related to Perplexity.

首先，Snap已签署一项重大协议，将Perplexity整合至其平台。

First, Snap has signed a huge deal to integrate Perplexity into their platform.

根据协议，Perplexity将支付4亿美元现金加股权，以获得使用Snapchat作为分发渠道的权利。

The deal will see Perplexity paying $400,000,000 in cash in equity for the ability to use Snapchat as a distribution channel.

收入分成协议将于明年生效。

A revenue sharing agreement will kick in next year.

尽管Snapchat可能落后于TikTok和Instagram，但它仍拥有近5亿日活跃用户。

While Snapchat might be behind TikTok and Instagram, it still has almost half a billion daily active users.

Perplexity将被集成到聊天功能中，根据新闻稿，其目标是在Snapchat内提供来自可验证来源的清晰对话式答案。

Perplexity will be integrated into the chat function and according to a press release, the aim is to deliver clear conversational answers drawn from verifiable sources all within Snapchat.

Snap首席执行官Evan Spiegel表示：'我们的目标是让人工智能更个性化、社交化和趣味化，融入用户的友谊、快拍和对话中。'

Snap CEO Evan Spiegel said, Our goal is to make AI more personal, social, and fun, woven into the fabric of your friendships, Snaps, and conversations.

总体而言，Snap正处于十字路口，通过'故事'和公共内容推送等功能，正与其他社交媒体平台趋同。

Now Snap, generally speaking, is at a crossroads, becoming closer to the other social media platforms with features like Stories and public content feeds.

他们还大力推动广告业务，上季度广告收入增长了8%。

They've also made a big push into advertising with ad revenue surging 8% last quarter.

虽然许多人质疑Snap的年轻用户群体是否适合Perplexity，但市场看好这笔交易，消息公布后Snap股价在盘后交易中上涨25%。

While many wondered if Snap's young audience was a good fit for Perplexity, markets liked the deal and Snap stock was up 25% in aftermarket trading following the announcement.

对Perplexity而言稍显不利的是，亚马逊正就其数据抓取行为提起诉讼。

In a slightly tougher story for Perplexity, Amazon is suing them over their data scraping practices.

周二，亚马逊向法院起诉Perplexity，试图阻止其代理程序访问亚马逊电商平台。

On Tuesday, Amazon filed a lawsuit against Perplexity in an attempt to block Perplexity's agents from accessing their e commerce platform.

据报道，亚马逊已于上周五通过停止侵权警告函提出投诉。

Amazon reportedly laid out their complaints in a cease and desist last Friday.

亚马逊声称Perplexity的网络爬虫未标明其与AI代理的关联身份。

They claimed that Perplexity's web crawlers failed to identify themselves as associated with an AI agent.

Perplexity本周通过题为《霸凌不是创新》的博客文章予以反击。

Perplexity fired back this week in a blog post called Bullying is not Innovation.

他们写道：过去五十年间，软件一直是一种工具，如同用户手中的扳手。

They wrote: For the last fifty years, software has been a tool, like a wrench in the hands of the user.

但随着自主人工智能的兴起，软件也正在演变为劳动力：成为助手、雇员或代理。

But with the rise of agentic AI, software is also becoming labor: an assistant, an employee, an agent.

法律明确规定，大型企业无权阻止你拥有扳手这类工具。

The law is clear that large corporations have no right to stop you from owning wrenches.

今日亚马逊宣称，它不认可你雇佣劳动力、聘请助手或雇员代为行事的权利。

Today, Amazon announced it does not believe in your right to hire labor, to have an assistant or an employee acting on your behalf.

这不是一个合理的法律立场。

This isn't a reasonable legal position.

这是恐吓像Perplexity这样颠覆性企业的霸凌手段，阻挠它们为人们创造更美好的生活。

A bully tactic to scare disruptive companies like Perplexity out of making life better for people.

亚马逊在回应博文中指出，其他公司的第三方代理会明确标识身份，并写道：我们认为第三方应用程序若代表客户在其他商家处进行采购，理应公开运作并尊重服务商是否参与的选择。

Amazon noted that third party agents from other companies identify themselves as such, writing in a response blog post: We think it's fairly straightforward that third party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions whether or not to participate.

他们在诉讼文件中更进一步写道：与任何入侵者无异，Perplexity不得进入被明确禁止的领域。

They took it a little further in their lawsuit writing: No different than any other intruder, Perplexity is not allowed to go where it has been expressly told it cannot.

Perplexity的侵权行为使用代码而非撬锁工具，但这丝毫不减其违法性。

That Perplexity's trespass involves code rather than a lockpick makes it no less unlawful.

当然，当前这场风波的有趣之处不在于两家科技公司的心理战，而在于它预示了关于自主购物更广泛的斗争。

Now, the dust up is interesting, of course, not as some psychodrama between two tech companies, but as a preview of a broader fight around agentic shopping.

多个实验室正准备在今年黑色星期五大促中释放自主代理，但作为最大电商平台，亚马逊目前动动手指就能关闭所有人的权限。

Multiple labs are preparing to let agents loose on this year's Black Friday sales, but Amazon, as the largest e commerce platform, can currently shut everyone down with a flick of a switch.

因此，这不过是更大战役中的一场小冲突。

So chalk this up as a skirmish in a larger battle.

不过就目前而言，今天的头条新闻就到这里。

For now, though, that is gonna do it for today's headlines.

接下来进入正片内容。

Next up, the main episode.

本期节目由超级智能为您呈现。

Today's episode is brought to you by Super Intelligent.

对于新来的或不了解的朋友们，超级智能其实是我的公司。

Now, for those of you who don't know who are new here, maybe Super Intelligent is actually my company.

我们创立它是因为每当我们与企业交流时，发现大家都在探索AI能做什么，但大多数建议都过于泛泛，没有针对性。

We started it because every single company we talk to, all the enterprises out there, are trying to figure out what AI can do for them, but most of the advice is super generic, not specific to your company.

因此我们通过部署语音助手访谈您的团队，了解当前工作流程和未来期望，从而为您绘制AI与智能代理的应用机会。

So what we do is we map your AI and agent opportunities by deploying voice agents to interview your teams about how work works now, and how your people would like it to work in the future.

最终成果是一张AI行动地图，包含高投资回报率用例和具体变革管理需求——基本上涵盖了实现AI价值所需的一切。

The result is an AI action map with high potential ROI use cases and specific change management needs, Basically everything you need to go actually deliver AI value.

访问bsuper.ai了解更多。

Go to bsuper.ai to learn more.

认识您的新AI队友Robo。

Meet Robo, your AI powered teammate.

Robo通过AI驱动的搜索、聊天和代理功能释放团队潜力，您还可以用Studio构建专属代理。

Robo unleashes the potential of your team with AI powered search, chat, and agents, or build your own agent with Studio.

Robo由您组织的知识驱动，运行在Atlassian安全可信的平台上，始终与您的工作场景无缝衔接。

Robo is powered by your organization's knowledge and lives on Atlassian's trusted and secure platform, so it's always working in the context of your work.

将Robo与您喜爱的SaaS应用连接，确保知识永不遗漏。

Connect Robo to your favorite SaaS app so no knowledge gets left behind.

Robo基于Teamwork Graph运行——这是Atlassian的智能层，能整合所有应用数据并从第一天起提供个性化AI洞察。

Robo runs on the Teamwork Graph, Atlassian's intelligence layer that unifies data across all of your apps and delivers personalized AI insights from day one.

Robo已内置在Jira、Confluence及Jira服务管理（标准版/高级版/企业版）订阅中。

Robo is already built into Jira, Confluence, and Jira Service Management Standard, Premium, and Enterprise subscriptions.

可曾感受过AI从工具转变为队友的瞬间？

Know the feeling when AI turns from tool to teammate?

若您拥有Robo，自然懂得。

If you Robo, you know.

认识Robo，您由Atlassian驱动的新AI队友。

Discover Robo, your new AI teammate powered by Atlassian.

访问rov,asinvictory,0.com开始使用。

Get started at rov,asinvictory,0.com.

AI发展迅猛。

AI changes fast.

您需要一个为长期合作而生的伙伴。

You need a partner built for the long game.

Robots and pencils与企业并肩工作，将AI抱负转化为真实的人类影响。

Robots and pencils work side by side with organizations to turn AI ambition into real human impact.

作为AWS认证合作伙伴，他们推动基础设施现代化，设计云原生系统，并运用AI创造商业价值。

As an AWS Certified Partner, they modernize infrastructure, design cloud native systems, and apply AI to create business value.

而且他们的合作关系不会止步于项目启动。

And their partnerships don't end at launch.

当AI技术迭代时，robots and pencils始终与您同行，助您保持领先。

As AI changes, robots and pencils stays by your side, so you keep pace.

区别在于紧密的合作伙伴关系能持续创造复合价值。

The difference is close partnership that builds value and compounds over time.

此外，通过遍布美国、加拿大、欧洲和拉丁美洲的交付中心，客户既能获得本地化专业支持，又能享受全球规模优势。

Plus, with delivery centers across The US, Canada, Europe, and Latin America, clients get local expertise and global scale.

要实现AI的实际进展而非空头承诺，请访问robotsandpencils.comaidailybrief。

For AI that delivers progress, not promises, visit robotsandpencils.comaidailybrief.

本节目由Blitzy赞助——拥有无限代码上下文的企业级自主软件开发平台。

This episode is brought to you by Blitzy, the enterprise autonomous software development platform with infinite code context.

Blitzy运用数千个专业AI代理，经过数小时思考来理解数百万行代码规模的企业级代码库。

Blitzy uses thousands of specialized AI agents that think for hours to understand enterprise scale codebases with millions of lines of code.

企业工程领导者通过Blitzy平台开启每个开发冲刺周期，输入他们的开发需求。

Enterprise engineering leaders start every development sprint with the Blitzy platform, bringing in their development requirements.

Blitzy平台会先制定计划，然后为每项任务生成并预编译代码。

The Blitzy platform provides a plan, then generates and pre compiles code for each task.

Blitzy能自主完成80%以上的开发工作，同时为冲刺阶段所需的20%人工开发工作提供指导。

Blitzy delivers 80% plus of the development work autonomously, while providing a guide for the final 20% of human development work required to complete the sprint.

上市公司将Blitzy作为集成开发环境前的开发工具，搭配自选的编码助手后，工程效率提升了5倍，实现了AI原生的软件开发生命周期。

Public companies are achieving a 5x engineering velocity increase when incorporating Blitzy as their pre IDE development tool, pairing it with their coding pilot of choice to bring an AI native SDLC into their org.

访问blitzy.com并点击'申请演示'，了解Blitzy如何将您的SDLC从AI辅助升级为AI原生。

Visit blitzy.com and press Get a Demo to learn how Blitzy transforms your SDLC from AI assisted to AI native.

欢迎回到AI每日简报。

Welcome back to the AI Daily Brief.

若想了解当前AI领域的整体舆论风向，只需对比过去半年两所常春藤大学发布的研究报告的反响差异。

If you are looking for a way of understanding where the general sentiment around AI is right now, Look no further than the difference in reception to two studies from Ivy League universities that have come out over the last six months.

我们整个夏天都在处理——说实话至今仍在应对——那个荒谬的MIT所谓'研究'（如果这也能称为研究的话），它随意访谈了52位高管，翻看企业财报中关于AI创收的声明，就妄称95%的AI项目都失败了。

We spent a huge portion of this summer, and frankly continue to have to deal with that inane MIT air quote study, if you can even call it that, that interviewed 52 executives seemingly chosen for convenience of being around and looked at public earnings statements for companies saying that they were seeing new profitability from their AI initiatives, all in the way to proclaiming that 95% of AI initiatives were failing.

这个统计数据被无数分析报告、媒体报道、博客文章和商业提案反复引用。

That statistic has been included in so many analyses and media pieces and blog posts and pitches.

说实话，尽管令人沮丧，这很可能是今年AI领域被引用最多、传播最广的'研究'。

It is honestly at this point, much to my chagrin, probably the most quoted, most ubiquitous study around AI shared this year.

与此同时，沃顿商学院采用更全面可靠学术方法开展的纵向研究已进入第三年，调查了约800位跨职能企业高管，却几乎无人问津。

Meanwhile, a longitudinal study from Wharton on its third year with a much more comprehensive and verifiable academic methodology that surveyed around 800 enterprise leaders across a variety of functions has gotten barely any attention at all.

讽刺的是，就AI的长期发展而言，行业集体关注怀疑论研究而非乐观研究或许是件好事——这说明尽管有泡沫论者的喧嚣，当前舆论远未过热。

Now ironically, it is probably the case that when it comes to the longevity of AI, it is probably a net better thing for the industry that everyone latched on to the skeptical study as opposed to the optimistic study because it shows that despite all the bubble screamers, the narrative at least is very, very far from overheated right now.

但无论如何，当我们超越对AI讨论的元分析后，这份沃顿研究仍包含大量极具价值的发现，非常值得细读——这也正是我们今天的主题。

But in any case, when we move beyond the meta interpretations of where the AI discourse is, there is still a ton of really valuable stuff in this Wharton study that very much deserves a review, and that is what we are doing today.

如我所言，这是沃顿GBK第三次年度企业生成式AI应用调研。

As I said, this is the third annual Wharton GBK study of Enterprise Gen AI adoption.

总体而言，研究描绘了AI成为主流技术、应用无处不在并融入日常生活，投资回报不仅开始可衡量且已显现成效的图景。

And the story, by and large, is of an AI moving mainstream, adoption becoming ubiquitous and integrated into the fabric of everyday life, and ROI not only beginning to be measured but also showing up.

一个重要主题，我们称之为日常AI或AI从探索阶段转向核心工作流程。

The big theme one, let's call everyday AI or AI moving from curiosity to core workflow.

这个主题本质上是说生成式AI已成为日常工作的一部分，而非实验性质。

The theme is basically that GenAI is now part of daily work, not an experiment.

82%的企业领导者每周使用生成式AI，近半数决策者（46%）报告称每天使用。

82% of enterprise leaders now use GenAI weekly and almost half of decision makers (forty six percent report using GenAI daily.

相比去年上升了17个百分点。

That's up 17 percentage points versus last year.

对生成式AI的认知度和熟悉度有所提升。

Knowledge and familiarity with GenAI has risen.

77%受访者表示至少对生成式AI有所了解，尽管营销和管理部门存在轻微滞后。

77% report being at least somewhat familiar with GenAI, although there are slight laggards in marketing and management.

我们开始看到功能性应用模式，关键业务任务中的生成式AI采用率更高。

We're starting to see functional adoption patterns where key business tasks are seeing higher Gen AI adoption.

营销内容创作增长，内部支持与帮助台服务增长，文档与会议摘要应用增长。

Marketing content creation is up, internal support and help desk is up, Document and meeting summarization is up.

演示文稿与报告制作增长。

Presentation and report creation is up.

创意生成与头脑风暴增长。

Idea generation and brainstorming is up.

数据分析与商业智能增长。

Data analysis and analytics are up.

虽然AI有诸多优势，但十大生成式AI用例中有半数直接提升员工生产力。

And while there are lots of different benefits of AI, half of the top 10 Gen AI use cases directly boost employee productivity.

2025年十大用例依次为：数据分析与商业智能、文档与会议摘要、文档与提案编辑撰写、演示文稿与报告制作、创意生成与头脑风暴、营销内容创作、客户服务与支持、邮件生成、内部支持与帮助台、销售内容创作。

Top 10 use cases in 2025 are in order: Data Analysis and Analytics, Document and Meeting Summarization, Document and Proposal Editing and Writing, Presentation and Report Creation, Idea Generation and Brainstorming, Marketing Content Creation, Customer Service and Support, email generation, internal support and help desk, and sales content creation.

不过有趣的是，AI智能体正开始崭露头角。

Interestingly though, AI agents are starting to emerge.

58%的企业正在测试AI代理，主要应用于流程自动化、数据分析和工作流编排等领域。

58% of enterprises are testing AI agents, mostly among this cohort, for process automation, analytics, and workflow orchestration.

然而这份调查的重点并非使用情况，而是投资回报率(ROI)。

And yet really, the big story of this survey is not about usage but about ROI.

企业正明显从使用阶段转向验证阶段。

Enterprises are very clearly shifting from use to proof.

首先，ROI评估已成为标准流程。

First of all, ROI measurement has become standard.

72%的企业正在系统追踪生成式AI的投资回报。

72% of companies are formally tracking their Gen AI ROI.

在结构化ROI追踪方面领先的部门包括人力资源(84%)和财务(80%)。

The functions that lead in structured ROI tracking including HR at 84% and finance at 80%.

可能是整份报告最重磅的数据：四分之三的企业报告了正向ROI。

Maybe the biggest headline or statistic of the whole report: three fourths of enterprises report positive ROI.

74%的企业获得了中度或显著正向回报，其中年收入在50-20亿美元的中型企业比年收入超20亿的大型企业获得更高ROI。

74% overall are seeing either moderately positive or significantly positive ROI, with smaller firms between 50 and 2,000,000,000 in revenue seeing more ROI so far than enterprises with 2,000,000,000 plus in annual revenue.

过去几个月我们看到大量迹象表明，ROI的实现速度远超许多人预期。

Now we have seen over the past couple of months just a slew of indications that ROI is coming faster than people in many cases would have thought.

而且市场对ROI的预期持续上升。

And the perception of ROI continues to rise.

那么主要障碍会是什么？

So what are going to be the big blockers?

正如我们反复看到的，问题更多在于人而非技术。

Well, as we've seen over and over again, it's going to be more about people than employees.

虽然89%的受访者认为AI提升了技能，但仍有43%担心技能退化。

While 89% of respondents said that AI enhanced skills, 43% still fear skill decline as well.

尽管绝大多数决策者表示过去一年对生成式AI更加乐观，但他们仍保持谨慎态度。

And while overall, the vast majority of decision makers are saying they feel more positive about Gen AI over the past year, they are also still cautious as well.

这一切让我清晰地认识到，2026年的核心主题将不仅是衡量和展示投资回报率，而是要理解其相对表现。

What all of this sets up very clearly in my mind is that 2026 where the key theme of the year is going to be not only about measuring ROI and demonstrating ROI, but about understanding how it compares.

目前我们已收集到700多个AI投资回报率基准研究的应用案例，获得了极其详尽的关于效益来源的数据。

We're now up over 700 use cases that have been contributed to the AI ROI benchmarking study and getting just a huge degree of granular information around where the benefits are really coming.

我发起这项研究的部分原因是希望建立基准，让企业能理解：第一，他们应该从AI中获得何种效益；第二，

Part of why I wanted to launch this study is that I want to start to have benchmarks where organizations can understand, A, what type of benefit they're supposed to get out of AI, And B.

他们看到的结果是否确实与同行和同事相当。

Whether the results that they're seeing are actually commensurate with their peers and colleagues.

关于第一部分，尽管我们常讨论AI与生产力，但AI实际能带来的效益和影响是多种多样的。

On the first part, as much as we talk about AI and productivity, there are actually a variety of different types of benefits and impact that AI can have.

包括时间节省、成本节约、新能力、提升吞吐量和产出、降低风险、改进决策、增加收入、新收入来源等，理解哪些应用场景对应哪些效益类型至关重要。

There's time savings, cost savings, new capabilities, enhanced throughput and output, reduced risk, improved decision making, enhanced revenue, new revenue lines, and understanding which use cases are relevant for which of those different types of benefits is really important.

其次，由于我们都处于新领域，不仅需要知道AI是否在改进现状，还要确认这种改进是否达到预期水平。

Second, because we're all floating in new territory, we don't need to just know whether AI is improving things, but whether it's improving things in a way that's commensurate with what we would expect.

例如，如果某个部署帮助团队将营销产出提高10%，这看似不错——直到你发现竞争对手的平均提升幅度是20%。

For example, if a particular deployment is helping your team increase marketing throughput by 10%, that might seem great until you find out that for all of your competitors, it's increasing marketing output by an average of 20%.

归根结底，这项ROI调查只是贯穿明年整个大主题中非常小的组成部分。

Look, ultimately the ROI survey is just one very small part of what is going to be a big theme for all of next year.

但如果你希望获取我们从中发现的所有信息，请访问roisurvey.ai参与贡献，本月底完成时你将获得结果。

But still, if you want access to all the information that we find from that, go to roisurvey.ai, contribute, and you will get the results when they are completed towards the end of this month.

让我们回到沃顿商学院的综述。

Bringing it back to Roundup on Wharton.

他们对2026年有什么预期？

What do they think about what 2026 will bring?

研究作者写道：2026年可能是从责任加速转向规模绩效的转折点，当今的ROI指标、操作手册和防护措施将让企业重构核心工作流，部署智能代理系统，并将预算重新分配给已验证的回报领域。

The study authors write: 2026 could be the turn from accountable acceleration to performance at scale, where today's ROI metrics, playbooks, guardrails let enterprises rewire core workflows, deploy agentic systems, and reallocate budgets towards proven returns.

他们指出，五分之四的人认为生成式AI投资将在两到三年内获得回报。

They point out that four out of five see Gen AI investments paying off in about two to three years.

88%的企业预计在未来十二个月内会增加生成式AI的预算。

88% anticipate increasing Gen AI budgets in the next twelve months.

每个人都在努力寻找如何培养或获取这个新时代所需的人才。

And everyone is trying to figure out how to make or get the talent that's required for this new era.

这就是沃顿商学院研究的故事——乐观、兴奋和投资回报率逐渐清晰。

So that is the story of the Wharton Study, optimism, excitement, and ROI coming into focus.

今天的AI每日简报就到这里。

For now, that's gonna do it for today's AI Daily Brief.

一如既往感谢您的收听，下次再见，祝您平安。

Appreciate you listening as always, and until next time, peace.