1万亿美元的人工智能建设 | 市场现状

全球市值排名前十的公司中，有八家是美国科技公司。

Eight of the top 10 valued companies in the world are US tech companies.

自从ChatGPT大约四年前问世以来，市场上涨了90%，年化收益率达到17%。

Since ChatGPT came out almost four years ago, the market's up 90%, which is 17% annualized.

人们自然会想，这肯定得回落吧。

The natural instinct is, well, that's got to come down.

我们确实处于一个火热的时期。

We're definitely in a hot period.

这让我们进入了一个原子的新时代。

This puts us in a new age of atoms.

到2040年，全球基础设施投资需求估计高达90万亿美元。

Global infrastructure investment needs are estimated at $90 trillion through 2040.

这远远超出了人工智能和数据中心的范畴。

This goes way beyond AI and data centers.

它涵盖了电力、水资源、道路和交通系统。

It includes power, water, roads, transit.

标普500公司的实际部署率已经达到69%。

Live deployments at S&P 500 companies, that's at 69%.

但如果看最根本的指标，也就是长期追踪的那个数据，其实只有2%。

Now, if you go to the ultimate barometer, which is a metric tracked over time, that's actually only at 2%.

AI正在带来可观的收入和节省。

AI is generating major revenue and savings.

但另一方面，采用率还处于非常早期的阶段。

On the other hand, adoption is still extremely early.

现在的机会主要在于利用这些能力，把它们转化为可靠的服务。

Today's opportunity is so much around taking these capabilities and harnessing them to build reliable services.

这些都很好，但这是泡沫吗？

That's all great, but is it a bubble?

AI正在推动现代历史上最大的投资周期之一。

AI is driving one of the largest investment cycles in modern history.

但支出是否已经超过了经济回报？

But is the spending getting ahead of the economics?

今天，David George、Sarah Wang、Santiago Rodriguez 和 Alex Immerman 将带我们浏览我们最新一期《市场状况》报告中的25张关键图表。

Today, David George, Sarah Wang, Santiago Rodriguez, and Alex Immerman walk through 25 key charts from our latest State of the Markets presentation.

他们先从宏观图景讲起，解释为什么市场在估值倍数下降的情况下仍然上涨，超大规模云服务商的资本支出如何逼近每年一万亿美元，以及为什么对计算能力的需求持续超过供给。

They start with the macro picture, why markets have risen even as multiples have come down, how hyperscaler CapEx is approaching $1 trillion a year, and why demand for compute continues to exceed supply.

然后，他们看看企业内部正在发生什么。

Then they look at what's happening inside businesses.

AI 的采用在增长，但可衡量的企业影响却出人意料地处于早期阶段。

AI adoption is growing, but measurable enterprise impact is surprisingly early.

与此同时，重度用户正在拉开差距，智能体消耗的令牌数更多，推理成本在下降，而行动最快的公司正在找到方法，将这些改进转化为增长和更好的利润率。

At the same time, power users are pulling away, agents are consuming more tokens, inference costs are falling, and the fastest-moving companies are finding ways to turn those improvements into both growth and better margins.

最后，他们展望未来。

And finally, they look at what comes next.

从消费级智能体和机器人技术，到自动驾驶、AI 和生物科技，再到下一波企业采用浪潮。

From consumer agents and robotics to autonomy, AI, and biology, and the next wave of enterprise adoption.

欢迎回到 a16z 播客。

Welcome back to the a16z podcast.

我是大卫·乔治。

I'm David George.

今天和我一起的是我的同事莎拉·王、亚历克斯·伊默曼和圣地亚哥·罗德里格斯。

I'm here with my colleagues Sarah Wang, Alex Immerman, and Santiago Rodriguez.

今天，我们将一起浏览我们最新市场状况报告中的25张关键幻灯片。

Today, we are walking through 25 key slides from our latest State of Markets presentation.

包括市场上涨背后的盈利、AI建设的规模、采用增长的证据，以及这个周期对硬件、软件和下一代私营公司意味着什么。

The earnings behind the market's rise, the scale of the AI build-out, the evidence of growing adoption, and what this cycle means for hardware, software, and the next generation of private companies.

我们会解释这些图表显示的内容，并讨论我们在企业内部的所见所闻，这样无论你是观看还是收听，都能跟上。

We'll explain what the charts show and discuss what we're seeing inside businesses along the way, so you can follow along whether you're watching or listening.

那么，莎拉、亚历克斯、桑蒂，谢谢你们加入我。

So Sarah, Alex, Santi, thanks for joining me.

当然。

Of course.

很高兴来到这里。

Glad to be here.

好的，这期播客上线的那天，我们会发布我们的《市场状况》报告。

Okay, so the day this podcast goes live, we will be releasing our State of Markets presentation.

这是我们增长团队的年度传统，汇总科技、人工智能、基础设施和市场方面的最大趋势。

So this is a now yearly tradition from our growth team, where we synthesize the biggest trends in tech, AI, infra, and markets.

今天，我们只是挑选一些有趣的幻灯片，然后进行讨论。

Today, we're just picking out a subset of interesting slides and having a discussion about them.

所以，我建议你们去看完整版，里面有很多干货。

So I would direct you to look at the whole version, which is filled with a lot of nuggets.

我们今天要讨论的领域包括宏观方面，比如资本支出、数据中心、高层次的加速需求，然后是实地观察，我们会讨论模型、应用和一些垂直领域的深入分析。

The areas that we're going to discuss today are macro, so where we'll talk about CapEx, data centers, accelerating demand at a high level, and then on-the-ground takeaways, where we will discuss models, apps, and some vertical deep dives.

科技正在推动整个经济的投资热潮，盈利增长支撑了市场上涨，而人工智能需求则让基础设施投资远超早前的预测。

Technology is driving an economy-wide investment boom, with rising earnings supporting market gains and AI demand pushing infrastructure spending far beyond earlier forecasts.

超大规模云服务商正在把近期的全部运营现金流投入到产能建设中，以满足持续超过供应的需求，这在我们看到的所有案例中几乎都是如此。

Hyperscalers are investing all of their near-term operating cash flow to build out this capacity for demand that continues to outstrip supply in almost every case that we see.

他们把这些投资用于芯片、电力、冷却、建筑、熟练劳动力等等。

They are channeling this investment into chips, power, cooling, construction, skilled labor, and much more.

这一扩张恰逢更广泛的物理基础设施现代化需求，为各行各业创造了机会，并可能降低企业和家庭的共同成本。

This expansion coincides with a broader need to modernize physical infrastructure, creating opportunities across industries and potentially lowering shared costs for businesses and households.

那么，我们就开始吧。

So with that, let's jump in.

好的。

Okay.

第一张幻灯片，科技就是一切周期。

First slide, tech is the everything cycle.

有些内容可能显而易见，但现在的数字确实相当惊人，对吧？

So some of this may be a little bit obvious, but the numbers are somewhat striking at this point, right?

高科技设备、软件和研发现在大约占美国资本支出的55%，这真是个惊人的数字。

So high-tech equipment, software, and R&D now account for roughly 55% of US capital spending, which is just a staggering number.

科技显然在推动整个经济的投资周期，从软件和模型到电力建设和工业产能。

Tech is driving the investment cycle, obviously across all areas of the economy, from software and models to power construction and industrial capacity.

科技几乎占美国股市总市值的40%，全球市值前十的公司中有八家是美国科技公司。

Tech is almost 40% of the aggregate value of the whole stock market in the US, and eight of the top 10 valued companies in the world are US tech companies.

所以这是一个很广泛的故事。

So this is a broad story.

你在资本支出上就能看到这一点。

You're seeing it in CapEx.

这在数据中心建设中被大量报道。

This is heavily covered in the data center build-out.

显然，在用于资助模型开发的资金量上也有大量报道。

It's obviously heavily covered in the amount of capital that's been going in to fund the model development.

模型公司总共筹集了，我想，超过3500亿美元。

Model companies together raising, I think, over $350 billion at this point.

而且，为了把它放到非常深远的历史背景中，因为这可能是我们见过最多的类比，这次建设在GDP中的占比已经超过了铁路。

And just to put it into very deep historical context, because this is probably the analogy that we've seen the most of, this build-out just surpassed railroads as a percentage of GDP.

所以是激动人心的时代，大规模的建设。

So exciting times, massive build-out.

我们都认为，如果你快进五到七年，也许十年后，我们再来看这些数字，它们累计可能会高出20倍。

We all happen to think that if you fast-forward five to seven, maybe 10 years from now, we'll be looking at these numbers, and they'll probably be 20x higher cumulatively.

所以——

So-

这确实像是马克的“软件吞噬世界”的下一章。

This does feel like the next chapter of Mark's "Software is Eating the World."

过去15年，软件改变了所有这些行业，但只有一小部分人能够构建它。

Over the last 15 years, software has transformed all these industries, but only a small fraction of the population could build it.

我们谁都不会。

None of us could.

但今天，看看我们。

But today, look at us.

我们每个人每晚都有几个自动化在运行。

We all have a handful of automations running every night.

所以如果你想想软件需求，那意味着更多的计算、芯片、电力、建设，所以看到大部分投资现在都集中在科技领域也就不足为奇了。

And so if you think about the software demand, that means a lot more compute, chips, power, construction, and so it's no surprise to see the majority of investment now covered in tech.

好的，所以我们经常从不同听众那里收到的一个问题是，好吧，这一切都很好，但这是泡沫吗？

Okay, so one of the questions that we get all the time from various audiences is, okay, that's all great, but is it a bubble?

所以市场创下了新高，但与此同时，市场的交易市盈率实际上是在下降的。

So the market has reached new highs, and at the same time that it has reached new highs, the trading multiples of the market are actually down.

所以股票上涨了大约20%，而市盈率却下降了大约20%。

So stocks are up about 20%, while multiples are down about 20%.

所以这意味着表现是由基本面盈利驱动的，对吧？

So what that means is the performance is driven by fundamental earnings, right?

而不是市盈率的上升。

Not increased multiples.

标普500的盈利市盈率低于20倍。

The S&P 500 earnings multiple is below 20 times.

所以如果你从这个角度看，这些在大多数情况下都是非常优质的企业。

So if you just start with that, these are, in most cases, very high-quality businesses.

这跟互联网泡沫时期完全不一样，那时候全球市值最高的公司，股价大涨靠的是市盈率的飙升，很多情况下市盈率能达到100倍。

It's nothing like the dot-com boom in that way, where some of the highest market cap companies in the world had their massive stock run-ups based on increases in their trading multiples and would trade, in many cases, for 100 times PE.

现在的情况不是这样。

That's not what's happening here.

相比之下，一些存储公司，它们也是周期性很强的，目前的交易价格大概是未来收益的六倍或七倍。

In contrast, some of the memory companies, which again are very cyclical, are trading for, call it six times, seven times forward earnings.

所以情况非常不同。

So very different.

而且我觉得这里有个重要的点，自从ChatGPT大约四年前推出以来，市场上涨了90%，相当于年化17%。

And I think there's an important double click here where since ChatGPT came out almost four years ago, the market's up 90%, which is 17% annualized.

我觉得任何时候出现连续四年17%的年化增长，人们自然会想，这肯定得回落，对吧？

And I think any time there's been a 17% annualized growth for four years, the natural instinct is, well, that's got to come down, right?

我们确实处于一个火热的时期。

We're definitely in a hot period.

但当你把它和你说的情况对比，市场以低于20倍的市盈率交易，增长15%，这确实感觉和最近的2021年或者2000年那段时期有点不同，那时候市盈率和增长并不匹配。

But when you compare that to, as you said, the market trading for below 20 times earnings growing 15%, this definitely feels a little different than maybe the 2021 period that was recent or the 2000 period when multiples and growth were not really going together.

对，完全同意。

Yeah, totally agreed.

所以我提到了资本支出建设的规模，是在铁路的背景下说的。

So I mentioned the scale of the CapEx build-out in the context of the railroads.

你看看那些超大规模企业，像Alphabet、亚马逊、Meta、微软和甲骨文，它们2026年的资本支出大约是7800亿美元。

If you just look at the hyperscalers Alphabet, Amazon, Meta, Microsoft, and Oracle, their CapEx in 2026 is about $780 billion.

这比2025年的4160亿美元有所增长，而且所有预期都指向它们从2027年开始每年支出超过一万亿美元。

That's up from $416 billion in 2025, and all expectations point to them spending over a trillion dollars annually from 2027.

所以，从模式上看，每一个计算平台都支持了更大规模的用户和使用场景。

So the pattern recognition is each of these computing platforms has supported a much larger population of users and uses.

在这种情况下，这种建设可以非常迅速地进行，但需求仍然可能超过供应，因为用户数量是由已有的分发渠道驱动的。

In this case, this build-out can happen so quickly, and demand can still outstrip supply because the amount of users is driven by the fact that there's already existing distribution.

这个趋势显然是建立在互联网、云计算和手机之上的，所以它能够迅速触达数十亿用户，这与之前的技术周期形成鲜明对比。

This trend is built on top of the internet and cloud computing, obviously, and mobile phones, and so sort of immediately could reach billions of users in contrast to previous technology cycles.

是的，我觉得这一点值得深入探讨，因为我们已经远远超出了那种偶尔查询的模式，对吧？

Yeah, and I think it's an important point to double-click on, just because we've moved well beyond this model of occasional queries, right?

想想智能体的兴起，你有并行任务，有长时间运行的任务。

If you think about the rise of agents, you have parallel tasks, you have long-running tasks.

所以，亚历克斯，你之前提到过这一点，对吧？

And so, I mean, Alex, you mentioned this previously, right?

有些任务会持续到晚上，白天你还在做其他事情的时候也在进行。

You have tasks going into the evening, during the day, when you're doing other things.

所以，如果你想想智能体自主写代码、搜索、执行这些任务，这就会让这些平台的计算需求比以往任何时候都大得多。

And so if you think about agents autonomously writing code, searching, carrying out these tasks, it just gives these platforms a much larger compute requirement than ever thought before.

对，绝对是这样。

Yeah, absolutely.

完全同意。

Totally agree.

这个例子也正好说明了资本支出方面正在发生的情况。

This one is another one that just points to what is happening in the CapEx side of things.

对五大超大规模云服务商的资本支出预测，在很短的时间内接连大幅上调。

Successive forecasts for the five largest hyperscalers' CapEx have moved sharply higher in pretty short succession.

所以，原本看起来像是几年或几个季度之后才会达到的支出上限，现在已经变成了近期目标，因为计算需求还在不断增长。

So spending that once looked like a ceiling, a number of quarters or years out has become near term as the demand for the compute keeps expanding.

对，你看这张图表，我觉得在任何时候，人的本能反应就是说，投资从这里开始就会趋于平稳。

Yeah, look, if you look at the chart here, I think at any point in time, I think natural instinct is just to say, like, the investment will kind of flatline from here.

但过去四年，作为一个经济体，我们一直低估了这个趋势的强劲程度。

Yet for the last four years, as an economy, we keep underestimating just the strength of the trend.

正如莎拉所说，随着模型的发展，我们现有产能的使用率一直在被提升。

And as Sarah said, with model developments, usage of our installed capacity keeps being taken up.

所以现在，大卫指出的那些数字就是当前的估计。

So right now, the numbers that David pointed to are the current estimates.

对，我是说，也许用DG的话来说，我们都认为对计算的需求现在是个“模型杀手”。

Yeah, I mean, maybe to use DG's language, I think we'd all call the demand for compute a model buster at this point.

我觉得有个我们都经历过的例子很能说明问题：山姆·奥特曼和莎拉·弗里尔大约一年前因为大规模的计算投入而受到很多批评。

I think one anecdote that's telling that we've all experienced is Sam Altman, Sarah Friar, they got a lot of flak a year or so ago for their massive compute commitments.

他们当时被认为鲁莽和激进，但现在我想每个人都会说，那个决定非常有远见。

They were being reckless and aggressive, and I think at this point, like everyone would say, they're incredibly prescient with that decision.

即便如此，两周前我们都看到他们不得不暂停专业版的新订阅，那可是一个2000美元的服务。

And even so, two weeks ago, we all saw that they had to pause new subscriptions on their pro plans, like showing up with a $2,000 service.

不，谢谢。

No, thanks.

我们现在面对的需求真是惊人，简直无法满足。

Pretty amazing insatiable demand that we're experiencing here.

对，确实如此。

Yeah, absolutely.

是的，基本上我们在供应链的每个环节接触到的每个人都在告诉我们同样的事情，或者说类似的情况，就是需求超过了供应。

Yeah, pretty much everyone we talk to at every stage of the supply chain is telling us the same thing, some version of the same thing, that demand outstrips supply.

数据中心供应链中有一些环节，你要等到2028年才能拿到材料或产品，所以这种情况并没有缓解。

There are certain elements in the data center supply chain where you can't get access to materials or products until 2028, and so this has not softened.

所以我想说，与此同时，你可以看看那些超大规模云服务商，需求端有一些高质量业务的证据，你可以放心依赖，对吧？

So I would say at the same time, you can look to the hyperscalers and see some evidence of high-quality business on the demand side that you can hang your hat on, right?

微软、谷歌和亚马逊的云积压订单加起来大约有1.7万亿美元。

So Microsoft, Google, and Amazon have about 1.7 trillion of combined cloud backlog together.

这些客户承诺正在快速增长，而平台也在大力投资于服务这些需求的能力。

And those customer commitments are building rapidly while the platforms invest heavily in the capacity to serve them.

所以，虽然目前在这个建设阶段自由现金流受到压制，但共识预测显示，从2028年开始会复苏，之后会有大幅增长。

So, while right now free cash flow is depressed during this build-out, consensus forecasts are showing a recovery from 2028 and substantial growth thereafter.

你可以看看亚马逊最近的财报电话会议，他们在那次会议上把这种J曲线动态解释得非常好，就是说GPU或TPU的实际使用寿命其实相当长。

You could look to Amazon's latest earnings call, where they did a really good job of explaining this sort of J-curve dynamic where the useful life of GPUs or TPUs is actually pretty long.

所以，你得先建好外壳，也就是数据中心。

And so, you have to build out the shell, the data center.

这需要一定的时间。

That's a certain amount of time.

你得购买芯片。

You have to buy the chips.

但这些芯片在很长一段时间内都会有非常高的经济价值。

But those will have a very useful economic life for a long period of time.

而且这段时间比我们任何人预想的都要长。

And it's been longer than I think any of us expected.

对，而且我知道一些超大规模云服务商确实因为为资本支出举债而受到批评。

Yeah, and I know some of the hyperscalers have frankly gotten dinged for raising debt for CapEx.

但和模型实验室的情况类似，我觉得那些犹豫不决、不够激进的公司现在都后悔了。

But similar to the model lab dynamic, I think the ones who have blinked and been less aggressive have regretted it.

我知道你最近和加文做的播客里提到了微软，但我觉得这个问题是普遍存在的。

I know on the podcast recently that you did with Gavin, Microsoft came up, but I think this is an issue across the board.

对，另一个经常被讨论的动态是现有GPU的现货市场价格，这又是一个信号，表明任何能上线的GPU都能以有吸引力的价格定价，而超大规模云服务商也能获得可观的回报。

Yeah, the other dynamic that's been spoken about a lot is pricing on the spot markets for existing GPUs, which is just another signal that any GPU that you can bring online is being priced at an attractive rate where the hyperscalers also are earning an attractive return.

对，这是个关键点。

Yeah, it's a key point.

这是我们经常讨论的事情之一，每一波技术浪潮都会创造巨大的用户或消费者剩余。所以目前实际发生的情况，至少据我们所知，是消费者和用户从中获得了巨大的价值。

This is one of the things that we talk about all the time is each one of these successive waves just creates a tremendous amount of user or consumer surplus And so what's actually happening right now, at least as far as we can tell, is consumers and users get a tremendous amount of value out of this.

这就是为什么他们用得这么多。

That's why they're using it so much.

服务这些用户的供应商赚了很多钱。

The vendors who are serving those users are making very good money.

然后，你一路往下看整个技术栈，直到芯片层面，你会发现现在获取芯片的成本比12、18、24个月前高得多，但你仍然可以保持很高的利润率，并从用户那里创造巨大的剩余价值。

And then, you go all the way down every level of the stack to the chips, where you have to pay much higher than you did 12, 18, 24 months ago to get access to them, and yet you could still make very high margins and create a tremendous amount of surplus out of the users.

当我们讨论超大规模云服务商的资本支出时，我们应该把它看作是别人的订单簿。

And as we talk about this CapEx for the hyperscalers, we should think of it as someone else's order book.

所以这些历史上利润最高的大型平台，正把现金大量投入AI基础设施。

So these big platforms with historically the largest profits are pouring their cash back into AI infrastructure.

正如我们刚才谈到的，这短期内给它们的自由现金流带来了压力。

That's putting pressure on their free cash flow short term, as we just talked about.

但对芯片订单、电力和建筑行业来说，这却是个福音，这也是为什么这轮更广泛的技术繁荣变成了工业繁荣。

But it's been a boon for chip orders, for power, for construction, and that's why this broader technology boom has become an industrial boom.

我们花了更多时间研究服务于这轮工业繁荣的企业，以及数据中心供应链上的各个环节。

And we've been spending more of our time looking at businesses serving this industrial boom and everywhere along the data center supply chain.

我觉得这很有意思，因为传统上，我们的领域更侧重于将现有知识产权变现，一次开发，无限销售。

I think it's interesting because traditionally, our world was more about monetizing existing IP, build once and then sell infinitely.

但这些企业面临更多复杂性，需要管理融资、供应商关系、产能预测和需求预测。

But with these businesses, there's many more complexities that the companies need to manage, like financing, managing vendor relationships, predicting capacity, and predicting demand.

我们看到有些团队在这方面很出色，有些则遇到困难或暂时停滞。

And we've seen there are some teams that excel at that and some teams that have struggled or taken a pause.

这和我们通常专注的领域有所不同，需要不同的专业能力。

And it's a little bit of a different expertise than we typically spend our time with.

这让我们进入了一个原子时代。

This puts us in a new age of atoms.

这页上的数字非常大。

Big numbers on this page.

到2040年，全球基础设施投资需求估计达到90万亿美元。

Global infrastructure investment needs are estimated at $90 trillion through 2040.

但重要的是，这远远超出了AI和数据中心的范围。

Importantly, though, this goes way beyond AI and data centers.

它涵盖了电力、水资源、道路和交通。

It includes power, water, roads, transit.

我们在整个投资组合中都能看到这一点。

We're seeing this across our portfolio.

比如在Anduril，你可以看到他们庞大的制造设施。

You can see this at Anduril with their massive manufacturing facility.

它的规模相当于87个足球场。

The scale of it is 87 football fields.

Waymo 正在大力扩展他们的站点。

Waymo, they're aggressively expanding their depots.

当然，SpaceX 在路易斯安那州投资了 1000 亿美元。

And of course, SpaceX with their $100 billion investment in Louisiana.

正如 Santi 所说，从我们的角度来看，今天押注未来，和过去的情况已经不完全一样了。

Betting on the future today, from our perspective, isn't what it exactly was in the past, as Santi said.

这意味着要建工厂、扩大基础设施、增加更多技术岗位，而且是在全国范围内。

It means building factories, expanding infrastructure, more skilled jobs, and across the country at large.

和我们刚才谈到的数据中心类似，我认为投资实体基础设施和投资软件是非常不同的。

And similar to what we just said on data centers, I think investing in physical infrastructure is very different than investing in software.

Elon 多年前就提出了“工厂就是产品”这个说法，难怪我们支持的很多优秀创始人都是从特斯拉或 SpaceX 出来的，因为他们深知，在执行工厂建设的过程中，随着规模扩大，这会成为巨大的竞争优势。

And Elon coined the term, "the factory is the product" many years ago, and it's no surprise that a lot of the great founders that we've backed have come out of Tesla or SpaceX because they just learned that executing on the factory ends up being a massive competitive advantage as they scale.

对。

Yep.

关于数据中心，我们经常听到很多误解。

So there are many myths that we hear all the time about data centers.

它们把美国的水都耗光了，有钱人不想住在它们附近，还有我们最喜欢的一个说法——我的电费要涨上天了。

They're draining all of America's water, rich people don't want to live near them, and then one of our favorites, my electricity bill is going to go through the roof.

这听起来可能有点反直觉，但数据中心其实能帮你降低电费。

It may seem counterintuitive, but a data center can help lower your electricity rate.

最近一项美国的研究显示，数据中心容量每增加10%，居民电价就会下降0.4个百分点。

A recent US study showed that for every 10% increase in data center capacity, residential rates went down by 40 bps.

你可以这样理解：电网是一个共享的固定成本基础，包括电线杆、电线和变电站，所以像数据中心这样的大型稳定客户，能帮助把这些成本分摊到更多的电量上。

You should think about it as a power grid is a shared fixed cost base, poles, wires, substations, and so a large stable customer, like a data center, can help spread those costs across more units of electricity.

简单来说，在共享系统上增加需求是件好事。

Simply put, more demand across a shared system is a positive.

我完全同意你的说法。

I totally agree with what you're saying.

对共享基础设施的投资，也能改善与之相连的家庭和企业的经济状况。

Investment in shared infrastructure can also improve the economics for the households and businesses connected to it.

对，而且我觉得这一点没有得到媒体足够的关注。

Yeah, and I don't think that point gets enough media attention.

迪娜·鲍威尔最近为Meta参加了一个播客，谈到了他们在路易斯安那州的站点，他们与社区合作降低了电费。

Dina Powell went on a podcast recently for Meta and talked about the Louisiana site that they did, where they worked with the community to lower electricity costs.

所以现在确实有这样的实际例子。

And so there's real examples of that today.

这不仅仅是空谈。

It's not just lip service.

完全同意。

Totally agreed.

那么现在我们来谈谈我们在模型和应用层看到的趋势。

So now we're going to talk about the trends that we're seeing at the model and application layer.

一方面，正如我们之前提到的，AI正在带来巨大的收入和节省。

On the one hand, as we've previewed, AI is generating major revenue and savings.

另一方面，采用率仍然非常早期。

On the other hand, adoption is still extremely early.

而且，我认为除此之外，趋势非常明显的是成本在急剧下降，我们之前已经提到了智能体，但现在智能体实际上已经经济实惠，适用于更广泛的工作范围。

And then, I think on top of this, the trend is very much so that costs are plummeting, and we previewed agents already, but agents are actually economical now for a much wider range of work.

当然，这些变化还在不断发生，对吧？

And of course, this is changing even as we speak, right?

像TypeSafe这样的新实验室推出的Jev等新创新，正在把这一点推向更极端的程度。

New innovations like Jev, by newer labs like TypeSafe, are taking this to a greater extreme.

你会开始看到，成本相差两个数量级的情况，确实会影响到应用场景的数量。

You're starting to see things like two orders of magnitude cost differences really impact the number of use cases.

这就是经典的杰文斯悖论，所以这个名字起得非常贴切。

It's classic Jevons paradox, which is why it's very aptly named.

总的来说，我认为这对模型层和应用层来说都是一个绝佳的机遇。

And I'd say, all in all, this is an awesome setup for both the model layer and the app layer.

这些改进正在为软件公司创造实实在在的增长和盈利机会。

These improvements are creating real opportunities for both growth and profitability among software companies.

这并不是说每个人都会成功，但人工智能吸引风险投资进入越来越多的行业，这并不令人意外。

And that's not to say that everyone will win, but it's no surprise that AI is attracting venture dollars into a very much expanding range of industries.

我们还看到越来越多的公司在私募市场中达到了巨大的规模。

And we're also seeing more companies just reach enormous scale in the private markets.

这张图不是新图，但它是我的最爱之一，展示的是OpenAI和Anthropic的合计营收规模。

So this is not a new chart, but it is one of my favorite charts, which is the combined scale of revenue for OpenAI and Anthropic.

这张图我们每次GP外出会议都会用，我记得过去几年都是这样，而且每个月都得更新，因为数据变化太快了。

And this is one that we've had in every GP offsite, as far as I can remember the last few years, and we have to keep updating it every month because it just gets out of date that quickly.

我觉得重点在于，你可以从数字里看到，OpenAI和Anthropic的合计年化营收已经攀升到了一个惊人的水平。

I think the point here, you can see it in the numbers, OpenAI, Anthropic, their combined annualized revenue has climbed to an extraordinary degree.

我们喜欢把它和历史上最伟大的软件公司做对比，你可以看到右边，如果你看看那些史上最优秀软件公司的营收增长估算。

We love to show this relative to the greatest software companies in history, and you can see that in the right that, if you look at the estimates for revenue added ... for the best software companies ever built.

而领先AI实验室的净新增营收估算已经远远超过了它们。

The estimates for net new for the leading labs has well surpassed that.

所以这真的很惊人，不仅规模大，而且达到这个规模的速度也快得惊人。

So it's really a striking combination of not only scale, but also speed to get there.

但我们现在还处于早期阶段。

But we're still pretty early.

我觉得相对于其他平台变革，AI变革的有趣之处在于，我们不仅在用户比例或使用AI的用户比例上还早，而且在那些用户的钱包份额上也非常早。

I think relative to other platform shifts, the interesting thing about the AI shift is we're not only early in terms of percentage of population or percentage of users that use AI, but still very early on the share of wallet gains within those users.

所以这和2008年有点不同，那时候我们会用拥有iPhone的人数比例来做模型。

So it's a little bit different than in 2008, we would model percentage of people with an iPhone.

现在，我们需要考虑的是会使用AI产品的人数比例，以及他们在多大程度上使用这些AI产品。

Here, we need to think about percentage of people that will use AI products and then to what extent they're using the AI products.

对，我觉得我们仍然处于早期阶段这一点非常重要，而且这里有很多指标都能说明这一点。

Yeah, I think that point that we're still so early is such an important one, and there's a bunch of metrics in here that I think elucidate that.

其中一个就是，如果你看看标普500公司的实际部署情况，这个比例已经达到了69%。

One is, if you think about live deployments at S&P 500 companies, that's at 69%.

所以，如果你像我们一样对AI充满热情，这个数字可能更接近我们的预期。

So, if you're AI-pilled, as we all are, maybe something closer to what we would all expect.

但如果你转向可量化的影响，这可能是衡量他们部署进展的一个好指标，这个比例是30%。

But then if you move toward quantifiable impact, which is probably a good metric of how far along they are in their deployment, that's 30%.

再来看最终的衡量标准，也就是随时间追踪的指标，实际上只有2%。

Now if you go to the ultimate barometer, which is a metric tracked over time, that's actually only at 2%.

所以我们相信AI正在带来成果，但在组织内部深化其使用并长期追踪这些成果，还有巨大的空间。

So we believe that AI is delivering results, but there is a huge amount of room to actually deepen its use inside the organization and track those results over time.

所以，坦白说，从你看到的这些单独部署，转向真正具有深远影响的常态化工作流程，是下一个阶段。

And so moving from these individual deployments that you're seeing to really deeply influential recurring workflows is the next stage, frankly.

而且，我想说的是，我们接触到的大多数企业，他们目前对AI的使用主要还是集中在微软的Copilot上，这正好说明他们还有很长的路要走。

And, I would just say that most of the enterprises that we speak with anecdotally, their exposure to AI is still mostly with Microsoft Copilot, which just shows you how far they have to go.

模型能做到什么和实际使用之间存在的差距，正是让我对应用层充满期待的原因。

This gap between what the models can do and how they're being used is what makes me so excited about the application layer.

Sarah，你和Databricks的Ali Ghodsi以及Martin有过一次很棒的对话，他提到AI虽然对世界了解很多，但对你公司的情况却知之甚少。

Sarah, you did this great conversation with Ali Ghodsi at Databricks with Martin, and he talked about how the AI can know a lot about the world but know very little about your company.

我们在应用层公司也看到了这一点，它们正在填补这一差距。

And we're seeing this in our application layer companies, where they're bridging this gap.

过去几个月里有个例子让我印象特别深：就拿Revolut来说，他们拥有非常专业的工程团队，但依然选择与Eleven Labs合作，采用Eleven业界领先的语音模型，将其整合到客户账户和银行工作流的安全连接中。这样当客户打电话时，问题就能得到顺畅、高效且安全的解决。

And one example that has stuck out to me over the last few months is if you were to take Revolut, who has a really sophisticated engineering team, they still had to partner with Eleven Labs, and take Eleven's best-in-class voice model and harness it to connect securely with customer accounts and banking workflows so that when I as a customer call, I can have my problem resolved smoothly, efficiently, and securely.

当下的机会主要在于利用这些能力，构建可靠的服务。

Today's opportunity is so much around taking these capabilities and harnessing them to build reliable services.

除了企业采用AI仍处于早期阶段这一事实外，另一个值得我们用数据来支撑的显著趋势是，重度用户的使用程度已经远远拉开了差距。

So in addition to the fact that enterprise adoption is still quite early for AI, the other really remarkable trend that we wanted to put some stats to is just that the power users are totally pulling away.

所以你可以看到，AI支出在各大公司普遍上升，但正如我们所说，重度用户，也就是使用最密集的用户，花费要多得多。

So you can think of AI spending rising generally across companies, but the power users, as we talked about, the most intensive users, are spending much, much more.

我们看了Yipit的数据，如果你看前1%的AI供应商中位数支出，大约是前10%的八倍。

So we looked at some Yipit data for this, and if you look at median AI vendor spending in the top 1%, that's roughly eight times the level of the top 10%.

令人印象深刻的是，前1%的支出几乎等于2%到10%的总和。

What's impressive about that is it's almost as much as the 2 to 10% combined is basically the 1%.

所以这些重度用户显然是早期采用者。

So you got these power users that are clearly early adopters.

对，确实如此。

Yeah, absolutely.

而且，即使在我们自己的投资组合公司里，这些公司可以说几乎完全是AI原生或AI重度依赖的，你也看到顶级用户每月花费大约7500到9000美元，而中位数用户，即使在这些最AI原生的公司里，可能更接近每月订阅费用200美元。

And I mean, anecdotally, even inside our own portfolio companies, which you could say are pretty much fully AI native or AI-pilled, you're seeing the top users spend anywhere from, call it, 7,500 to 9,000 a month, and the median users, again, for the most AI native companies, probably closer to the cost of a monthly subscription of 200.

也许你叠加两个订阅，所以是200到400美元。

Maybe you're stacking two subscriptions on top, so 200 to 400.

所以，如果比较中位数用户和顶级用户，支出差距超过20倍。

So over 20 times the spend if you think about median versus top users.

我们会和我们的投资组合公司交流，其中一个讨论的问题是他们的采用率有多高，以及如何衡量这个采用率。

We talk to our own portfolio companies and one of the questions that we discuss is how much adoption do they have and how do you measure that?

举个例子，如果你只看他们在AI工具上的支出占员工人数的比例，我认为相对领先的财富500强公司目前可能只有1%。

And, if you just look at the percentage of money that they're spending on AI tools compared to headcount, as an example, I think relatively forward-leaning, large Fortune 500-type companies are probably today at 1%.

而我们投资组合中最前沿、最拥抱AI的公司，这个比例可以高达10%。

And the most forward-leaning AI-pilled companies in our portfolio can be as high as 10%.

所以这些都是关于我们处于扩散的早期阶段有多早，以及这种扩散会深入到什么程度的问题。

And so all of these are just questions of how early are we into diffusion and how deep will that diffusion go?

在我们的投资组合中，另一个例子是，许多企业在第一次购买Cursor或Cognition时非常兴奋。

Another example we've seen within our portfolio, many enterprises are really excited when they buy Cursor or Cognition for the first time.

但现实是，这只是旅程的开始。

But the reality is that's just the beginning of the journey.

从仅仅采购和试用这些工具到完全采用，中间的路程是巨大的。

The runway from just procuring and trying these tools to full adoption is massive.

而且，莎拉，你早些时候提到过这一点，但我们开始看到可量化的案例研究，我们在这里也重点介绍了几家公开报告这些指标的公司，它们通过AI看到了显著的改进。

And I mean, Sarah, you touched on this early, but we're starting to see quantifiable case studies and we've highlighted a couple here of public companies that actually report on the metrics where they've seen meaningful improvements with AI.

我们特别提到两个例子，一个是在成本方面。比如Chime报告说，过去四年里，他们的服务成本每年都降低了超过10%。

Two that we call out, one on the cost side, for example, Chime has reported that they've reduced their cost to serve by over 10% a year for the last four years.

这样算下来，服务成本几乎降了一半，这背后有我们投资的公司Decagon在支持。

Compounded, we're talking about almost a 50% reduction in cost to serve, happily supported by one of our portfolio companies, Decagon.

但总的来说，很多降低服务成本的举措都在进行中。

But just in general, many initiatives to lower the cost to serve.

在收入方面，我觉得Shopify的例子挺有意思。他们推出了AI助手，这个AI助手能帮助商家更快地上手，速度快很多。

On the revenue side, an interesting one that I found was Shopify here, who they launched their AI sidekick, and this AI sidekick helps merchants get up to speed much, much faster.

所以，入驻后15天内达到5个订单的客户比例增长了8%。Shopify追踪的一个指标是，一旦你达到5个订单，你就更可能留在Shopify上，继续使用。

So, the percentage of customers that reach five orders within 15 days after onboarding has grown 8%, and that is kind of a metric that Shopify tracks is once you reach five orders, you're going to stick around and you're going to retain on Shopify.

这对他们的业务来说也是一个很大的推动力。

So it's been a meaningful tailwind to their business as well.

对，完全同意。

Yeah, absolutely.

而且我觉得你刚才提到的服务成本下降，确实是个关键点，对吧？

And I think to your point on just the cost to serve coming down, right?

你想一下，降低服务成本带来的好处，就是你有了更多的竞争空间。

If you think about the advantages that lower cost to serve gives you, you just have more room to compete.

所以你可以提供更好的价格，或者更广泛的服务。

So you have better pricing, maybe broader service.

你可以把资金重新投入到增长中。

You can reinvest in growth.

所以我觉得这些好处不仅惠及客户，也改善了利润率，然后他们再把利润重新投入进去。

And so I think the benefits are reaching the customers, not only the customers but also the margins, and then they rotate that back in.

另外我觉得非常有趣的一点是，当然，各家公司情况不同，不能一概而论，但老牌企业在变现方面也做得相当不错，ServiceNow就是一个很好的例子，对吧？

And then I think the other thing that's been really fascinating to see is, of course, they vary, so you can't throw everything into one bucket, but the incumbents have done a pretty nice job on monetizing this as well, and ServiceNow is a good example, right?

他们报告了超过十亿美元的AI年度合同价值，而且代理式部署增长了九倍。

They've reported more than a billion in AI ACV, and actually a 9X increase in agentic deployments.

所以你可以看到，从老牌企业到新兴公司，整个行业都在发生这样的变化。

So you're kind of seeing it across the stack, incumbents to newer companies.

我们经常讨论的一个有趣话题，而且像Stripe这样一些最前沿的公司也跟我们聊过，就是他们到底把增量AI投资花在了哪里。

One of the interesting things that we've talked about a lot and, again, some of our most forward-leaning companies like Stripe have discussed with us is, where are they actually putting their incremental AI investment dollars?

他们是把钱花在为客户开发新产品、从而可能带来更高收入上，还是花在成本端的优化或效率提升上？

Are they putting it toward things like building new products for customers that could drive higher revenue, or are they putting it toward optimization or efficiency gains on the cost side?

这就像是我们的一块试金石，看看这些公司的创始人或CEO们认为最大的机会在哪里。

And this is sort of a litmus test for us of like, where are the founders or CEOs of these companies seeing the most amount of opportunity?

比如，推动收入增长的机会是上不封顶的。

Like the opportunity to drive revenue growth is unbounded to the upside.

哦，对。

Oh, yeah.

而成本改善的机会呢，虽然你可以把省下来的钱再投回去，但那更像是一个潜在的、一直存在的机会。

Whereas the opportunity for cost improvement, yes, you could take that and reinvest it, but that's sort of a latent opportunity that will continue to exist.

如果你觉得今天最该把钱花在优化成本结构上，那这又说明了什么呢？是不是意味着你的收入机会有限？

And if you think that the best and highest use of your dollars today is to optimize your cost structure, what does that say about the revenue opportunity for you?

而且就算在成本优化内部，也有不同的类型，对吧？

And even within cost optimizations, there's different flavors, right?

比如一种是降低服务成本，这样你就有余力再投资，让业务变得更好。

Like one is lowering your cost to serve, which allows you to reinvest and makes you a better business.

我觉得六个月前，我们这个行业真的在专注于内部重建核心系统，如果你的工程师们专注于重建核心系统只是为了省几十万美元，我觉得这些资源应该有更好的用途。

I think six months ago, our industry was really focused on rebuilding systems of record internally, and if your engineers are focused on rebuilding a system of record to save a couple hundred thousand dollars, I expect there to be better uses of those resources.

对，完全同意。

Yeah, totally agree.

Sarah提到了ServiceNow，看到了大量的智能体使用。

Sarah referenced with ServiceNow, seeing massive agentic usage.

智能体已经来了。

Agents are here.

它们正在执行任务。

They are performing tasks.

任务需要多个步骤，而每个步骤都需要多次模型调用。

Tasks require multiple steps, which require multiple model calls.

这有助于解释OpenRouter上智能体令牌使用量增长14倍的原因。

That helps explain the 14X growth, in agent token usage on OpenRouter.

不过，这些步骤的成本正在变得越来越低。

Those steps, though, are becoming more affordable.

我们现在看到缓存技术的应用，系统可以复用背景信息，而不是每次都从头处理。

We're seeing caching, so the system can reuse background information instead of processing it from scratch each time.

我见过Heavyai利用这一点。

I've seen Heavyai take advantage of this.

他们的金融聊天工作负载的运行成本降低了10倍。

They've seen their financial chat workloads become 10X cheaper to run.

他们之前同样的预算现在可以支持更多的工作。

The same budget that they had before can now support more work.

对，我觉得你更广泛的观点是，成本降低使得现在让代理尝试、检查结果、再试一次变得切实可行，你可以开放大量任务，而这些任务以前可能因为推理或工具使用成本过高而无法进行。

Yeah, I think your broader point is just that lower costs make it practical now for an agent to try, check its work, try again, and you can just open up a ton of tasks where maybe reasoning or tool use would've been just prohibitively expensive.

我认为这在可靠性成为决定是否使用代理的关键因素的场景中尤为重要。

And I think this is especially important in cases where reliability is maybe the de facto reason you would or would not use an agent.

而且，我们之前也提到过类型安全，但想象一下，当成本降低一个数量级、两个数量级时会发生什么。

And again, we sort of talked about type safe previously, but imagine what happens when you go an order of magnitude cheaper, two orders of magnitude cheaper.

我认为由此开启的用例数量实际上是难以想象的。

I think the amount of use cases that open up are actually unimaginable.

完全同意。

Totally.

而且延迟方面也有大幅改善。

And with much greater improvement in latency.

没错。

Absolutely.

所以现在公司确实越来越多地在考虑如何在大规模场景下优化延迟或性能，同时兼顾成本。

So companies are definitely thinking more and more today about how to optimize latency or performance at large with cost.

就拿 Databricks 举例吧。

So take Databricks as an example.

他们利用路由机制，为每个具体任务选择合适的模型，以提升性能和降低成本。

They're leveraging routing, choosing the appropriate model for each specific task to improve performance and cost.

他们的智能路由器表现更好。

And so their smart router performed better.

它解决了更多问题，成本比最强的单一模型还低 35%。

It solves more problems at 35% lower cost than the strongest individual model.

我们现在看到很多另一个方法是微调。

Another approach we're seeing a lot of right now is fine-tuning.

我们在Harvey那里看到了这一点。

We've seen this with Harvey.

我们在Elise AI那里也看到了这一点。

We've seen this with Elise AI.

以Elise为例，他们微调了一个较小的模型，所以成本大大降低，便宜了60%，而且延迟也低得多，所以从音频角度的实时用例变得可行了。

In the case of Elise, they fine-tuned a smaller model, so it became much more affordable, 60% cheaper, but it also had way lower latency, so live use cases from an audio perspective became tenable.

因此，这些工程上的进步让AI更有用，也更实惠。

And so, these engineering gains are making AI more useful but also more affordable.

对，我觉得如果你把这些汇总起来，从应用业务的角度看，他们真正关心的单位是完成客户任务的成本。

Yeah, and I think if you sort of roll that up, and think about for application business, the relevant unit for them is really the cost of getting the customer's job done.

如果你能把这个成本分离出来，对吧，你完成了客户的任务，你为此收费，而且你可以通过更好的路由让产品更经济，那真的很神奇，因为现在你不是每一步都用最贵的模型，但你仍然完成了任务，我认为最好的应用公司会看到更多的利润提升。

And if you can decouple that, right, you get the customer's job done, you charge for that, and you can use better routing to make the product more economical, that's pretty magical because now you're not using the most expensive model for each step, but you are getting the job done, and I think we're going to see more margin improvements in the best app companies.

这是个很好的过渡。

It's a nice segue.

这是个很好的过渡。

It's a nice segue.

我们一直在讨论企业内的使用情况，但消费者方面也很重要，毕竟那里是我们取得一些最大成果的地方，也是我们投入大量时间的地方。

We've been talking mostly about usage inside of the enterprise, but important to talk about consumer, given obviously that's the place where we've had some of the largest outcomes and where we spend a lot of our times.

不，我们认为现在还为时过早。

No, we think it's still pretty early.

最近一项调查显示，只有略超过2%的美国家庭真正订阅了AI付费服务。

There was a recent survey that showed that only just over 2% of US households actually have a paying subscription for AI.

也许订阅制不是向大多数家庭收费的最佳方式，但到目前为止它确实是最有效的。

Maybe subscription is not going to be the best place to monetize the majority of households, but it has been so far.

订阅收入的有趣之处在于，它的用户留存率是我们见过的消费者领域中最高的。

And the interesting thing about the subscription revenue is it has some of the best retention we've ever seen in consumer.

亚历克斯喜欢谈论微笑曲线，而像这样不仅保持平稳，反而随着产品改进和用户回归而上升的留存曲线，真的非常罕见。

Alex likes to talk about smile curves, and it's really, really rare to see a retention profile that not only flat lines but actually smiles as the product improves and people come back to it.

但这就是我们在AI领域看到的情况，这恰恰证明了订阅制能为付费用户带来的价值。

But that's what we've seen with AI, which just demonstrates the value that the subscription is able to deliver to the person who pays.

顺便说一下，关于订阅，这些数字让我印象最深的是，特别是如果你看右下角的图表，它们真的非常小。

By the way, just on subscription, what struck me about these numbers, especially if you look at the bottom right chart, is just how small they are.

想想亚马逊Prime。

Like think about Amazon Prime.

那有超过2亿家庭。

That's over 200 million households.

Netflix。

Netflix

Netflix是个很好的例子。

Netflix is a great one.

7000万家庭。

70 million households.

我不得不查一下。

I had to look it up.

我当时想，"这个K对吗？"

I'm like, "Is that K correct?"

所以完全同意，这才刚刚开始。

And so totally agree that it's just getting started.

我们的合伙人乔什·埃尔曼有一个定义，我挺喜欢的，但可能跟这个有点相反，就是说消费级AI是我日常生活中用的，不一定是我报销的那种。

Our partner, Josh Elman, does have a definition I like that would be counter to that a little bit, which is consumer AI is what I use for my daily life and not necessarily what I expense.

所以在他看来，97%使用AI的家庭还没付费，这也许并不太让人意外。

So in his view, he would say, it's maybe not super surprising that 97% of households that are using AI aren't yet paying for it.

你说得对，桑蒂。

You are right, Santi.

我确实喜欢那种上扬的留存曲线。

I do like smiling retention curves.

我们确实喜欢看到用户越来越频繁地回来。

We do like to see users coming back more and more.

大型消费平台的另一个共同特征是用户使用时长。

Another common characteristic of large consumer platforms is time spent.

比如你看Facebook、Instagram、TikTok、Snap，它们的活跃用户每天都会花30到60分钟在这些平台上。

So if you were to look at Facebook, Instagram, TikTok, Snap, they all get 30 to 60 minutes per day from their active users.

目前市面上涌现出许多优秀的AI助手，它们都希望能成为我的首选应用。

And while the best AI assistants, and there's many up and coming right now, all aspire to be my go-to application.

我希望它们能保持极高的持续性，随时待命，并且非常主动，这样一来，未来我花在它们身上的时间可能反而不会是最多的。

I hope that they are going to be incredibly persistent and always on and super proactive, such that they may not be the place where I spend the most time in the future.

当AI代理在后台运行，我们不再能通过外部数据来观察用户参与度，只能看社交应用的屏幕使用时间时，我们的工作难度会增加。

It's going to make our job harder when we can't look at engagement via outside-in data because the agents are working on the background instead of us just looking at screen time on a social media application.

但这确实很重要。

But it is important.

我记得当年最新模型刚出来的时候，哪怕是在我们的业务场景里，我们也会启动一个深度研究任务或者智能体任务，然后开车进城之类的。这可不是那种一边开车一边打电话的特斯拉模式。

I remember when the newest models come out in December, even in our workloads, we would set off a deep research task or an agentic task and then drive up to the city or something, and that's not drive and dial Tesla.

这也不是你以前那种被盯着看的屏幕使用时间，那是上一代消费者产品的玩法了。

That's not screen time that you're being tracked as would've been in prior generations of consumer.

完全同意。

Totally agreed.

虽然现在还早，但消费者领域确实迎来了一个全新且令人兴奋的阶段。

It's early, but it is a really new and exciting time in consumer.

Muse和Instinct无疑是新来的，但它们发展得很快。

Muse and Instinct are no doubt the new kids on the block, but they've grown really quickly.

它们和ChatGPT一起，越来越多地取代了我对传统搜索的查询。

Those, along with ChatGPT, are taking more and more of my queries away from traditional search.

但这是一个非常动态的时期，所以如果你是一个现有的消费者发现平台或现有的市场，你真的需要现在就开始考虑你的棋局。

But it's a really dynamic time, and so if you are an existing consumer discovery platform, an existing marketplace, you really need to be thinking about your chess moves right now.

我会问几个问题。

There's a couple questions I'd be asking.

首先，我能从AI代理那里获得多少增量需求？

First, how much incremental demand can I get from an AI agent?

它们能给我带来多少额外的订单？

How many more orders can they bring me?

第二，我的业务中有多少，我的利润池中有多少来自于拥有客户关系和发现？

And two, how much of my business, how much of my profit pool comes from owning the customer relationship and discovery?

上周有很多新闻，亚马逊对Muse说了“不，谢谢”。

Last week there was a lot of news where Amazon said, "No, thank you" to Muse.

但Instacart却表示，非常乐意。

But Instacart said, "Yes, please."

如果你想想亚马逊（不包括AWS）和Instacart，看看它们的广告收入，会发现广告收入超过了它们的全部营业利润。

And if you think about Amazon, ex-AWS, and Instacart, and look at their advertising revenue, it outpaces all of their operating profit.

所以，广告收入和掌握客户关系极其重要。

And so advertising revenue, owning the customer relationship is incredibly important.

但对亚马逊来说，接入Muse能带来多少新增订单或客户呢？

But for Amazon, how many incremental new orders or certainly customers am I going to get from connecting to Muse?

其实没多少。

Not that many.

而Instacart就不一样了，线上杂货渗透率还比较低，还有很多订单可以去争取。

Whereas if you look at Instacart, online penetration of groceries, still relatively early, a lot more orders to go get.

所以乐观来看，这些应用可能会带来大量新增订单。

And so the optimistic possibility here is that there's going to be a lot more orders coming from these applications.

以前，我得通过很多不同的步骤才能完成一个订单。

Historically, I've had to click all these different ways through to process an order.