Aug 7, 2025 · 45m · no-priors
No Priors Ep. 126 | With Cloudfare CEO Matthew Prince
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Cloudflare CEO Matthew Prince explains how generative AI is dismantling traditional web economics, advocating for new crawler standards, dynamic publisher revenue-sharing, and distributed edge infrastructure to power autonomous AI agents.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 16.9% of the talking time here. How this is scored →
The hosts as informed peer 5.8 Guest teaching 5.8 Guest disagreement 2.9 The hosts pushing back 3.1
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 18:28 Rejection of the human content ceiling hypothesis
Matthew directly rejects Elad's suggestion that AI performance benchmarks like MedPalm II mean we are running out of valuable human-generated content.
Hardest push from the hosts ▶ 20:00 Guo questions AI lab willingness to match publisher ad economicsSarah directly questions Matthew's premise of large payouts to creators, pointing out the massive economic discrepancy between open web ad revenues and existing data licensing budgets.
Biggest teaching moment ▶ 4:45 Data on referral traffic collapse across AI enginesMatthew educates the hosts with precise internal and industry metrics showing that getting clicks from OpenAI is 750x harder and from Anthropic is 30,000x harder than legacy Google search.
The host holds their own ▶ 18:03 Gil cites MedPalm II benchmarks to challenge human content scalingElad demonstrates deep technical command by citing Google's MedPalm II medical benchmark outperforming physicians to question the long-term dependency on human expert writing.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Founding Cloudflare and Building Beyond the Traditional CDN | 4 | 5 | 3 | 2 | Sarah introduces Cloudflare's scale and describes them as the largest CDN. Matthew politely corrects the framing, explaining that Cloudflare was built as a cloud security firewall rather than a traditional CDN. | |
| The Demise of Search and Shifting Web Economics | 5 | 6 | 2 | 2 | Sarah and Elad query the shift from search to agentic AI interfaces. Matthew provides concrete empirical figures showing how much harder it has become for creators to get referral clicks from Google, OpenAI, and Anthropic. | |
| Copyright Limitations and Scarcity via Content Independence | 6 | 6 | 3 | 2 | Matthew breaks down why relying on copyright law fails content creators due to fair-use protections on derivative works, outlining Cloudflare's crawler-blocking initiative. Elad draws an apt technical parallel to robots.txt evolution. | |
| The Swiss Cheese Metaphor and Knowledge-Based Incentives | 6 | 6 | 3 | 4 | Matthew uses a Swiss cheese metaphor to explain how AI models prune redundant data and need unique knowledge. Elad pushes back by comparing this to existing data labeling companies like Scale and Mercor, which Matthew distinguishes using Spotify's demand-fill model. | |
| Future Publisher Monetization and Avoiding Monolithic AI Silos | 7 | 6 | 4 | 6 | Elad pushes Matthew on whether human content reaches a ceiling by citing Google's MedPalm II surpassing human doctors, while Sarah challenges AI lab willingness to pay compared to open-web ad scale. Matthew firmly rejects the content ceiling premise and critiques naive lump-sum licensing deals. | |
| Edge Inference Architecture and Support for Open Models | 5 | 5 | 3 | 3 | The hosts inquire about edge inference architecture and open versus closed models. Matthew explains Cloudflare's edge GPU deployment and dismisses existential security arguments against open-source AI as flimsy. | |
| Leveling Rules for Google and Long-Tail Content Markets | 5 | 6 | 2 | 2 | Matthew outlines the need to strip Google of privileged crawler status to establish an open content marketplace for long-tail publishers. Sarah asks for concrete advice for publishers who cannot predict model incrementality. | |
| Inference Efficiency, Model Pruning, and Agentic Infrastructure | 7 | 7 | 4 | 5 | Matthew critiques Nvidia's lack of focus on power efficiency and hyperscalers' unaligned GPU rental models, praising DeepSeek's computer science innovations. Elad presses on the separation between underlying chip architectures and model compression techniques. | |
| Micropayments, Agent Identity, and Cryptographic Web Permissions | 7 | 5 | 2 | 2 | The discussion covers cryptographic agent identity, microtransactions, and browser scraping. Elad actively contributes by highlighting zero-knowledge proofs as the natural credentialing mechanism for autonomous agents. |
Statements from this episode (21)
Prince: Cloudflare built its registrar after an attempt to steal its domain
“Someone almost hacked or stole our domain at some point as a way of hacking into us. So the next thing you know, we built our own registrar.”
Prince: Google Clicks Are 10x Harder to Get Than 10 Years Ago
“The data that we have also substantiates that and shows that compared with 10 years ago, it's become 10 times harder for the same piece of content to get a click from Google than it was before.”
Prince: OpenAI and Anthropic Refer Traffic 750x and 30,000x Less Than Google
“In the case of someone like OpenAI, it's 750 times harder than it was with the Google of old. With the case of Anthropic, it's 30,000 times harder than the content of old”
Prince: AI Traffic Drop Could Starve Web Content and AI Training
“What I worry about is that if the value creation model of the web has been all about how do I get traffic The new interface of the web isn't going to send you traffic, and if that's the case, if content creators can't get value from selling a thing or a subscr…”
Prince: 80% of major AI companies are Cloudflare customers
“80% of the major AI companies are Cloudflare customers.”
Cloudflare blocks AI training crawlers by default across all customer sites
“And so what we did on July first was we announced what we called the Content Independence Day, where you could actually have independence from these AI companies. And we, for free, across all of our customers, whether they paid us or they didn't pay us, we sta…”
Prince: Prominent AI companies use state-hacker tactics to bypass scraping blocks
“In fact, some really prominent companies That do some very, very, very shady things where they basically say, absolutely, we follow the rules of the road, but when push comes to serve, if it turns out they're blocked, then all of a sudden they're doing a bunch…”
Prince: Cloudflare will soon publicly expose misbehaving AI scraping companies
“So, you know, watch our blog, and I have a feeling that some prominent AI companies that are misbehaving are Can get called out pretty soon.”
Prince: AI Markets Should Reward Creators for Filling Knowledge Gaps Over Traffic
“Their very algorithms, as they come across a piece of content, they prune off that content Which is already kind of part of the meaty part of the cheese, whereas the parts that are in holes are actually super valuable to them, and so I actually think that if w…”
Prince: Musicians make millions writing songs to fulfill unmet Spotify queries
“Spotify actually looks at queries that people have run that they don't have good answers for, where there's, you know, somebody searches for, I don't know, I want a disco song about like how fun it is to dance with your dog, right? I don't know. And if somebod…”
Prince: Fixed-Sum AI Licensing Deals Are a Mistake for Content Creators
“I think the mistake that a lot of content creators did was they actually did deals that don't scale as the business models of the AI companies scale with them. So if you do a deal that's like, twenty million dollars and you get all my content, I mean, that's a…”
Prince: In-network edge inference will handle models too large for end devices
“We believe that a lot of inference is going to happen on your end device, but there will always be some model which is too big or too resource intensive. And in that case, the next best place to run it is going to be on at the inside the network at the edge.”
Prince: Regulate pathogen printers, not open-source AI models
“I'm actually quite skeptical of the, if we allow open source models, the world is going to end arguments. That seems histrionic to me. There are things we should worry about. Like it is, you know, I think some of the you know synthetic pathogens and other thin…”
Prince: AI companies fear Google most due to its privileged content access
“Why Google is so important is when you talk to all of the other AI companies, Google is the one company that they're the most afraid of. And that the reason that they're most afraid of them is because they think that they have privileged access to content in a…”
Prince: Cloudflare could negotiate pooled AI licensing deals like Spotify
“Where I think we can play a role is when you have either a large content provider trying to make a deal with all of the AI startups that are out there, which they really do want to do, and they want to do it in a way that scales, but they can't do one-off deal…”
Prince: Nvidia's power stance mirrors Intel's hubris in 2010
“When we talk with the folks at Nvidia, it feels a little bit like talking to Intel back You know, in our case in 2010, or Apple's case in 2005, where they were like, you're doing it wrong if you care about power efficiency.”
Prince: Smartphones will soon run local models matching current ChatGPT
“It is not particularly long before you're running something which is the equivalent of kind of the current generation of ChatGPT on your iPhone or your, or at your Android device.”
Prince: GPUs will speed-run 30 years of CPU efficiency in ten years
“We're going to speed run the last 30 years of CPU efficiency gains. In, in the next, you know, five to 10 in, in GPUs.”
Prince: Cloudflare will probably enter the agentic infrastructure space
“We will probably compete in that. Agentic infrastructure space.”
Prince: Cloudflare processes roughly 15 trillion requests per day
“I mean, we do something like 15 trillion requests every day”
Prince: Infrastructure regulating web bots will also regulate AI-connected browsers
“I think that that's, it, it'll probably be the same fundamental infrastructure that regulates how bots access the web, that ends up regulating how browsers access the web, and a browser that Takes data and then immediately feeds it back to an LLM might have mo…”