Agents Don't Click Ads: AI Kills the Old Internet
60sIt distills a massive industry shift into one shocking idea: AI agents don't click ads, so the internet's entire ad-supported model breaks.
▶ Play Clip"Sensational 'extinction event' framing overshoots a nuanced interview where the CEO says he's actively playing for the opposite outcome."
Cloudflare CEO Matthew Prince discusses how AI is transforming the internet's business model, from bot traffic surpassing humans to the need for microtransactions. He also explains why Cloudflare cut 20% of its workforce and the risks AI poses to small businesses.
Cloudflare is riding the AI tailwind as a generative AI cloud company. CEO Matthew Prince sat down at Cannes Lions to discuss growth and the company's role in the shifting internet economy.
Google defined the internet's business model for 28 years, but the next 28 years will be different. Cloudflare is engaging with brands and advertisers to understand the new model.
AI is a platform shift comparable to browser, social, and mobile. Consumers will increasingly get information and do commerce through AI interfaces.
A third of the world's population already uses generative AI tools within 3.5 years, which is twice the adoption rate of mobile technology.
For the first time, automated/agent traffic passed human traffic two months ago. If it accelerates, automated traffic could be 1000x more in 5 years, making humans a rounding error.
Cloudflare processes about 500 million internet requests per second. 1–10% of that could warrant microtransactions, requiring a financial system that scales from 10 million to 100 million transactions per second.
Cloudflare's 20% workforce reduction was AI-driven. AI tools can now measure and audit more efficiently than humans, reducing the need for middle-management and measurement roles.
Previously Cloudflare audited 6 of 100+ risk areas quarterly due to human constraints. Now it can audit all risk areas continuously, entirely through AI-built tools.
Cloudflare is increasing direct reports per manager from 6 to up to 12, which improves productivity. The company is still hiring builders and sellers aggressively.
AI agents will optimize for the best deal globally, ignoring local convenience and emotional attachment. This could threaten small businesses and new entrants that lack established reputation.
Prince argues that AI could massively consolidate power, but Cloudflare is building tools and payment infrastructure to keep the internet open and allow anyone to reach a global audience. He remains optimistic that with the right mechanisms, small businesses can survive the shift.
What percentage of the world's population uses generative AI tools within 3.5 years of launch?
About a third (1/3), which is twice the adoption rate of mobile technology.
01:36
What happened two months ago regarding online traffic?
Automated/bot/agent traffic passed human traffic for the first time.
02:28
How much automated traffic is predicted in 5 years?
A thousand times more automated traffic than today, making humans a rounding error on the internet.
02:41
How many internet requests per second does Cloudflare process?
About half a billion requests per second.
03:08
What percentage of Cloudflare's traffic might be suitable for microtransactions?
Between 1% and 10%.
03:15
What scale does Cloudflare think its microtransaction system needs to reach?
10 million transactions per second at day one, scaling to 100 million transactions per second.
03:22
According to Peter Drucker, what are the three functions in any organization?
Build products, sell products, and measure things.
04:57
How has Cloudflare changed its audit function?
It now audits all 100+ risk areas continuously instead of picking six each quarter.
05:36
What is Cloudflare's new manager-to-report ratio target?
From six direct reports per manager to up to twelve.
06:04
Why does the CEO say they did the RIF early?
Because it was the kindest thing for the team, allowing laid-off people to get placed in great roles before the market floods.
07:10
What does the CEO fear for small businesses in the AI future?
Agents will choose the best deal globally based on price and product, ignoring time efficiency and emotional attachment, so new entrants without reputation may fail.
08:14
Generative AI adoption rate
Provides a concrete, surprising statistic: a third of the world uses generative AI in just 3.5 years, doubling mobile's adoption pace.
01:36Humans as a rounding error
A striking projection that automated traffic could make human internet usage negligible, underscoring the scale of the shift.
02:41Continuous audit with AI
Illustrates a concrete operational transformation: moving from quarterly sample audits to real-time continuous monitoring, enabled by AI.
05:36The agent's cold calculus
It captures the heart of the small-business threat: AI agents optimize purely on price and product, ignoring human factors like loyalty and convenience.
08:14[00:02] >> Cloudflare is riding the artificial intelligence tailwind as generative AI cloud company. Yahoo Finance's Gator Baiter Brian Sozzi sat down with Cloudflare's CEO Matthew Prince at Can Lions to learn more about the growth
[00:16] >> Cloudflare is is firing on all cylinders. It's just been amazing to see that as we've built up what is one of the world's largest networks that more and more companies are trusting us to provide their network security to make
[00:29] think the thing that's been kind of the real unlock for us and the reason why we're here in Can is the internet's business model is changing. Um Google defined the business model of the internet for the last 28 years and um
[00:43] the business model of the next 28 years is going to be different and I am here talking to brands, talking to advertisers, talking to uh companies going to look like and I think Cloudflare is going to play a role in in
[00:55] was listening to I think it was an investor day for you all uh in earlier June and you told me what you said at the time on the investor call or analyst will be different over the next 10 years. What or how will the internet be
[01:10] >> You know, I think the internet's had platform shifts before. Uh we went from platform shifts before. Uh we went from a browser to social to mobile. AI is a platform shift. The way that consumers are consuming information is
[01:24] through that AI interface and that's the way that they're consuming information to uh do commerce going forward and the adoption rate has been just amazing. A third of the world's population is
[01:36] already using generative AI tools in just 3 and 1/2 years. Um that to give you some sense, that's twice the adoption rate that we saw with mobile, technology that's out there. And so as these things become the the place where
[01:50] we consume information, what that means is if you're a platform, if you're Yahoo Finance or or or the New York Times or or anyone else, And so the old internet business model of showing advertising or selling
[02:03] subscriptions isn't going to work in this new world where information is consumed in the AI platforms themselves. Um your agents don't click on ads. And so that's going to require us to find some other way to make sure that content
[02:16] creators like yourself and and others continue to get compensated for what is >> What is the way? What is the way towards survival? I I I don't know that I know exactly the
[02:28] answer. But I But I have a feeling it's a couple of things. So first, um for the first time in history, 2 months ago, automated traffic, bot traffic, agent traffic online passed human traffic. And if it continues to accelerate at the
[02:41] rate that we're seeing, in 5 years it will be a thousand times more automated traffic online. In other words, humans will be a rounding error on the internet uh effectively. What that means is there's going to be a whole bunch of
[02:55] additional infrastructure, which is needed in order to build that up, in to have to pay for that. And so I think that there's going to be some version, very beginning, of if an agent is accessing a page, there's going to be
[03:08] back. But to give you some sense, Cloudflare, um we process about half a billion internet requests per second all all day long. We think between 1 and 10% of that will be appropriate for some
[03:22] sort of microtransaction, which means that we have to build a little financial system that can do tiny little transactions at day one, 10 million transactions per second, to scale very in a really reasonable time to 100
[03:34] million transactions per second. To give you a sense, Visa, the largest payments transactions per second. So there's a lot of really cool technology that we have to build in order to support this future. But I think micro payments, I
[03:47] think having a marketplace for content, making sure that you get to choose who gets access to that content, and then maybe, you know, an ads platform but directed at agents rather than being directed at humans. These are all things
[03:59] that our team is working on and thinking about, and they're going to be necessary in the future that is able to support the agentic traffic that we're seeing. at my desk this early morning, and the news crossed of your
[04:13] RIF, reduction in force. And I sat back, and I'm like, all right, I'm not going tough day, and I get it. Don't want to be that guy. But I I'm like, this >> Yeah. >> Why would you need
[04:26] to let go of 20%? Like that is That's a That's a large number for a company did these people come from? >> Yeah, so so So, first of all, incredibly emotional day. A lot of amazing people who who we parted ways with. Um and and
[04:41] now to say this, but it really was driven by AI. And what we found was If you you go back to old Peter Drucker books, like on like, you know, here, you know. >> Yeah, sorry. Um but, you know, he talked
[04:57] about there are sort of three different functions within any organization. There are people that build products, there are people that sell products, and that's where really kind of the business drives value is in in building and
[05:09] whole bunch of other people inside of an organization that actually just measure things all the time. The auditors, the finance people, you know, a bunch of the the marketing team, all the operations roles that are there. And what we've
[05:22] found over the course of the last couple years is that a lot of these tools are able to better measure than humans were before. So, I'll give audit function just like any public company does. Yeah, we have a hundred
[05:36] plus risk factor areas, and every quarter we would pick sort of six in through that process. That was constrained because it required people in order to do it. We've now moved to we're moving to a world where
[05:51] we can take all of those different risk areas and audit them continuously all the time. And we can only do that because we have built these tools. The other thing is powerful is the number of direct reports that can work for a
[06:04] manager scales significantly more as you have more tools to do that. So a lot of again, great middle managers on our team who are going to be amazing senior managers at a startup or or another organization. A lot of them we were just
[06:17] organization. We need to go from six direct reports per per manager to up to 12. That's better for everyone. We found that this has made us more productive. accelerating our ability to deliver products. And we are hiring like crazy
[06:31] builders and sellers, but we're using the machines in order to do much more of >> middle managers under pressure, those jobs will get re-purposed. But again, these are amazing people. They're yeah, I mean and and and I think that
[06:43] the thing is when I talk to peers everyone says, yeah, I'm going to do that. I'm we're going to have to do the same thing. But they're like, I'm going to do it when everyone else does it. And honestly, I think that that's that's
[06:57] poor leadership because in 6 to 12 months when we're going to see more of harder for these those people laid off at that time to get a job. It'd be behind everyone else and just say, listen, I'm a sheep in the flock and
[07:10] everyone's doing it, so I'm doing it, too. I think once we realized it was decision it was the kindest thing that we could do for the team was to do as really hard to make sure that those people then get placed with great roles
[07:22] cuz they're great people. >> I'm thinking, you know, get through it. But like what happens to these small businesses up and down this boardwalk who don't have the capital to invest in AI? To me, they're going to
[07:34] their next 5 years look like? >> You know, last year when I was at media was really at risk. We really needed to reinvent um the business the last year we've actually kind of got some good ideas and some traction and
[07:48] and we're seeing the big media companies like Condé Nast and and People Ink are creating new revenue streams for them. And I'm I'm proud of the fact that we've helped catalyze some of that. This year what I'm talking about is how dangerous
[08:00] to small businesses AI is going to be. Think about it this way. In the future more commerce. And if you think about why why do you do business with a small business today? Usually it's because it's it's time efficient, it's
[08:14] the local bodega when you're, you know, in New York. Or because you have some emotional attachment to that place. Your agent doesn't care about any of those things. It's going to literally just say, "Hey, I'm going to find the best
[08:27] deal for the best product wherever it is in the world. And if I have to search that in a second." >> It's like a killer. It's just like >> And so what I am deeply worried about is
[08:40] brands are fine cuz they have an established relationship with consumers. They have a reputation which is online. But how does a new entrant, how does hanging my shingle up for the first time." How do they get the reputation?
[08:54] stands out? And if we don't create a way for that to happen, if we don't have new then we're going to see massive consolidation over time. And that's bad for everyone. We want to have anyone be able to survive. And so I'm spending a
[09:08] lot of time meeting with, you know, the leaders at at Visa, American Express, Shopify, Salesforce saying like, "We've got to do something in order to make tools in order to succeed." >> is kind of is there there are no small
[09:20] >> That's right. And and we're again I was I was meeting with the the um the guy who used to run uh PayPal up until recently and he ran small businesses at Intuit and he said, "I worry that in the future there are five companies left.
[09:32] One that holds the real estate, uh one that holds the money, one that builds company. Now, that's not what's going to happen, but I do think that AI, given its natural kind of tendencies, is this massive consolidating force. And so,
[09:46] what we're playing for is not a world where there's, you know, two to five AI companies. Not a world where there's, you know, a very small number of media it's consolidated massively, but where everyone can actually they have a good
[10:00] reach a global audience. And not a world where we have a few mega like a small business, wherever you are in the world, can do what the internet only can't lose that, and that's the world that we're playing for.
[10:14] scared the hell out of me, but we're going to we're going to leave it there. We continue this conversation in Davos, hopefully. All right, good to see you.
⚡ Saved you 0h 10m reading this? Transcribe any YouTube video for free — no signup needed.