The Front Door Has Moved
For over two decades, the internet ran on a remarkably stable loop.
People had questions. Search engines pointed them somewhere. Publishers built content to capture that traffic. Advertisers paid to sit alongside it. The money flowed back into more content, and the cycle repeated.
It wasn’t elegant, but it worked. Search became the de facto front door to the internet. Publishers learned to optimize for it. Digital advertising grew into one of the most valuable industries on the planet – all because users had to travel through certain chokepoints to get what they wanted.
That journey is getting a lot shorter.
More and more, people skip the search results entirely and just ask an AI. They get a synthesized answer in seconds, no clicking required. It feels like a minor convenience upgrade. But underneath that convenience, the economic logic that powered the web for two decades is quietly unraveling.
The old model was built on navigation. AI is eliminating the need to navigate.
Why This Breaks Advertising
Digital advertising has always been an interception game.
A user wanted something. Advertisers placed themselves between the user and that thing – in search results, on publisher pages, before videos. The ad value was tied directly to the journey. You couldn’t skip the tollbooth.
AI assistants remove the tollbooth entirely.
When someone asks an AI a question and gets a direct answer, there’s no search results page, no publisher website, no natural slot for a sponsored message. The interception point that advertising was built around simply doesn’t exist in that interaction.
We’re already seeing the downstream effects. A significant share of searches now end without any click to an external site. AI summaries handle the question on the spot. Even when sources are cited, most users don’t follow through. Why would they? They already have what they came for.
For publishers, fewer clicks means fewer impressions, which means less revenue. For advertisers, the question becomes more fundamental: how do you reach an audience that’s no longer traveling through the spaces you used to own?
The attention hasn’t gone anywhere. People are still curious, still engaged, still consuming information constantly. But the pipes that carried that attention – and monetized it – are corroding.
The Content Crisis Nobody’s Talking About Enough
Here’s the tension that I think deserves more serious discussion than it typically gets.
AI systems are built on human knowledge. Not metaphorically – literally. The articles, research papers, forum threads, technical documentation, and professional analysis produced over decades form the substrate these models learned from. Without that material, there’s no intelligence to speak of.
And yet the same AI systems are steadily dismantling the financial model that made producing that material worthwhile.
Traditional search had an imperfect but real bargain built into it: index the content, send traffic back to the creator, let them monetize that traffic. Publishers grumbled about Google’s power, but at least the referrals showed up.
AI breaks that bargain. The better these systems get at synthesizing content, the less reason users have to visit the source. Creators see their traffic decline exactly as demand for their knowledge reaches an all-time high. That’s a genuinely strange situation.
And it compounds over time. If original, high-quality content stops being economically viable to produce, the supply of it shrinks. Which means the quality of future AI outputs gets harder to sustain. The industry is, in a slow and indirect way, eating the foundation it depends on.
Expertise Is the New Scarce Resource
For most of the internet era, distribution was the hard part. Creating content was accessible; reaching an audience wasn’t. That’s why platforms became so powerful – they sat between creators and users and extracted value from that position.
AI flips that dynamic.
Distribution is rapidly becoming a commodity. Any decent model can summarize, translate, reformat, and produce content at scale. That capability isn’t a moat anymore – it’s table stakes.
What AI can’t manufacture is genuine expertise. Firsthand experience. Original reporting. The specific insight that comes from someone who’s actually done the thing they’re writing about. Independent analysis grounded in real domain knowledge.
These things don’t emerge from a language model. They come from people and organizations that invest real time and resources into knowing something deeply.
As AI-generated content floods the web, that kind of original expertise becomes rarer and more valuable. The internet spent years rewarding scale and volume. The next phase may reward something harder to fake.
A New Economic Layer Is Taking Shape
The industry isn’t standing still on this.
Licensing deals between AI companies and major publishers are becoming more common. Legal frameworks around training data are being contested across multiple jurisdictions. These negotiations are messy and often contentious, but they signal that the old “free extraction” model is facing real pushback.
At the infrastructure level, experiments like Cloudflare’s Pay-Per-Crawl initiative point toward something genuinely new: a direct commercial relationship between AI systems and the content they depend on. The premise is simple – if a crawler is extracting value from your content, it should pay for access, just like any other commercial transaction.
Whether that specific model scales is an open question. But the underlying logic matters. For the first time, we’re seeing serious attempts to build mechanisms that connect content creation to economic compensation in the AI era – not through the indirect, ad-supported path of the past, but as a direct value exchange.
That’s a different kind of internet than the one we’ve operated in for the last twenty years.
What Actually Comes Next
Advertising won’t die. It never does – it adapts.
Display gave way to search. Search expanded into social. Social spawned influencer economies and creator platforms. Each transition felt disruptive from the inside, and each one eventually produced new business models that captured the shifted attention.
AI will do the same thing, just faster and with more structural disruption along the way.
The models that emerge will probably look different from what we’re used to. Native AI advertising, direct brand relationships with credible experts, premium subscription tiers for high-quality information, licensing arrangements between platforms and creators – some mix of these will likely replace the impression-based economy we’ve relied on.
What won’t survive is the strategy of aggregating attention without creating genuine value. The businesses built purely on optimizing for traffic – capturing clicks without contributing knowledge – are the ones most exposed to what’s coming.
For content creators, this is a complicated moment. The traffic-based model they’ve relied on is under real pressure. But new compensation paths are emerging, and expertise has rarely been more differentiated from the noise around it.
For AI companies, the stakes are just as high, even if it’s less obvious. Their long-term product quality depends on a healthy ecosystem of original human knowledge continuing to exist. That’s not a given.
The internet’s advertising era isn’t ending this year. But the assumptions underneath it – that attention flows through predictable channels, that content gets made because traffic pays for it, that distribution is where value lives – those assumptions are already cracking.
The next chapter probably belongs to whoever creates knowledge worth paying for directly.
That’s a harder thing to build. But it might also be a more honest internet than the one we’ve had.
