Fifteen Years of Cloudflare, and a Shift in the Internet’s Center of Gravity
Cloudflare launched 15 years ago this week. The company typically marks its birthday by shipping products that give back to the Internet, and it will do plenty of that again this week. This year, though, the founders’ letter pauses to look at what has actually changed on the Internet over that period — and what stubbornly has not.
Some metrics show clear progress. In 2010, less than 10 percent of the Internet was encrypted; today, well over 95 percent is. Cloudflare is quick to credit its own role in driving universal SSL adoption. Other areas have moved far more slowly: IPv6 adoption has grown steadily but painfully slowly, a problem that matters because scarce and expensive IPv4 addresses hold back new entrants and inflate the cost of networking and cloud computing.
The Business Model That Built the Content-Rich Web
One thing has remained remarkably consistent: the Internet’s fundamental business model. The formula has been the same for 15 years — create compelling content, get discovered, and then extract value from the resulting traffic. Whether through ads, subscriptions, merchandise, or sheer ego, that traffic-based reward system filled the Internet with content. The Internet was never free; there was always a transfer of value from consumers to creators, and that transfer is why the web is as vibrant as it is.
That philosophy explains a trivia question: why Cloudflare never built an ad blocker despite countless requests. Ads, for all their flaws, are the only micropayment system that has worked at scale to fund an open Internet while compensating creators. Cloudflare’s mission is building a better Internet, and its conviction is that undermining the existing economic engine is not the way to do it.
Traffic Was Always a Flawed Proxy for Value
The problem is that traffic has never been a clean measure of value. The same reward system that funded the web also incentivized clickbait and ragebait. Entire media companies built businesses on headlines engineered to spike cortisol, because maximum outrage generates maximum traffic. Cloudflare has periodically faced calls to intervene in content moderation at the infrastructure layer, but it has always held the line that editorial decisions do not belong at that level. The real problem, in Cloudflare’s view, is not a moderation gap but a broken incentive structure.
That structure may finally be shifting. A change fundamental to the Internet’s operation has unfolded over the last year, driven by AI, and it has the potential to reshape the web’s underlying economics.
The Discovery Layer is Changing: From Maps to Answers
For 15 years, search engines were the Internet’s discovery system. They scraped content, built an index, and sent users to sites with a treasure map. Creators were fine with being scraped because search engines sent value back in the form of traffic — the Internet’s historic currency. That arrangement is now breaking down. The next discovery system is the answer engine, which does not hand out a map but simply supplies the answer, with no click-through required.
For the vast majority of users, most of the time, that is a better experience. ChatGPT, Anthropic’s Claude, and other AI startups are not search engines; they are answer engines. Even Google now serves AI Overviews in place of ten blue links. The sci-fi future has never hinted otherwise: the helpful robot does not reply with a list of links, it gives an answer. The future, whether anyone likes it or not, is answers, not searches.
Short-Term Pain for Traffic-Dependent Industries
This transition is going to hurt. Media companies are already feeling it, while ecommerce and social apps have not yet seen a dramatic decline — you still need to buy the thing the answer engine recommends, and humans still value talking to other humans. But for news, the answer engine often gives you the whole story in summary; you do not need to read it. The traffic loss to media has already been dramatic. Research shops at investment banks, industry analysts, and consulting firms are all watching people stop landing on their content because they increasingly receive answers instead of search results.
Some argue agents are simply acting on behalf of humans. Cloudflare’s response: so what? An agent that summarizes twenty sources without visiting a single one undermines all twenty. Agents do not click ads. And if agents aggregate information on behalf of many users, subscription revenue also collapses — why subscribe to the Wall Street Journal when your agent can freeload off another subscriber?
Unless content creators are expected to work for free, something has to change. Agent visits are not human visits, and they warrant different rules. An agent’s visit to a media ecosystem that loses its traffic may well kill the business model that built the content-rich web. Cloudflare calls this an existential threat to one of humanity’s most important creations.
A Different Kind of Content Wins
There is, however, reason for optimism. Content is the fuel for every AI system, and the companies running those systems know they must ultimately support the ecosystem financially. Using tools like Cloudflare’s AI Crawl Control, content creators can now wall off their content from AI robots that take it without compensation, and a market is emerging with better deals being struck between AI and content companies.
Who is getting the best deals? Not the ragebait headline writers. Not the political hot-take factories. Not the spammy content farms. The winners are Reddit and the quirky corners of the web that best recall the Internet of old — the Internet of 35 years ago, not 15. The early web had something, much of it has been lost, but the incentives may finally be aligning to bring it back. In an AI-driven Internet — assuming AI companies step up and pay for the content that is truly valuable to them — the creative, local, unique, original content will command the most value. That is precisely the content consumers say they crave and creators say they most want to make.
The Shape of a New Payment Model
Here is how a new model might take shape. For the first time in history, humanity has a solid mathematical representation of its collective knowledge: the sum total of all the large language models. It is not perfect, but it is good. That same mathematical model also maps the gaps in human knowledge — like a block of Swiss cheese, there is plenty of cheese but also plenty of holes.
Imagine an Internet that rewards the creators who fill those holes rather than the ones who generate outrage for traffic. A portion of AI subscription fees, plus a share of the ads those AI companies inevitably serve, could flow back to content creators who most enrich the collective knowledge base. Think of it crudely as a per-monthly-active-user sum from each AI company going into a collective pool, distributed based on who best fills the holes in the cheese.
The mechanisms are plausible. An AI company might tell creators it needs more content on a niche topic its users keep asking about. The very pruning algorithms used today to curate what is valuable already map out the content worth paying for rather than trimming. As GPU costs and talent pools commoditize, the differentiator between AI products will inevitably be access to quality content. The math of the algorithms provides a roadmap to what matters most, and that is the seed of a healthy market.
Cloudflare’s Role: Leveling the Playing Field
Cloudflare sees its role as catalyzing this better model, not defending the status quo — creating a level playing field where many AI companies and many content creators can coexist. The failure mode would be a world where a new AI entrant must pay for content while a legacy search engine gets it for free. The right fix is not that no one pays; it is that everyone who benefits from the ecosystem contributes back in proportion to their size.
That may sound impossible today. Cloudflare’s bet is that once a few significant market participants tip — either voluntarily because it is the right thing or under pressure — the entire market will follow quickly. The ingredients for a healthier bargain are on the table; the question is who blinks first.
An Open Internet Needs Partners
Cloudflare’s mission statement is deliberate: not to “build a better Internet” but to “help build a better Internet.” The distinction matters because the infrastructure required for a healthy market cannot be proprietary or siloed. The solutions have to be open, collaborative, standardized, and shared across many organizations.
That philosophy is driving a series of partnership and collaboration announcements this week. The Internet functions as an ecosystem, and its continued health depends on contributions from a range of actors: infrastructure providers, AI companies, and content creators. Cloudflare is positioning itself to work with any organization ready to take responsibility for that shared environment.
There is a realistic path to what Cloudflare calls a new golden age of the Internet, provided the ecosystem collaborates. Conversations with leading AI companies increasingly include acknowledgment that they owe something back, both to the infrastructure that powers them and to the creators whose work they use. On the other side of that equation, major publishers report that negotiations over content licensing have become far more constructive.
To make sure smaller publishers are not left out of those deals, Cloudflare is announcing new tools this week that give even the smallest operators direct control over who can use the content they produce. The goal is to make content governance accessible, not just for enterprises with legal teams but for independent creators as well.
It might sound ambitious. Cloudflare frames it as an obvious step. Fifteen years in, the company sees the gap between its accomplishments and its mission as still wide—which is why the closing message of its annual letter is less a summary and more a directive.
