Cloudflare CEO：互联网需要加密来应对AI

The internet for a long time is growing like crazy. And then it kind of stalled out for the last little bit. And the number of like new websites that were being created, number of new domains, kind of plateaued from about 2012, 2011, 2012 until actually quite recently. What's changed is I think that because a lot of these tools have allowed more people to be creators, to be, you know, coders, to be developers, they actually for the first time in really the last 15 years, the number of websites that are being created, the number of new things that are happening is actually is actually growing again and growing [music] at rates that look like what the early 2000s growth rates of the internet um look like.

So I actually think that's kind of optimistic [music] that there's more things that are being created that are out there and are and are are coming online.

[music]

Bankless Station very excited to introduce you to Matthew Prince. He's the co-founder and CEO of Cloudflare. That's a $70 billion company that sits between [music] the internet and roughly uh 20% of the world's websites run on top of Cloudflare. I think he's one of the most thoughtful voices on the question of how is the business model of the internet changing? He's also recently joined X402, got some stablecoin stuff to talk about.

Matthew, there's so much to discuss today. Welcome to Bankless. Thank you for having me. So we've seen estimates that um at a certain year AI bot traffic is going to exceed all human traffic on the internet. I don't know if that's 2027, 2028, 2029, or maybe even this year. Uh what do you what when do you think that happens and is the internet ready for it? Yeah, you know, I think that based on what we've seen, um we were saying it was 2027, we're now saying it's the first half of 2027 uh that that will will cross.

It's just been it's been extraordinary to watch the growth of agentic traffic, bot traffic uh over over the last little bit. And for a long time bots were bad, um but you increasingly are seeing these these agents which are actually helping humans trying to do things which are which are really useful. Just driving an enormous amount of traffic. And I think any of us who have used any of the AI tools can see how that happens.

If if I'm personally going to go like think about buying a new digital camera, I might like go to site or two, you know, a few a few pages that are that are out there and then I and I'll make a decision. Someone who's maybe more more thorough than I might go to five, you know, sites and compare reviews. You The agents are you can watch it as you ask, you know, a chat GPT or a Grok or a or a Claude to do work on your behalf.

They're they're taking, you know, often times a thousand times more in-depth research. And that means there's just a lot more traffic that is that is hitting hitting the internet. In terms of whether the internet's ready, I think that's that's a a little bit of an open question. I think one of the challenges, you know, from a from a technical perspective, I think we can manage that. And and we can scale through it. We've we've seen explosive growth in internet traffic before.

I I I hearken back to like April of 2020 when COVID hit and all of a sudden overnight internet traffic doubled over the course of two weeks. And it was extraordinary how much the system still held together. And while it it it created challenges, it was able to scale and we were able to keep the internet functioning and help us get through that that time. I think this is the same sort of moment where you're going to see a lot more more traffic.

So I think technically we'll be okay. I think the big question though is what's the business model going to be and who's going to pay for it because there's there's a lot of servers, there's a lot of infrastructure, there's a lot of stuff which is behind all of that. And if the business model of the internet for the last, you know, 30 years has been ads and subscriptions, the problem is agents don't click on ads. And you know, buying one subscription and then having agents be able to basically pick pick all of the content back up from that, that's not going to that's not going to help make sure that the people who are creating that content get compensated for it, the people who provide the infrastructure to power that compensate that that content get get actually compensated for the work that they're doing.

And so we have to figure out some new business model, and that's what that's a lot of the what I'm I'm spending my time thinking about these days. I can't wait to discuss that with you. I think that's going to be the bulk of the episode. Before we get there into the the new business model that might be required, why ads are breaking down, why subscriptions are breaking down. Talk about Cloudflare for just a second because I think our audience may not know exactly what Cloudflare does even though they're probably a daily active user of Cloudflare because if if Cloudflare turned off tomorrow, big chunks of the web they would just go completely dark.

And you guys have been part of the story of how the internet has scaled over the last decade or so. One one way I think of Cloudflare is sort of like, you know, that ice wall from Game of Thrones where it's just like this giant ice wall that kept the the White Walkers out and all of the the bad bots and you know, Russia and and North Korea and just malicious things. You guys play a role in that. I know that's not all you do though.

Talk about how would you explain Cloudflare to I guess, you know, a normal tech-savvy audience member? So, you know, I think it's it's kind of like, you know, any any superhero needs an origin story. And so our origin story started pretty simply, which was we saw that all of the world of software was going to the cloud. We saw that all the world of like servers were going to cloud. And so it seemed obvious to us that sort of networking and the security appliances that were out there before the firewalls of the world were going to go to the cloud as well.

And so Cloudflare started with a really simple idea, which is how could you take a firewall and put it in the cloud? And and and one of the challenges we had was we knew that in order to be successful we had to then someday sell to, you know, big banks and governments and healthcare organizations and a and things like that. Um but but we were tiny. And so, um you know, the the the the challenge was we didn't have a big network.

Uh we didn't have the data to really stop the most sophisticated threats of the time. And so, one of the things that we did was we made uh a stripped-down version of the service available to everyone in order to basically help us bootstrap getting the network built and also um uh to to help get the data to to be able to identify what the threats and challenges were. And um and and and the combination of decisions um really caused us to then have what were just a series of almost endless problems.

Um you know, the next thing we knew, uh we had every human rights organization in the world that signed up, which meant that then every sort of authoritarian regime in the world was attacking us trying to knock them offline. And so, we had to build, you know, not just a firewall, but now we had to do DDoS mitigation. We had hacker kids that were, you know, trying to, you know, literally steal our domain. So, we had to build a registrar.

Um we had to our own employees that were getting, you know, attacked as they would go online. So, we had to build our own VPN service that was there. We even had to build our own um uh developer platform in order to um be able to keep up with all the feature requests and be able to build things ourselves. And so, the story of Cloudflare has really been like start with this simple idea, how do you put a firewall in the cloud, make it available to as broad a set of people as possible, have a series of problems that that the

原文超出正文长度上限，此处截断——上游还有内容，完整版见上方「原文 ↗」。