The Internet Is Splitting in Two. Which Side You're On Will Define the Next Decade of Your Business.

The internet is no longer one thing. It is splitting into two distinct layers: an information layer that is collapsing and a transaction layer that is growing. Businesses built on the information layer are losing 30% to 97% of their traffic. Businesses built on the transaction layer are seeing AI-driven traffic convert at four times the normal rate. Which layer your business sits on will determine whether AI is the best or worst thing that ever happened to you.

This isn't speculation. The data is already in.

What Is Actually Happening to Internet Traffic?

The headline numbers look fine. Global internet traffic grew 17% in 2025, reaching over 520 exabytes per month, according to Cloudflare's Year in Review. Underneath that aggregate number, the composition of that traffic is changing in ways that are breaking the business models the internet was built on.

Three shifts are happening at the same time.

First, AI-driven traffic is exploding. HUMAN Security's 2025 AI Traffic Benchmark found that AI-generated traffic grew 187% year-over-year. Monthly volumes tripled. And here's the part nobody expected: autonomous systems traffic (AI agents that don't just read the web, they transact on it) grew 7,851% year-over-year. AI isn't just consuming content anymore. It's buying things.

Second, zero-click searches are now the default behavior. According to Semrush's 2025 data, 58.5% of all U.S. Google searches now end without a single click. On mobile, that number is 77%. When Google's AI Overviews appear on a search result, the zero-click rate jumps to 83%. Eight out of ten people who see an AI-generated answer never visit a website. They get what they need and move on.

Third, publisher traffic is in freefall. This isn't a gentle decline. Digital Trends went from 8.5 million monthly visitors to 264,000, a 97% drop. ZDNet lost 90%. The Verge lost 85%. Business Insider lost 55% of organic search traffic and cut 21% of its staff. HuffPost lost half its search referrals. Ten major tech publications went from a combined 112 million monthly visits to under 50 million. None of them got penalized. None of them did anything wrong. AI simply started answering the questions their content was built to answer.

Why This Is Happening: The Inverted Incentive

For over 500 years, since the invention of the printing press, creating better information meant reaching more people. Write something useful, accurate, and comprehensive, and you'd be rewarded with attention, readers, and revenue. That relationship has gone the other direction for the first time in human history.

It's actually very simple. AI reads your content, absorbs it, and uses it to answer someone's question directly in an AI search result. The person gets the answer. They never click. You get nothing. No visit. No ad revenue. No sale.

The better your content is (detailed, specific, exactly what the person was looking for) the more likely AI is to use it as the definitive answer. Mediocre content that doesn't quite answer the question? AI doesn't even bother citing it. Great content? AI absorbs it, delivers the answer, and the person is done.

Quality content is now punished more than mediocre content, because AI is more likely to use it as the definitive answer. The best content creators are feeding the machine that's replacing them.

This is why publications that "did everything right" are the ones losing the most traffic. They were too good. AI absorbed their work and cut them out of the equation.

The Two Layers: Information vs. Transaction

The internet isn't dying. It's splitting into two layers with very different futures.

The Information Layer (Collapsing)

The information layer is websites that exist to inform: news publications, how-to guides, review sites, educational content, reference material. These sites make money through advertising. People visit, see ads, and the site earns revenue from that attention.

AI is systematically dismantling this layer. When someone asks "how do I fix a leaky faucet" or "what are the best headphones under $200," AI now answers directly. The information sites that used to answer those questions are being absorbed into AI's training data and then bypassed.

The numbers tell the story. Google's AI Overviews now appear on 13-14% of all search queries, and that number doubled from 6.49% in January 2025 to over 13% by year-end. For queries where AI Overviews appear, click-through rates drop from 15% to 8%. The expansion is accelerating.

For businesses built on the information layer (media companies, content publishers, ad-supported blogs, affiliate review sites) this is an existential threat. Their business model depends on people clicking through to read content. AI is making that click unnecessary.

The Transaction Layer (Growing)

The transaction layer is websites that exist to sell: ecommerce stores, SaaS platforms, service businesses, marketplaces. These sites don't make money through ad impressions. They make money when someone buys something.

AI cannot replace that transaction. It can summarize a product review, but it can't ship you the product. It can recommend a piece of furniture, but you still need to visit a store to buy it. The transaction layer is protected because AI handles the discovery and still needs the merchant to fulfill the purchase.

The data on this is striking. AI traffic to retail sites grew 4,700% year-over-year. Shoppers who use AI tools to find products convert at 12.3% compared to 3.1% for non-AI-assisted shoppers, a 4x conversion rate lift. Cart-to-checkout rates jump from 26.3% to 49.3% when AI shopping assistants are involved.

This makes intuitive sense. Someone who asks ChatGPT "what's the best teak outdoor dining set under $3,000" and gets a specific recommendation is a pre-qualified buyer. They've already done their research through the AI. By the time they arrive at your store, they're ready to buy. That's why the conversion rate is four times higher.

Only 2% of ecommerce product searches triggered AI Overviews as of late 2025. As of early 2026, that number has risen to about 14%. Growing fast, but still a fraction of informational queries. AI is absolutely crushing informational websites. Product searches? It barely touches them. Because AI can answer a question, it cannot buy the product for you.

What Actually Survives the AI Information Collapse

Not everything on the information layer is doomed. Five categories of websites have structural protection against AI displacement.

Transactional sites survive because AI needs somewhere to send the buyer. You can't complete a purchase inside a chatbot (yet). Ecommerce stores, booking platforms, and SaaS products are the destination, not the source.

Social platforms survive because they generate real-time, human-generated conversation that AI can't replicate at the same pace. Reddit, YouTube comments, and niche forums produce the kind of raw, opinionated, experience-based content that AI models actually depend on for training data.

Academic and research institutions survive because they produce original, peer-reviewed data. When Nature publishes a study on model collapse, that's a primary source. AI cites it. It doesn't replace it.

Consulting and service businesses survive because their content is a demonstration of expertise, not the product itself. A digital agency publishing case studies and methodology breakdowns uses content to earn trust and generate leads, not to make money from page views.

Government and institutional sites survive because they are authoritative sources of record. Tax codes, regulatory filings, census data. AI has to cite these because there is no alternative source.

What do all five categories have in common - None of them depend on advertising revenue. Their business models don't break when AI eliminates the click.

Why Shopify Store Owners Are on the Right Side of This Split

If you run a Shopify store, you are on the transaction layer by default. Your business model is built on selling products, not selling attention. AI can't break you the way it can break a publisher, because at the end of the chain, someone buys the product from your store.

Being on the transaction layer isn't enough. You also need to be visible to AI.

When someone asks ChatGPT "what's the best handmade ceramic mug for pour-over coffee," the AI recommends specific products and links to specific stores. If your store isn't in that answer, your competitor's is.

This is where the opportunity is widest right now. AI shopping is growing explosively (traffic to retail sites up 4,700% year-over-year) and most Shopify stores haven't done anything about it. They're still focused on Google search, where the rules are changing underneath them. The stores that position themselves for AI discovery now will capture a disproportionate share of this new, high-converting traffic channel.

One stat should change how you think about this: 80% of products shown in Google's AI Overviews don't rank in the traditional top 10 search results. AI visibility is a different game than SEO. Stores that never ranked on page one of Google are showing up in AI recommendations. And stores that dominated Google's first page are being bypassed entirely.

The window for this is open right now. The early movers in SEO captured advantages that compounded for a decade. The same dynamic is forming in AI shopping. The stores that figure this out in 2026 will own the channel.

The Model Collapse Problem: Why This Gets Worse Before It Gets Better

Model collapse, documented in a landmark 2024 Nature paper, occurs when AI models are trained on data that was itself generated by AI models. Each successive generation of training produces models with less diversity, less accuracy, and less connection to reality. The tail of the distribution (the specific, minority-perspective content) disappears first. Eventually, the models converge on bland, homogenized outputs that don't reflect the real world.

This matters for the internet because AI-generated content is now flooding the web. As more websites publish AI-written articles, and as AI companies train on that content, the quality of AI outputs degrades. The snake eats its own tail.

The implications are the stuff we all see in the doomsday movies.

First, the internet gets dumber. As original human-created content disappears behind paywalls or stops being created (because there's no economic incentive to create it), AI training data gets progressively worse. An estimated 60% of internet traffic is already bot-generated or AI-generated, and that proportion keeps growing. The share of genuine human insight shrinks.

Second, original human content becomes more important, not less. If AI-generated content is converging toward mediocrity, the content that stands out (the content AI actually wants to cite) is content with original data, genuine expertise, and real-world experience. The very thing AI is destroying (quality content) is the thing it needs most to maintain its own quality.

Third, the brands that create original content now will have compounding advantages. AI systems preferentially cite sources with first-party data, practitioner expertise, and specific real-world examples. A Shopify store that publishes original content about its product category (with real customer data, real use cases, real results) builds the kind of content moat that AI can't replicate from training data alone.

What This Means for Your Business in 2026

The internet is not dying. The internet as we've known it, where you create content, people find it through search, and you make money from that attention, is over for most businesses in the information layer.

If your business depends on people clicking through to your website to see ads, you are on the wrong side of this split. The information layer is collapsing, and no amount of SEO will reverse a structural shift in how people access information.

If your business depends on people buying your products, you are on the right side. You need to make sure that when AI recommends products in your category, your store is in that recommendation. The transaction layer is growing, and the businesses that show up in AI-powered discovery will get the traffic.

The businesses that win the next decade will understand three things:

The information layer is being absorbed by AI. Content that exists solely to inform will increasingly be summarized and delivered by AI without any traffic reaching the original source.

The transaction layer is being amplified by AI. Ecommerce businesses that are visible to AI will receive higher-quality, higher-converting traffic than any channel has ever delivered.

Original, expert content is the bridge between the two. The Shopify stores that publish genuinely useful, experience-based content about their product categories will be cited by AI as authorities, driving both brand awareness and qualified buyers to their stores.

The internet is splitting in two. The question isn't whether you'll be affected. It's which side you're on, and whether you're ready.

Over the next few weeks, I'll be walking through every layer of this shift: the model collapse problem in detail, what specific publishers are experiencing, why AI licensing deals won't save anyone, and exactly how Shopify stores can position themselves to win. If you want to follow along, subscribe to the weekly email where I break this down one layer at a time.

Steve Merrill is the founder of WRKNG Digital, a digital agency that helps Shopify stores get their products recommended by AI shopping tools including ChatGPT, Google AI, Perplexity, and Copilot.