The Engine Is Slowing. The Driver Hasn't Noticed.

A hypothesis on what happens when the world's most valuable attention machine optimizes for the wrong variable

There is a pattern in technology history that repeats with such consistency it has almost become boring to cite. A company builds a dominant position. It defends that position with the tools that created it. The world moves in a direction that those tools were not designed for. The company, still holding its tools, watches the gap widen, usually with a press release expressing confidence in the fundamentals.

We have seen this with Nokia, which controlled over 40 percent of the global mobile phone market at its peak and sold its handset division to Microsoft in 2013 after failing to recognize that software had become the product and hardware was merely the vehicle. We have seen it with Yahoo, which at one point was valued at 125 billion dollars, passed on the opportunity to acquire Google for a fraction of that, and was eventually sold to Verizon for less than five billion. We have seen it with Blackberry, which built a loyal enterprise following on the strength of physical keyboards and proprietary messaging, then watched both become irrelevant within a single product cycle after Apple introduced the iPhone.

The specific failure mode varies across these cases. The structural failure is identical. Each of these companies defined its competitive advantage in terms of what it currently did well, not in terms of what users would eventually require. Each optimized its position until the position itself became the problem.

I am going to argue that Google is now at the beginning of that same arc.

The Revenue Architecture That Creates the Problem

To understand why Google's current situation is structurally precarious, it helps to understand what Google actually is as a business.

Google is widely described as a technology company. It is more precisely an advertising distribution system that uses a search engine as its primary demand generation mechanism. According to Alphabet's fiscal year 2024 financial data, Google Search and related properties generated approximately 198 billion dollars in revenue, representing 56.6 percent of Alphabet's total revenue. When YouTube advertising and Google Network revenue are included, total advertising contributes roughly 76 percent of the company's entire top line.

The search engine is not a product Google sells. It is the mechanism through which Google sells attention. Every query entered into Google is a signal of intent that Google converts into an advertising opportunity. The business only functions at its current scale if people continue to use Google to ask questions.

For approximately two decades, that assumption was as close to a given as any assumption in global commerce. Google held more than 90 percent of global search market share for most of its commercial history. Entire industries, including the one YPYM operates in, were built on the premise that Google's position was structurally permanent.

That premise is now being actively challenged by data, not by prediction.

What the Numbers Actually Show

According to research published by Graphite.io in March 2026, AI assistants collectively generate sessions equivalent to 56 percent of global search engine volume. After correcting for task-execution and creative sessions that do not directly compete with search behavior, the true search-competing footprint of AI platforms is approximately 28 percent of global search volume globally, and roughly 17 percent in the United States specifically.

The methodology behind this figure is more rigorous than most comparisons circulating in marketing discussions. Previous estimates of AI versus Google traffic were based primarily on desktop web sessions, which capture only a fraction of actual AI usage given that approximately 83 percent of AI interaction happens through mobile applications. Correcting for this produces numbers that are four to five times larger than most published estimates suggested.

Google's share of the combined information discovery landscape, meaning search plus AI platforms, fell from 89.3 percent to 57.6 percent between December 2022 and December 2025. That is not a projection. It has already happened. According to data from Datos and SparkToro, Google searches per user in the United States declined by approximately 20 percent year over year in Q4 2025. And according to research from Demandbase, 25 percent of B2B buyers now use generative AI for vendor research, typically before a shortlist is formed.

ChatGPT alone now holds approximately 20 percent of total global search-related traffic, making it the second-largest information discovery platform in the world by session volume. Grok grew from 1.2 million monthly sessions in January 2025 to 1.4 billion by December 2025, a 116,000 percent increase in twelve months.

The scale and speed of these shifts have no precedent in the history of search.

The Optimization Problem

Here is where the structural argument becomes interesting, and where I want to be precise about what I am and am not claiming.

Google is not ignorant of this threat. The company has invested heavily in AI, has integrated AI Overviews into its search results, and has launched Gemini as a competitive AI product. Google is aware that the landscape is shifting. The question is not whether Google understands the threat. The question is whether Google can respond to it without dismantling the business model that generates 76 percent of its revenue.

This is the trap that Nokia, Yahoo, and Blackberry each fell into. Nokia understood that smartphones were the future. Its internal research had identified the trend early. The company's inability to respond was not primarily a failure of intelligence. It was a failure of organizational will, specifically, the difficulty of cannibalizing a profitable existing business to invest in a future that would disrupt it. Nokia's hardware expertise, its manufacturing infrastructure, its partner relationships, all of these were assets that were simultaneously liabilities in a world where software had become the differentiator.

Google's equivalent constraint is its advertising model.

An AI assistant that provides a direct, accurate answer to a user's question does not, by design, generate the same advertising opportunity that a list of ten links does. When a user asks an AI what the best option is for a given need and the AI synthesizes an answer, there is no obvious slot for a sponsored result. There is no click-through to an advertiser's landing page. There is no keyword auction. The interaction that makes AI useful is structurally at odds with the interaction that makes Google profitable.

Google can build better AI. What it cannot easily do is replace the advertising mechanism that depends on users clicking through to external destinations with a mechanism that answers the question before the click happens. Every improvement to its AI product potentially erodes the behavior that funds the entire enterprise.

The Process Fixation That Compounds the Problem

There is a second layer to this argument, and it connects directly to why YPYM exists.

In the absence of a clear answer to the structural revenue problem, Google has doubled down on what it can measure and enforce: the characteristics of content that appears in its index. The practical expression of this has been an increasing focus on how content is produced, specifically, the detection and de-prioritization of content that appears to have been generated with AI assistance.

The logic is understandable but, I would argue, strategically misplaced. The concern is that AI-generated content at scale will flood the index with low-quality information. This is a legitimate concern. The response, however, conflates the process with the output. A page that contains accurate, well-structured, technically verified information about a niche industrial product, generated with AI assistance by a subject matter expert who knows the domain deeply, is categorically more valuable to a user than a page written entirely by a human who has no expertise in the subject and no interest in accuracy.

Google's framework for evaluating content quality (reference in Bahasa Indonesia), specifically E-E-A-T, which addresses Experience, Expertise, Authoritativeness, and Trustworthiness, already provides the correct evaluative framework. Accuracy, depth, and demonstrated expertise are the relevant variables. The production method is not. A page is not valuable because a human spent hours writing it. It is valuable because the information it contains is accurate, complete, and useful to the person asking the question.

The obsession with process over outcome is a distraction from the actual competitive threat. While Google spends resources detecting whether content was written with AI assistance, its competitors are training their systems to evaluate whether information is correct. The evaluation that matters in the new discovery landscape is not "was this written by a human?" It is "is this true, and does it answer what was asked?"

That is a substantially higher standard, and one that does not care about production method.

The Countdown and What Comes After

I want to be careful here about the scope of this argument. Google is not going to disappear. The company operates an ecosystem of extraordinary complexity: cloud infrastructure, productivity tools, hardware, mapping systems, enterprise software, and advertising technology that is deeply embedded in the operations of millions of businesses globally. The collapse of its search dominance would be a significant disruption to one part of a much larger organization.

But the part that is being disrupted is the one that generates three-quarters of the revenue.

When Nokia lost its position in mobile phones, it had other businesses to fall back on. Nokia today is a significant player in telecommunications infrastructure. It survived, profoundly changed, by pivoting to an area where its capabilities were still relevant. The question for Google is what the equivalent pivot looks like when the business that needs to be preserved is not a product line but the primary monetization mechanism of the entire company.

Yahoo had the same structure problem. Its advertising revenue was tied to search dominance. When search dominance eroded, the advertising business eroded with it. There was no hardware division, no infrastructure business, no adjacent product that could absorb the loss. The company that had been valued at 125 billion dollars was sold for less than four percent of that.

Google is a more complex organization than Yahoo and a more adaptable one. But the structural dependency is comparable. If the behavior that generates advertising revenue, specifically, users entering queries into a search interface and clicking on results, continues to shift toward AI-based discovery, Google faces a revenue problem that no amount of AI product investment can fully solve, because the AI product and the advertising business are in fundamental tension with each other.

The countdown on that tension has already started. The data shows it has been running since late 2022.

What This Means for Anyone Building on the Internet Today

This is not an argument for ignoring search engine optimization. Google still delivers the majority of organic discovery globally, and that will remain true for the foreseeable future. The argument is about direction and preparation.

The businesses that will navigate this transition well are the ones that are building page quality for the standard that the new discovery landscape demands, not the standard that was sufficient three years ago. That standard is high-fidelity information: accurate, deep, technically verified, structured for both human readers and AI interpretation, and fast enough that the technical overhead of accessing it is invisible.

This is what we describe at YPYM as the real meaning of "Your Page, Your Money." A page that is built to that standard performs in Google search, performs in AI discovery, performs in direct conversion, and compounds in value over time. A page that was built to satisfy a keyword density metric from 2018 is already failing, and will fail faster as the discovery landscape continues to shift.

The transition that is underway is not from search to AI. It is from low-fidelity information retrieval to high-fidelity information retrieval. Every platform involved in that transition, including the ones that will eventually replace current category leaders, will reward the same underlying quality. The investment thesis does not change. The delivery mechanism does.

Build the page correctly. The platforms will follow.

— A personal reflection from the founder of YPYM. Written with AI assistance (26-44%), because why wouldn't it be?

References:

1. Graphite.io / Ethan Smith, AI is Much Bigger Than You Think, March 2026.
2. Harvard/NBER Working Paper 34255, ChatGPT and search behavior data, July 2025.
3. Alphabet Inc. Form 10-K / Bullfincher segment analysis, Fiscal Year 2024: Google Search revenue 198.08B USD, representing 56.63% of total Alphabet revenue.
4. Datos / SparkToro, Q4 2025 State of Search: Google searches per US user, year-over-year change.