Cloudflare 首席执行官预测：AI 智能体数量将超过人类 1000 倍

你最近怎么样？

How are you doing?

各位，很高兴见到你们。

Gentlemen, good to see you.

我挺好的。

I'm doing well.

见到你真好。

Great to see you.

见到你。

To see you.

你那边有什么新动静？

What's new in your world?

你收购了一些公司。

You made some acquisitions.

你收购了一家公司。

You made an acquisition.

给我们详细讲讲吧。

Take us through it.

是的。

Yeah.

我们收购了一家叫 Void Zero 的公司，他们开发的 Veat 是目前最受欢迎的开发者平台之一。

So we we bought a a company called Void Zero, which, makes Veat, which is one of the most popular developer platforms that's there.

他们的团队简直太棒了。

It's just an incredible team.

创始人 Evan Yu 是个非常出色的人，我们的团队跟他合作特别兴奋。

Evan Yu, who's the founder, is just a just a first class human being, someone who who, our team is super excited to work with.

他组建的团队也是一流的。

The team that he's assembled, is is just great.

我觉得，这个平台正越来越成为支撑互联网上众多智能代理的核心工具。

And I think that this is increasingly becoming the platform that, is being used to power a lot of the, agents that are running around around the Internet.

而且很多这类代理都在 Cloudflare 上运行，所以我们觉得这两者结合得特别自然。

And, and a lot of those agents are running on Cloudflare, and so we think it's just a really natural, combination.

这种协同效应有多简单？

How simple is the synergy?

你是要把每个月下载VEET的那1.3亿用户都引导过来吗？

Is it you'll funnel those 130,000,000 users who download VEET every month?

我觉得是每周1.3亿次下载。

I think it's 130,000,000 weekly downloads.

每周。

Weekly.

天哪。

Wow.

是的。

Yeah.

这数量真不小。

That's a lot.

然后让他们默认使用Cloudflare？

Into, you know, preferring Cloudflare, defaulting to Cloudflare?

还是说，在开发者集成和公司整合方面，背后还有更多协同效应？

Or is there more synergy under the hood around developer integration, company integration?

你是怎么看待这件事未来发展的？

Like, how are you thinking about this playing out?

是的。

Yeah.

我们会继续把它作为开源项目来维护和投入。

We continue we we plan to continue to leave it as an open source project Mhmm.

并且以这种方式提供支持和资源。

And support it and invest in it that way.

我们希望把它深度整合进 Cloudflare 的开发者平台，确保 Cloudflare 成为运行任何 Vite 项目的最佳选择。

We wanna integrate it closely with Cloudflare's developer platform and make sure that Cloudflare is the best place to run any sort of Vite project that that you that you have.

不过它也能在其他各种平台上正常运行。

But it'll work in any any of the different platforms as well.

我们真正想做的，是确保 Evan 和他的团队能得到充分支持，继续把这样一个了不起的平台做得更好，我们也相信这会吸引更多开发者使用 Cloudflare Workers 平台。

And so we just really wanted to make sure that Evan and his team have the support to make sure they could continue to to really invest in in what was in what what's just an incredible platform, and and we think that that is gonna drive more developers to Cloudflare Cloudflare Workers platform as well.

现在开发者最头疼的问题是什么？

What are the headaches for developers these days?

我知道大家都很担心令牌成本和使用预算。

I I know everyone's concerned about token costs and token budgets.

但这些往往不会直接体现在开发者身上。

Sometimes that doesn't show up for the developer.

更可能是老板在操心这个。

It's more like the CEO that's worried about it.

那系统的稳定性是不是更难维持了？

But is uptime more difficult to maintain?

我们最近看到不少在线状态监控页面，红色和黄色的警告越来越多。

We've been seeing screenshots of different uptime tracker status pages that have more and more red and yellow on them.

现在最前沿的问题是什么？你们又是怎么帮上忙的？

Like what's at the top of the stack and how are you helping?

是的。

Yes.

我觉得，面对即将到来的智能代理时代，我们需要一种全新的架构，而不是沿用上一代应用的模式。

I think that you need a different architecture than than we've than we've had to build sort of the last generation of applications for what's coming with with agents.

你可以想象一下，美国大约有一亿名知识工作者。

If you imagine, you know, there are about a 100,000,000 knowledge workers in The United States.

如果

If

所有

all

这每一亿名知识工作者都配备一个替他们工作的代理，而每个代理都运行在像 AWS 或 Google Cloud 那样的传统容器里，

of those knowledge workers had one agent which was which was working on their behalf, that would that would and and each of those was each of those agents was running in a container, a traditional container like something you get at a AWS or a a Google Cloud.

光是运行这些代理所需的 CPU 资源，就差不多占了当今全球所有芯片制造商总产能的百分之五十。

The amount of just CPU resources that would be needed to to run just those those agents, assuming they've just had one agent per person, is about 50% of the total CPU capacity that's generated, you know, by all the different CPU manufacturers that are out there today.

这还只是美国的知识工作者。

And that's just The United States knowledge workers.

如果扩展到全世界，那所需的算力将是当前全球所有 GPU 和 CPU 总容量的三四十倍。

If you take that to the world, then it's several times, you know, 30 or 40 times the capacity of of GPUs, and and CPUs that that are existing today.

所以我们认为，当这些代理在生成代码时，你需要一个全新的平台来支撑它们。

And so what we really think is that as these agents are creating code, you need a different platform for it.

Cloudflare Workers 并不是主要基于容器，而是采用了一种叫隔离环境的技术，效率高得多。

And and Cloudflare was built Cloudflare Workers was built not on on containers as much, but on something called isolates, which is a much more efficient technology.

我们看到，随着人们开发和使用这些代理，Cloudflare 成为了运行它们更自然的选择。

And so what we're seeing is that as people are building these agents, as they're using them, it's just a much more natural place to be running them.

这或许也是为什么越来越多的大型AI实验室都选择 Cloudflare 作为代码部署的首选平台。

And it's I think why you're seeing more and more of the big AI labs have Cloudflare as the preferred target for where where their code gets run.

OpenAI 之前为他们的企业用户发布了一个项目，目标正是我们。

You know, openAI released a project for their enterprise users a little while ago that, again, is is targeting us.

我们希望把 V8 这样的技术作为 Cloudflare 的核心能力，来推动这一未来的发展。

And and and we wanna make sure that with with things like Veat as as first class citizens on Cloudflare that we can help power that future.

因为这不会再是过去和大型云服务商一起构建的那种系统。

Because, again, it's not gonna be kind of the same system that we that we built with the hyperscalers.

它会是完全不同的东西。

It's going to be something different.

再说一次，我觉得我们很有机会打造出这种不同的方案。

Again, And, I think that we have a really good shot of of building that different thing.

那你能帮我更深入地了解一下 CPU 瓶颈这类问题，以及相关的解决方案吗？

So can you help me understand more about the problems of, like, CPU bottlenecks and then maybe some of the solutions?

我刚才在想，带宽肯定是个问题，CPU 负载也是。

I'm just thinking about, like, I I I would think bandwidth would be an issue, obviously, CPU load.

不过，有没有可能大家会就某些东西达成某种共识，比如说是机器人？

But is there a world where we get to some sort of convention around maybe it's the robots.

是纯文本，还是用户代理？

Txt or just the user agent?

当 Google 爬虫或者其他任何 AI 系统、AI 代理出现的时候，你提供的其实差不多就是个 MCP 服务器，只不过你交付的是一个看起来更像 JSON 包的东西，更轻量，对 CPU 的占用也更低。

And when Google bot shows up or any other AI system, AI agent shows up, you're just delivering a it's almost an MCP server, but you're just delivering something that looks more like a JSON package, something that's a little lighter, little less CPU intensive.

这有没有优化的空间？

Is there a path there to optimization?

感觉我们现在就处在这样一个时代，恨不得把每一分性能都榨干。

It feels like we're in the era of, like, squeeze every ounce of performance out of everything.

是的。

Yeah.

但接下来到底会发生什么？

But what's actually going to happen here?

我们是真没辙了，还是有转机？

Where where oh, like, are we just screwed, or is there an opportunity?

不是。

No.

我觉得是的。

I you know, I think so.

我觉得这是两个不同的问题。

I think two two different different problems.

首先，如果你让这些不同的AI系统替你完成任务，尤其是任务变得复杂时，背后必须有某种协调机制。

So the first is, if you if you're as as as you ask these the these different AI systems to perform tasks on your behalf, what what has to happen behind the scenes, especially those tasks get complicated, is that they need to be coordinated in in some ways.

比如你让它们帮你规划一次旅行，背后其实需要各种任务之间的协调。

So if you say plan a vacation for me or something like that, what what goes on behind the scenes is that there there's coordination there.

而要让这个过程尽可能高效，智能代理和这些大语言模型最擅长的，其实就是写代码。

And the best way to be make that as efficient as possible, what what agents are really good and and and and these various LLMs are really good at is actually writing code.

所以这些代码需要有个地方存放。

And so that code needs somewhere to live.

问题是，如果把这些代码放在容器里，就得带上整个操作系统。

And the problem is that if that lives in a container, then you've gotta bring in an operating system.

还得带上所有工具和其他依赖，这实际上非常笨重。

You've gotta bring in all all the tooling and everything else, and that's actually extremely heavyweight.

因此，这就是第一个同时出现CPU和GPU瓶颈的地方。

And so that's the first place that you've got both a CPU and a GPU bottleneck that's there.

而像Cloudflare Workers和隔离环境这样的技术，系统要轻量得多。

And so with with something like Cloudflare Workers and isolates, you it's just a much lighter weight system.

这意味着你可以在同样的CPU基础设施上运行更多代理，提供所需的服务。

And so that means that you can have more agents running on the same CPU infrastructure and and be able to provide that.

而且，我觉得这也是为什么许多下一代工具都选择基于我们的平台来构建。

And, again, I think that's why a lot of these next generation, tools are actually built using our our platform.

你提到另一点，就是当这些代理开始与网络其他部分互动时，你得确保整个过程尽可能高效。

You mentioned something else, which is, you know, as as these agents go out and interact with the rest of the web, they they you wanna make sure that that is done in the most most efficient way.

如果它们下载整页的 HTML，而这些网页本来就是为人类浏览设计的，里面就会有很多无关紧要的冗余内容。

And so if they're pulling down, you know, all of the HTML from a from a web page, if if and and it's and those web pages are designed kind of to be consumed by by humans, there's just a lot of cruft on that that isn't isn't necessarily as important.

所以我们为那些希望内容被代理高效使用的客户，会自动把内容转成 Markdown，这样简单得多。

And so some of the things that we're doing are, you know, for those those customers of ours that wanna make sure their content is consumed by by agents, wanna make sure it's as as accessible as possible, we are automatically converting things into markdown, which is a much simpler Sure.

这种做法能大幅节省 token 使用量。

System that that saves you a ton of tokens.

也能减少大量处理开销。

It saves you a ton of processing.

这样一来，你的上下文窗口就能容纳更多有用的信息。

It means that your context window, can be can you can fit more useful information into a into a context window.

我觉得还有很多优化空间，我们正在从开发者和内容提供者两边入手，让这些代理能力更强、效率更高。

So I think there's a ton of optimizations, and we're helping both on the developer side as well as on the content side, making sure that we can, have have these agents be as powerful as possible and and get as much done as possible.

约翰·格鲁伯，完胜。

John Gruber, total victory.

你知道他是发明 Markdown 的人吗？

You know he invented markdown?

不知道。

No.

不是他。

Didn't.

Grubernator。

Grubernator.

是的。

Yeah.

不可能吧。

No way.

对。

Yeah.

这难道不惊人吗？

Isn't that amazing?

是的。

Yeah.

不过说真的，这属于那种超前于时代的发明，你知道的，结果它就这么成了。

Anyway And, I mean, it's it's one of these things that just, you know, it it it turned out it was it was ahead of its time.

但它对信息压缩至关重要，能让我们把内容尽可能精简。

But it's it's such a key for making sure that we can take information and make it, you know, as compact as possible.

约翰·格鲁伯为这个格式奠定了基础，真是了不起。

John Gruber created the the the format for God.

我们生活的这个世界，真是挺奇妙的。

It's a funny funny world we live in.

上周你们提到，代理流量已经超过了人工流量，能聊聊这个时刻吗？

Talk about, it was last week you guys announced that the threshold had been passed around agent versus human traffic.

说说当时的情形。

Talk about that moment.

这个转折点，是比你预期的来得早，还是晚？

Did it happen sooner or later than you expected?

早得多。

Much sooner.

早得多。

Much sooner.

是的。

Yeah.

我是说，去年年底，也就是2025年底的时候，我就觉得我们会在整体流量上超过人类访问量。

I mean, it was I I at the end of last year, so end of of twenty twenty five, I said that I I thought that we would pass he it bought traffic, and that's that's, you know, across the board.

这不仅包括谷歌的爬虫，也包括那些新推出的智能代理。

So that's, like, Google's crawler, but also, you know, the new agents which are which are coming out.

我当时预计会在2027年底之前超过人类流量。

I thought we'd would pass human traffic by the end of twenty twenty seven.

大约三个月前，根据我们观察到的流量趋势，我把预测提前到了2027年上半年。

About three months ago, I revised that based on the traffic that we were seeing to say that it would actually be in the first half of twenty twenty seven.

结果现在2026年上半年就实现了，这简直太惊人了，也说明了这个趋势增长得有多快。

And so the fact that it actually happened in the first half of twenty twenty six is just is just it's just been extraordinary, and it just shows how quickly this this is is growing.

关键在于，如果我们普通人想买台新相机之类的东西，

And and the and the real key here is that I think that if, you know, you or I as humans, we're researching to go buy an additional camera or something.

可能会逛五个网站，做点研究，比比价格，

We might visit five websites and do, you know, a little bit of research and some price comparison.

你只要看看这些智能代理是怎么工作的就知道了，

You know, you just watch as you use these agents.

它们能无休止地浏览多达五千个网站，只为找到你想要的——最划算的价格、最快的配送、最优的服务，所有相关信息一网打尽，

They have boundless attention to be able to just go to maybe 5,000 websites to find exactly what what you're you're looking for, the best price, the best delivery, the best service, and everything that's that's there.

这就导致了互联网的使用量急剧上升，

And so that's just driving an enormous amount of consumption of of of the Internet.

与此同时，从2015年左右开始，网页数量长期停滞不前，

At the same time, the other thing that's happening is that for a long time, since about 2015, the the web has kind of plateaued.

那段时间，关闭的网站比新建的还要多，

There were not there were there were more websites that were, being shut down than were being created, during that time.

但过去十八个月，网页数量又开始以惊人的速度增长了。

In the last eighteen months, though, we're we're back to the the web growing, at a rate which is is exponential.

而且这种情况，看起来跟两千年初的时候网页增长的态势挺像的。

And and what and and it looks sort of similar to what what was happening back in the early two thousands in terms of in terms of growth of the web.

所以你看，一方面在线内容的使用量在增加，另一方面在线的内容本身也在不断增多。

And so I think you're seeing both more sort of consumption of what's online, but you're also seeing more things online as it as it goes forward.

这两方面都在推动互联网的使用量持续增长。

And so in both of those directions, you know, that's that's that's gonna continue to just drive, you know, more and more use of of the of the Internet.

如果五年后，机器人流量达到人类流量的一千倍，我一点都不会觉得意外。

And and I wouldn't be surprised if, you know, going going, you know, forward, say, five years, that that bot traffic will be a thousand times human traffic online.

我们必须确保互联网能适应这个全新的未来。

And we've and we've gotta make sure that we make the Internet work for for that new future.

是的。

Yeah.

每年，互联网流量中机器人占比都会再增加一个九，达到99.999%以上都是机器人。

Like, every year, we're just gonna add another nine to the nine nine 99.999% of Internet traffic is bots.

对。

Yep.

你知道 baseline 是多少吗？

Do you do you know what the baseline was?

就是说，在 AI 出现之前，机器人流量大概占 1% 吗？

Like, pre AI, were we at, like, 1% bot traffic?

我之前说过

I mean, I've said

不是的。

No.

比那高多了。

Was it was more it was more than that.

大概有 20% 左右，没错。

Was about 20% Okay.

就为了

For for for

很长一段时间了。

a long time.

所以没错。

So it yes.

而且你知道，谷歌是最大的那一个。

And and it's you know, Google was the was the largest.

另外，网上还有很多恶意的程序在到处活动。

And then and then, obviously, there's a lot of of malicious things that that run around online.

但至少在Cloudflare能监测到的这段时间里，这个比例一直维持在差不多的水平。

But that was that was around what what it was for and it was pretty stable over, you know, the history of Cloudflare at least, so where where we could measure it.

所以从2010年开始，这个比例就一直稳定在20%左右。

So so since 2010, you know, it was always kind of around that 20% range.

对。

Mhmm.

但之后就开始增长了。

And then and then it's and then it's grown.

最近几年，它稳步上升，到了2026年上半年，增长速度明显加快了。

You know, it's it's in the last few years, it started growing steadily, and then it really accelerated in the first half of of twenty twenty six.

我想聊聊推理栈，我觉得。

I wanna talk about the inference stack, I guess.

我们现在看到两种趋势在分化，比如苹果在 WWDC 上推出了设备端推理，虽然不是最前沿的，但肯定能在手机和电脑上正常使用。

Like, we're seeing two things play out of sort of a bifurcation, like a WWDC, Apple's launching on device inference that's it's not gonna be frontier, but it's gonna be usable for sure on your phone, on a computer.

他们当然也可以用私有云。

They can obviously go to their private cloud as well.

然后就是新出的 NVL 72 模型。

And then you have, like, the new NVL 72 models.

中间还有像 Cerebras 芯片这样的方案，速度快但价格昂贵。

There's stuff in between on a cerebris chip, fast but expensive.

再往下就是那些能在普通硬件上跑的，不过模型本身还是挺大的。

Then there's everything from, oh, it runs on commodity hardware, but it's still pretty big.

你得有个像样的台式机才行。

You need a real desktop for it.

我想知道，你觉得 Cloudflare 在这里面能扮演什么角色？

And I'm wondering about how you see Cloudflare fitting into that.

在边缘节点上做推理，对 Cloudflare Workers 来说是个很自然的延伸。

It would be a logical extension to Cloudflare workers to have inference on the edge, inference in different places.

你们打算在哪些地方提供推理服务呢？

Like, where do you see yourself offering inference if at all?

是的。

Yeah.

说起来还挺有意思的。

Well, we we we it's it's actually kind of funny.

2022 年的时候，我们发过一篇新闻稿，说要和一家图形芯片公司合作，在网络边缘部署 GPU。

Back in, 2022, we we issued a a press release that there was a graphics chips company that we were gonna partner with in order to put GPUs at the edge edge of our network.

对。

Yeah.

结果那家图形芯片公司

And that graphics chip company turned out

就是 NVIDIA。

to be NVIDIA.

NVIDIA。

NVIDIA.

对。

Yeah.

当时简直没人搭理，一片寂静。

And and and and what was what was amazing was at the time, it was just crickets.

根本没人关心。

Like, nobody cared.

是啊。

Sure.

我们其实做了一点，但没大规模推广。

And but but we we had we had and and so we we sort of did a little of it, but we hadn't really rolled it out broadly.

有趣的是，两年后，突然所有人都开始关注了。

And and what was funny was and then you fast forward two years later, and all of a sudden everyone cared.

所以我们干脆把原来的新闻稿重新发了一遍。

And so we we basically just reissued the exact same press release.

是的

Mhmm.

而现在，你知道，这已经成了我们业务中相当重要的一部分。

And and and now, you know, that's become that's become a pretty pretty important part of our our our business.

所以今天，你可以在 Cloudflare 的边缘运行推理任务。

And so today, you can run inference at the edge of Cloudflare.

对的

Yep.

因为我们遍布全球350多个城市，距离世界上绝大多数人口都只有毫秒之遥，这就让我们成了执行推理任务非常自然的选择。

And because of the fact that we're, you know, in over 350 cities worldwide, you know, we're within within milliseconds of the vast majority of the world's population, we become a very natural place for inference to happen.

我一直默认，大约一半的推理任务会发生在设备端，不管是你的手机、笔记本，还是其他设备。

I I my my working assumption has always been that about 50% of the inference that happens will be on device, whether that's your your phone or your or your laptop or or whatever.

但同时，也需要一个标准协议，让手机、笔记本这些本地设备能把那些耗时较长或规模较大的任务交给附近的网络节点，而我们正好适合承担这个角色。

But that there needs to be some standard protocol where your phone or your laptop or whatever that local thing is can hand those either long running tasks or larger tasks off next to the network, and so that would be to us.

如果某些任务超出了我们的处理能力，那就可以再转交给更强大的中心化数据中心来处理。

And then if if for some reason you need something that's more than that, that it could handle it back to, you know, some centralized data center with, you know, with with more more capacity than than we may have.

我觉得，我们正越来越多地看到这种情况。

And I think that that that's what that's what we're increasingly seeing.

我想，我之前的假设可能有点偏差。

I think my my assumption was a little bit probably off.

实际上，我觉得现在在设备端运行的推理任务会更少，因为越来越多的任务都是这种长时间运行的类型。

I think actually, I think less is gonna happen on device today because I think more and more of the tasks are going to be these long running

是的。

Yeah.

不再是简单地问，纽约今天气温多少这类问题。

Tasks where it's not gonna be just, you know, what what's, you know, what's the, you know, what's the temperature in in New York today.

对。

Yeah.

而是像帮我去规划一次度假，综合考虑各种因素，比较不同地方的酒店，安排好各个地点之间的行程。

Instead, it's gonna be something like, help help me plan a a vacation, go you know, take into account all of these different things, shop for, you know, different hotels in different places, plan all the plan all the, the the travel between between the different locations.

这是我的一些要求。

Here are the criteria that I have.

这可能需要不少时间，肯定不是几秒钟就能完成的。

And that might be something that takes, you know, maybe, you know, certainly not not gonna be seconds.

有时候可能得花上几分钟、几小时，甚至几天，我觉得你会需要长期运行的系统来处理这些任务。

It it might be, you know, minutes or hours or days in some cases to I feel like you're have

我觉得你会需要专门针对帕克城的雪情预报和报告系统，持续追踪每一趟滑雪路线，我目前就在这么做。

I feel like you're gonna have long running agents just for Park City, snow forecasting and reporting, tracking like individual I already do.

每一条单独的路线，你知道的。

Individual runs, you know.

也许会有一群无人机，实时监测哪些地方出现裂缝。

Maybe a network of maybe a network of drones that are identifying what's cracked out,

还有卫星图像，来自行星实验室的数据。

what's Satellite photos, planet lapsed.

帮你优化在山上的行进路线。

Optimizing your routes on the mountain.

很多事情都要做。

A lot of stuff.

没错。

Exactly.

完全正确。

That's that's that's exactly right.

希望今年能下雪，别像去年那样。

Hopefully hopefully it snows this year unlike last year's.

是的。

Yeah.

在边缘设备上做推理时，你们会做哪些权衡？

So, what are the decisions when you're doing inference on the edge?

要说节省个三四百毫秒、六百毫秒，可你面对的是二十分钟甚至十秒的流程，提升也就3%，这真能带来实质差别吗？

Because it's sort of a hard pitch to say, I'm going to shave off three hundred milliseconds, six hundred milliseconds, when you're talking about a twenty minute workflow or even a ten second workflow, you're getting me a 3% boost.

这点提升真能起关键作用吗？

Is that gonna make the difference?

但你是否觉得，未来会有一种中间状态——推理必须在请求周期内完成，且延迟必须控制在一秒以内，那时候延迟才真正重要？

But are you optimistic that there will be sort of like a in between state where there will be inference that happens and it needs to happen within the request loop under a second and then the the latency matters?

是的。

Yeah.

我觉得大概有三种不同的原因，

Well, I think I think there are there are sort of three different

对。

Yeah.

人们选择在 Cloudflare 上运行推理任务。

Buckets why why people are choosing to run inference tasks on Cloudflare.

我觉得第一个原因，跟你刚才说的一模一样。

I think I think the first bucket is exactly what you said.

就是延迟问题。

It's just it's latency.

尤其是那些涉及人机交互的场景，任何延迟都会让人感觉不舒服，所以大家都在努力追求最佳性能。

And so especially for those things that have human and computer interaction where where, you know, any delay feels like it's it's it it hurts you that people are are trying to make sure that they can get the best best performance

能做到最好。

possible.

对于语音模型来说，确实如此。

For voice models, I mean, it like that has yeah.

语音模型必须得在本地运行。

Voice models have gotta be local.

是的。

Yeah.

我很喜欢这一点。

I love that.

对。

Yeah.

所以，确实也有这样的理由。

And so and so that's and so that you know, again, there there there's a case for that.

我觉得

I think

第二个原因是，出于隐私或合规要求，人们希望数据尽可能保留在本地。

the second case is that a lot of times for either privacy or regulatory reasons, people wanna keep things as close to where they physically are Sure.

尽可能就近处理。

As possible.

所以，我觉得尤其是在欧洲和其他地方，很多人认为当初互联网1.0和2.0时代把所有数据都集中到弗吉尼亚州的阿什本、集中到美国，是个错误。

And so, you know, I think that a lot of especially in Europe and otherwise, of places think that, you know, they made a mistake with the Internet kind of one point o and two point o of having everything go back to Ashburn, Virginia, everything go back to The United States.

因此，现在确实有一种强烈的主权意愿，希望把数据留在本地，这一点很重要。

And so I think that there's there's a real kind of sovereign desire to keep things local, and that's important.

第三点是，我们实际上往往能更节省成本，因为我们本身就是一个网络。

And I think the third thing is we we actually can be oftentimes much more cost effective because of the fact that, you know, we we are this we are a network.

而且由于我们与全球各地的网络深度互联，带宽对我们来说几乎是免费的。

And and because of how we're interconnected with networks around the world, bandwidth for us is effectively free.

所以把数据接入我们的网络非常容易。

And so it's very easy for us to get things onto our network.

再加上我们部署系统的位置通常不需要支付场地或电力费用，这些节省下来的成本就能直接让客户受益。

And because of where we deploy, systems, we often are in places where we don't have to pay for the space or the power, which allows us to then pass those savings on to customers.

因此，使用我们的服务进行推理运算，成本要比在其他地方低得多。

So it can be significantly more cost effective, to use inference with us than it can be in some other other places.

这听起来有点反直觉，但这也正是我们所构建的东西的本质，更是Cloudflare能持续提供如此强大服务的关键所在。

That's somewhat counterintuitive, but but but it's sort of the nature of what we've what we've built and kind of the secret to what's allowed Cloudflare to to to to to deliver as much as we have over time.

关于数据中心的建设方式、规模大小、需求多少，以及各方的反对声音和监管问题，争论可太多了。

There there's a ton of debate ton over data centers, how they get built, how big they should be, what we need, the pushback, regulation, etcetera.

这些因素对你们的业务有影响吗？

How are you is any of this affecting your business?

你们有没有考虑过如何在现实世界中布局自己的网络节点？

Are you thinking about how you position your footprint in the real world?

你们是怎么看待当前关于数据中心建设的这场持续演变的讨论的？

How are you processing the the evolving discussion around data center construction?

确实如此。

Yeah.

首先，我们肯定还需要建设更多数据中心。

I mean, I so so first of all, I mean, we're we're gonna need more data centers.

我觉得很多讨论其实有点荒谬。

I think a lot of a lot of the discussion is is somewhat silliness.

我的意思是，用水量——这些其实都是闭环系统。

I mean, the the water, consumption I mean, these are closed loop systems.

一个高尔夫球场一年的用水量，可能比美国所有数据中心加起来还要多。

I mean, a a a golf course uses more water than than probably all the data centers in The United States over over the course of over a year.

是的。

Yeah.

我觉得很多所谓的担忧其实挺荒谬的。

So I I think that I think, you know, there there's there's there's a lot of kinda silly concerns that are there.

但确实也存在一些真正值得关注的问题。

And but but there are but there are real concerns as well.

我们得确保给这些设施供电的方式足够高效，别挤占电网本可以支持的其他系统资源。

We've gotta make sure that there are efficient ways to get power to these things, that it's not taking power from from other other systems that the grid can support support that.

所以我觉得这些都是很合理的考量。

And so I think those are all all all good considerations.

对Cloudflare来说，我们的情况有点不一样。像AWS、谷歌，或者SpaceX这类公司，通常会建一两个超大规模的设施，把大量设备全塞进去。

For Cloudflare ourselves, you know, it it we're we're we're a little bit different where in in the case of, you know, if you're a a AWS or a Google or, you know, a SpaceX, you're building one big facility or a handful of very big facilities and putting a ton of machines in that that one facility.

Cloudflare的情况不一样。

Cloudflare is different.

我们有大量设备，但把这些设备分散在全球各地。

We have a ton of machines, but we spread them massively all around the world.

我们常常希望进驻网络交汇的地方。

And oftentimes, we we wanna go into the places where networks come together.

这些地方可能是世界上一些最老旧的数据中心。

And so those those can be, you know, some of the oldest data centers in the world.

在这些设施里，我们可能只部署几百台设备。

And in any of those facilities, we may not have we may we may have only hundreds of machines.

但把全球所有设备加在一起，我们的总量实际上远超任何一个单独的数据中心。

But over collectively around the world, we've got, you know, what is a, you know, effectively much larger than any individual data center, which is which is out there.

所以我觉得，我们受新建数据中心的影响比较小。

So I think we're less less impacted from, you know, the new builds of data centers.

我们更受制于如何融入现有的数据中心，以及如何确保部署的设备尽可能节能，因为通常人家会给我们一个固定的空间，让我们把设备塞进去，而我们往往是作为客人，入驻当地ISP或者合作方的设施——全球成千上万栋建筑里，我们都是这样运作的。

I think we're much more impacted by how do we find our way into the existing data centers, and then how do we make sure that the equipment that we're deploying is as power efficient as possible because we're often given sort of here's the envelope that you have to fit your fit your your equipment in, and we're often guests of, you know, whoever the local ISP is or or or or whoever, you know, partner is that we're working within in the, you know, the the thousands of buildings that we're in all around the world.

意大利那边有什么新情况？

What's new with Italy?

是啊。

Oh, yeah.

对。

Yeah.

意大利和西班牙。

Italy and Spain.

就是说，就是说，就是说——

Like, it's it's it's it's

但你是在为意大利争取资源吧。

But you're So you're fighting for Italy.

你也在为法国争取资源。

You're fighting for France.

你们夏天还得应对全球性的需求。

You're summer We need world

这里和平得很。

peace here.

别开玩笑了。

Come on.

任务是有期限的。

Are limited duty.

没有期限。

Unlimited.

是的。

Yeah.

我不去伊比萨，真的吗？

No no no Ibiza for me or or Really?

多重海岸。

The Multi Coast.

不行。

No.

说实话，我觉得这事挺有意思的。

It's it's you know, I think it's it's interesting.

对我们来说，这其实一直挺让人困惑的。

It's been actually kind of puzzling for us.

所以这一切都要追溯到以前——如果有人还不知道的话，在意大利和西班牙，有人一直对我们个人或者Cloudflare采取非常强硬的手段。

So all of this stems back so so for those who don't know, in in both Italy and in Spain, there there's been pretty aggressive tactics to come after either me personally or or Cloudflare.

而这一切主要都是由那些地方的足球联赛推动的。

And it's all been driven largely by the football leagues, the the soccer league in those in those places.

他们担心的是盗版问题。

And what they're concerned about is is piracy.

讽刺的是，我们自己也不喜欢网络上的盗版行为。

The the thing that's been ironic about it is, like, we don't like piracy on our network either.

我们从盗版里一分钱也赚不到。

We don't make any money off of it.

盗版只会浪费带宽。

It's it takes bandwidth.

它浪费资源。

It steals resources.

所以我们专门有个团队全天候工作，专门打击盗版行为。

And so we have a whole team that's working all the time to shut shut the pirates down.

但盗版者挺聪明的，总能找到办法利用我们的系统。

And and yet, you know, the Pirates are clever.

毕竟，互联网上很大一部分流量都经过Cloudflare。

They find ways to to use our our system because, you know, huge percentage of the Internet sits behind Cloudflare.

时不时总会有些内容我们没及时捕捉到。

You know, there's gonna be stuff from time to time that we don't we don't capture.

要发现这些内容，我们需要一些时间。

It takes us some time to catch.

在世界其他地方，比如NBA、NFL、MLB、英超联赛等，我们都和他们紧密合作，只要他们发现盗版直播，我们就立刻下架。

With leagues everywhere else in the world, the NBA and NFL and and MLB and and the Premier League in in The UK and and others, we work really closely with them to if they identify a pirate stream for us to, you know, pull that down.

因为这同样会给我们带来成本。

Because, again, it costs us money.

我们不喜欢这样。

We don't like it.

但不知为什么，西班牙的西甲和意大利的联赛却决定不跟我们合作打击盗播，反而选择起诉我们。

But for whatever reason, La Liga in in in in Spain and and the and the and the league in Italy have have decided that instead of instead of partnering with us to take this down, they they would sue us instead.

所以，没错，这确实打乱了我一些暑期安排。

And so so again, yes, it it it does cramp some of my summer plans.

我想聊聊上市和成为上市公司后的日子。

I wanna talk about IPOs, life as a public company.

今年有几位创业者正打算上市。

There's a few entrepreneurs that are trying to go public this year.

你们的IPO非常成功。

You had a very successful IPO.

自上市以来，你们的股价已经涨了超过十万基点。

The stock is up more than a 100,000 basis points since you went public.

表现太出色了。

Fantastic run.

我们现在建议把所有内容都换算成点数。

We're advice for putting everything in points now.

听起来更棒了。

Sounds even better.

我的意思是，这场谈话讲的是百分之十万的增长。

I mean, this talk's about thousand percent.

太出色了。

It's fantastic.

但说十万点，听起来更震撼。

But a 100,000 basis points sounds even better.

点数可真不少。

Basis points a lot.

再多一百点，对。

100 more yeah.

十一万点。

110,000 basis points.

谁在数啊？

Who's counting?

不过话说回来，你的流程是怎样的？

But, anyway, what was the what was your process like?

我们刚才在聊股权锁定期的问题。

We were talking about the level of float lockups.

比如，哪些决定特别难做？

Like, what were the hard decisions?

现在这些大公司上市时讨论的那些事，当初在你待办清单上连前十都没进过吧？

Were there things that we're talking about now with these big IPOs that weren't even on the top 10 of your to do list?

什么样的做法对成功上市和运营上市公司最重要？

What is important to a successful IPO and then running a public company?

是的。

Yeah.

我觉得，有那么几件事特别值得我们借鉴。

I think, you know, there are a handful of things that that that that are, you know, interesting lessons for us.

首先，我真的特别喜欢做一家上市公司。

So first of all, like, I love being a public company.

说实话，我觉得这挺有意思的。

It's you know, I I think that it's interesting.

你知道吗，在那些没有无过错离婚制度的地方，配偶杀人案的发生率要高得多——这其实很像私人市场投资者和公开市场投资者之间的区别。

You know, the the the in in jurisdictions where, yeah, you don't have no fault divorce, spousal homicides are much higher, which is I think kind of an analogy to the difference between private market investors and public market investors.

想解雇你的风投，真的很难。

It's really hard to fire your VC.

你的风投想解雇你，也不容易。

It's hard for your VCs to fire you.

所以结果就是，这种关系其实挺不健康的。

And so as a result, like, actually, it's kind of dysfunctional.

但公开市场就不一样了，我特别喜欢我们的股东，因为要是我干了件蠢事，他们会直接打电话来告诉我：你这事儿办得太蠢了。

Whereas public market, like, I love our investors because if I do something stupid, they call me up and they say, that was stupid.

然后就把我的股票卖了。

I sold your stock.

然后我们会聊一聊，有时候他们会说，哦，现在想想确实有道理，就又买回去了。

And then we and we have a conversation and sometimes they say, oh, actually, that makes sense now, and I bought it back.

对吧？

Right?

是的。

Yeah.

但我觉得，你其实能进行更坦诚、更真实的对话。

But I think you can have actually a much more honest, much more real conversation.

而且这种互动感觉要健康得多，比那种在私募市场里的情况强多了。

And and and that's and that that's felt actually just a lot healthier than what you see in the third Yeah.

确实如此。

It's so true.

几乎所有天使投资人，只要投过几家有影响力的企业，都会遇到一些项目，早就放弃希望了，但还是尽量保持礼貌，能帮就帮一把。

Mean, there's everybody that's angel invested in any number of meaningful companies is going to have portfolio companies where you've fully given up on the company and you you just try to be try to be nice and, yeah, help if you can.

但到了某个阶段，你跟这笔投资已经完全脱钩了，这其实就是现实。

But at at some point, you're just totally disassociated with the investment and you're just like and and that's that's that's the reality.