Chapter 087　What Did Google Redefine?

What did Google redefine? This is the hardest question in the series. The reason is simple. This company developed, and then published, a technology capable of destroying its own largest source of revenue. Search advertising as a structure. Conversational AI as a technology. Both grew up inside the same company at the same time. What we take up here is not a story of success. It is a tension still in progress. We neither celebrate nor condemn. We separate what public information confirms from what it does not.

1 Why this company is worth the question

Start with the numbers. The numbers are not the conclusion. Alphabet’s total revenue for the fiscal year ended December 2025 was $402.8 billion. Google Search and other accounted for $245.6 billion of that. YouTube ads came to $45.0 billion, Google Network to $31.3 billion. Advertising in total was $321.9 billion. Google Cloud revenue was $68.8 billion, with operating income of $18.3 billion. Total operating income was $129.0 billion, and capital expenditure $91.4 billion (earnings release of February 4, 2026). Just over sixty percent of revenue comes from a single business. One entry point: search. Advertising as a whole approaches eighty percent. Few enterprises of this size rest on one point so heavily. In the following quarter, ended June 30, 2026, revenue was $119.8 billion, up 24% year over year. Search and other reached $63.3 billion, up 17%. Google Cloud reached $24.8 billion, up 82%, with operating income of $8.8 billion (earnings release of July 22, 2026). Search is not shrinking. It is growing. That is precisely why the company is worth the question. There are three reasons. First, it produced the technology of its own destruction. The paper setting out the mechanism called the Transformer was posted to arXiv in June 2017. The same year, an explanation appeared on the Google Research blog. Most of today’s conversational AI descends from that structure. The company opened to the outside world a technology capable of shaking its own revenue base. Second, value creation and value capture diverged for a long time. The company that produced the technology and the company that first won a large consumer market with it were not the same. That divergence concerns every enterprise that invests in basic research over long horizons. Third, law and regulation are now shaping the form of the business. In the United States, Europe, and Japan, this company’s distribution structure has become the object of judicial and administrative decisions. When managerial freedom is being narrowed from outside, how far can redefinition go? What we want to read here is not a list of reasons for winning. It is the structure of a tension.

2 The conventional answers and their limits

Three explanations of this company circulate today. Each is partly right. Each drops the part that matters. The first conventional answer: “It is an advertising company protected by a search monopoly” As a description of the revenue structure, this is accurate. Advertising revenue for the fiscal year ended December 2025 was $321.9 billion, approaching eighty percent of total revenue. The concentration is a fact. Judicial decisions have followed. Judge Mehta of the US District Court for the District of Columbia issued a remedies ruling on September 2, 2025. According to public reporting, exclusive default search agreements were prohibited and the sharing of certain data with competitors was ordered. Divestiture of Chrome or Android was not ordered. Final judgment was entered on December 5, 2025. Google appealed in January 2026, and the Department of Justice and several states cross-appealed in February 2026. As of August 1, 2026, the appeal is pending. One caution here. Monopoly is a description of a state, not an explanation. Why the state arose, and why it persists, are separate questions. An argument that ends at monopoly misses what the company is doing now. The second conventional answer: “It fell behind on AI and is playing catch-up” That the company was not the first mover in the consumer market for conversational AI is a fact. But the phrase “fell behind” confuses a technology question with a product question. There is an in-house processor design called the TPU. On the February 2026 earnings call the company’s chief executive said it had been under development for ten years. The Transformer paper dates from 2017. On the accumulation of technology, “behind” is a hard reading to sustain. What came late was not the technology. It was the order in which the technology was applied to the company’s own core business. That is where this chapter focuses. For an enterprise with an existing revenue source, choosing where to apply a new technology is not a technical problem. It is a management problem. The third conventional answer: “Conversational AI kills search” This is the most widely repeated prophecy. If the answer comes back directly, there is no need to follow a link. Therefore search advertising shrinks. That is the logic. Public information as of August 1, 2026 does not yet bear the prophecy out. On the July 2026 earnings call the company said monthly active users of AI Mode had passed one billion. It went on to say this was driving an increase in total search queries. Search revenue in the same quarter was up 17%. But this is the company’s own account. It cannot be verified independently from outside. We treat it as a claim, not as a fact. The prophecy has not been refuted. The question is simply not yet settled. What all three miss All three ask what kind of company this is by asking what it sells. But the second dimension of Enterprise Redefinition does not ask what a firm sells. It asks what value the firm delivers. Restate the question and the picture changes. What this company has delivered is not a list of links. It is a reduction in the time it takes to reach the information you want. On that reading, conversational AI is not the enemy. It is a way of delivering the same value over a shorter path. The problem is that the shorter path does not mesh with the existing unit of revenue.

3 What was redefined — an analysis across the five

dimensions We read the case along the five dimensions of Enterprise Redefinition. The order follows the canon: Purpose, Business, Organization, Capital, Leadership. What follows is what public information supports. It does not assert anything about internal decisions. Purpose — from organizing to generating The purpose the company has stated is to organize the world’s information and make it universally accessible and useful. That wording can still be found on its official site. What matters is that the Core Purpose has not been swapped out. A Core Purpose can remain stable while its expression and its means of realization evolve. The word organize presupposes information that already exists. Generation produces text that did not exist. The means changed. The core of the purpose reads as unchanged. The two nonetheless pull against each other. Organizing sends people to sources. Generating answers in place of sources. On the July 2026 earnings call the company said its AI features in Search were sending billions of clicks to the web every week. That such a statement becomes necessary is itself evidence of the tension. Business — from the business of the entry point to the business of the substrate The rewriting of the business dimension is running in two directions at once. One is the form of search itself. On November 18, 2025 the company announced Gemini 3 and, the same day, put it into AI Mode in Search. The frontier model went into the largest revenue source on day one. Search was treated not as something to be carefully protected but as a place to experiment. The other is the multiplication of revenue lines. Google Cloud’s backlog reached $514.0 billion at the end of June 2026. The company said it expects to recognize somewhat more than half of that as revenue over the following twenty-four months. Backlog at the end of 2025 stood at $240 billion, so it more than doubled in six months. Put the two together and the shape of the business redefinition appears. Search advertising earned from the brokerage of attention. Cloud and AI earn from the supply of computation. If the second grows while the first shrinks, the enterprise still stands. The company placed the technology capable of destroying it at the center of a second revenue structure. This is not business expansion. It is the doubling of the unit in which value is measured. Organization — the distance between research and product Two features of the organization dimension are visible in public information. One is that the output of basic research has been opened to the outside. The Transformer paper was published both at a peer-reviewed conference and as a preprint. A corporate research organization published a result capable of becoming its own competitive advantage. That behavior conforms to the norms of the academic community. From the standpoint of value capture, it requires explanation. The other is that the computing substrate is being supplied externally. On April 6, 2026, Anthropic announced an agreement with Google and Broadcom for multiple gigawatts of TPU capacity, to be taken up progressively from 2027. The company is selling compute to a firm that competes with its own conversational AI. The organizational boundary has widened to include collaboration with competitors. As the canon defines it, an organization is not a set of people. It is a value-creation system made of people, AI, partners, universities, and customers. This company’s organization approaches that definition. But a structure in which the competitor is simultaneously the customer carries a permanent internal tension. Capital — capital gets heavier The capital dimension is the one moving most visibly in this case. Capital expenditure was $91.4 billion in the fiscal year ended December 2025. In February 2026 the company guided full-year 2026 capital expenditure to $175–185 billion. On the July 22, 2026 earnings call it raised that guidance to $195–205 billion. More than a doubling in two years. The shape of cash changed as a result. In the quarter ended June 30, 2026, operating cash flow was $39.1 billion and capital expenditure $44.9 billion. Free cash flow was an outflow of $5.9 billion. Net income of $112.2 billion in the same quarter was up 298% year over year, but it included $99.0 billion of gains on equity securities. Accounting profit and cash on hand are different things. The canon’s third equation defines capital as follows. Future Capital = Financial × Human × Learning × Trust × AI × Knowledge × Ecosystem × Purpose This is multiplication. If any single term is zero, the whole product is zero. In this company’s case the Knowledge term compounded over a long period. The Transformer belongs to it, and so do the ten years of the TPU. What the company is doing now is converting that accumulation, in bulk, into the Financial and AI terms. Leadership — the choice not to protect The rewriting in the leadership dimension shows up in sequencing. An enterprise with a revenue source to protect tries a new technology in peripheral businesses first. Application to the core comes last. This company did not do that. It put the frontier model into its largest revenue source on the day of announcement. Whether the judgment was right is not yet settled. The character of the judgment is clear. The canon’s fourth equation defines leadership as follows. Leadership = Purpose × Question Design × Capital Allocation × System Architecture × Trust This too is multiplication; a zero anywhere zeroes the product. What stands out in this company’s leadership are the Capital Allocation and Question Design terms. The question asked was not “what does it take to protect search” but “when the route to information changes, what should we be?” As the sixth of the First Principles states, Enterprise Exists to Redefine Itself. But leadership that declines to protect is extremely exposed on transitional revenue. The speed at which the old unit shrinks, and the speed at which the new unit grows. If that gap opens, managerial freedom disappears quickly.

4 Structure — maturity and the value chain

4.1 Where it sits on the Enterprise Redefinition Maturity Model

Within what public information supports, maturity varies by dimension. No firm placement is possible. In the business dimension the model of the business is evolving continuously. The company rewrote the form of search itself and, at the same time, stood up a separate revenue structure in the supply of computation. That behavior resembles what is seen in a Continuous Redefinition Enterprise. It is redesigning itself before external disruption requires it. In the capital dimension the direction of allocation also points toward Future Value. The sharp expansion of capital expenditure is an allocation toward future demand rather than past success. Whether the scale is appropriate cannot be judged from outside. Only the direction can. The leadership dimension is harder to assess. A Future Value Enterprise actively shapes future industries rather than reacting to external change. That the company was not the first mover in the consumer market for conversational AI makes that level impossible to assert. The organization dimension is not adequately observable from outside. Recall the canon’s notes here. The Enterprise Redefinition Maturity Model (ERMM) evaluates organizational coherence rather than isolated excellence. An organization may hold Level 4 AI capability while remaining Level 2 in leadership. The reverse is equally possible. Placing this company at any single level is itself a misuse of the model. Maturity is also assessed across all five dimensions in balance. Exceptional technological capability with weak leadership redesign does not produce higher maturity, and strong purpose without adaptive organizational systems remains insufficient. And Level 5 must not be treated as a target to reach as fast as possible. The appropriate level differs by industry and by environment.

4.2 Where in the Future Value Chain the value arose

The canon fixes the causal order as follows. Purpose → Learning → Redefinition → Creation → Enterprise Value Lay the company over that order and the singularity of this case appears. Learning was extremely high. The Transformer and the TPU are both products of Learning. But there was a period in which Redefinition lagged Learning. The technology was inside the company. Time passed before the company used it to redefine itself. In that gap, others advanced Creation. Value creation and value capture separated. The cause was not a shortage of Learning. It was a delay in the connection between Learning and Redefinition. The second equation shows the same structure. FVCC = Purpose × Learning × Redefinition × AI Integration × Ecosystem × Capital Allocation × Trust Seven terms, multiplied. However large Learning grows, a small Redefinition makes the product small. Measuring Future Value by the amount spent on research and development is incompatible with this equation. And Enterprise Value comes last. The order must not be read out of sequence.

4.3 Time as a term

The fifth equation places time as an independent term. Future Value = Future Time × Future Capability The company built Future Capability over more than a decade. The question is whether the Future Time in which to use that capability remains. Capability without time drives the product toward zero, because this is multiplication. The Value Equation says the same thing. Value = Purpose × Trust × Capability × Time Multiplication again. If any one of the four collapses, no value remains. Here the Trust term bites. The role of sending people to sources, and the role of answering in place of sources. How to hold trust between those two is not a technical problem but a design problem. However good the answers, if the ecosystem of sources thins out, the material for answers disappears. As the eighth of the First Principles states, Trust Compounds Faster Than Capital. Trust compounds faster than capital. It is also lost faster.

5 What it looks like in practice — the tension of selfdisruption

Redefinition is not the act of adding something. It is the act of deciding what to keep and what to let go. We set out the company’s turning points together with what was given up. Turning point 1 — 2017, publishing the Transformer A paper setting out a new sequence-transduction architecture was published. What was given up was exclusivity over the technology. Publication made the structure a shared asset of researchers and companies worldwide. The classic problem of basic research appears here. The party that created the value and the party that captured it were not the same. Publication is a contribution to the academic community and a device for attracting talent. It is also the surrender of part of a competitive advantage. We call the judgment neither an error nor a success. We confirm only that long-horizon investment carries this form of cost. Turning point 2 — November 2025, same-day deployment into the largest revenue source On the day Gemini 3 was announced, the model entered AI Mode in Search. What was given up was the short-term stability of the existing search experience. On the July 2026 earnings call the company said it was testing several advertising formats inside AI Mode, including formats that place contextually relevant links within the answer. The work of moving the unit of revenue from the click to something else is under way. There is no guarantee it will be completed. Turning point 3 — 2026, supplying compute to a competitor An agreement to supply computing substrate at scale to a firm competing with the company’s own conversational AI was made public. What was given up was the option of reserving the computing substrate for internal use. Enclose it and the advantage is protected. Sell it and the market widens. The company appears to have chosen the second. That choice may narrow the relative advantage of its own models. The conditions under which this structure breaks On that basis, we set out four conditions. Celebration is the enemy of analysis. First, the conversion of the revenue unit may not arrive in time. If an answer-shaped experience does not establish a unit of revenue to replace the click, the contraction of the old unit becomes a contraction of revenue. As of August 1, 2026, there is little to test this against beyond the company’s own account. Second, the weight of capital. Capital expenditure guidance for full-year 2026 is $195–205 billion. Free cash flow in the second quarter of that year was an outflow of $5.9 billion. If demand comes in below expectations, the investment remains as fixed cost. Third, law and regulation. The US search case is pending on appeal. In the advertising-technology litigation, a ruling came on April 17, 2025: the US District Court for the Eastern District of Virginia found unlawful monopoly maintenance. A remedies phase followed. As of August 1, 2026, we have not been able to confirm a final remedies decision in public information. In September 2025 the European Commission imposed a fine of €2.95 billion for selfpreferencing in advertising technology. On April 15, 2025 the Japan Fair Trade Commission issued a cease-and-desist order concerning contracts related to Android. All remain in dispute or in progress, and no outcome can be asserted. Regulation carries a special meaning in this case. The more the distribution of search is constrained, the more important the new entry point of conversational AI becomes. Regulation can push selfdisruption forward. That is a structural observation, not an inference about the company’s intent. Fourth, loss of the entry point. If the form of default search agreements is constrained, another firm may hold the first point of contact with the user. That is a route to losing on distribution rather than on the quality of search. All of these concern the future, and none can be asserted. What we can say is that this enterprise is shaking itself without waiting to be disrupted from outside. The textbook version of the innovator’s dilemma (Christensen, 1997) has the incumbent ignoring the new technology. That is not this case. A company that looked straight at it has entered a difficult transition precisely because it looked.

6 What transfers, and questions for the executive

What transfers from this case, and under what conditions? Not the imitation of a technology strategy. There are four things to take. First, design value creation and value capture separately. Basic research creates value. It does not follow that the creator captures it. Many incumbents hold excellent research results and have been late in the sequence of commercialization. The problem is not the quality of the research. It is the connection between research and redefinition. Second, the order of self-disruption. Do you apply a new technology to the core business, or test it at the periphery? That order is not a technical problem but a management problem. Testing at the periphery looks safe, but it does not keep up with the speed at which the assumptions of the core business change. Either way, you need to know that you are choosing. Third, a second revenue structure. This company placed the technology that threatens it at the center of a different revenue source. When incumbents say they will protect the core business, most are protecting the way the core business is sold. What deserves protection is not the way of selling but the value being delivered. Fourth, treat regulation as a design condition. Antitrust decisions narrow managerial freedom from outside. Read only as a constraint, the response closes into defense. The regulatory environment is also part of the design conditions of the future. And one caution above all. This company must not be imported as a model. The appropriate level on the Enterprise Redefinition Maturity Model differs by industry and by environment. Making the fastest possible arrival at Level 5 the objective is something the source papers explicitly warn against. Three questions to close. Each can be answered at your next executive meeting. Question 1 — What technology could destroy your largest source of revenue? Is it inside the company or outside? If it is inside, who can make the decision to apply it to the core business? If it is outside, what can you do from the moment you notice? Question 2 — Of the value your company has created, what has it failed to capture? Research, standard-setting, developing people. Activities that create value become gifts to others unless capture is designed alongside. The gift is not the problem. The unwitting gift is. Question 3 — When the unit of revenue changes, what will your company earn from? This is the question of Recognize, the first stage of Enterprise Redefinition: what assumptions about our enterprise are becoming obsolete? AI cannot answer it. AI Optimizes. Humans Define. AI optimizes; humans define value, purpose, and direction. What Google redefined is not search. It is the form in which the route to information is delivered. And the company is running that redefinition on top of its largest revenue source. Not testing somewhere safe and then migrating. Testing on the earner. Whether that choice was right is not settled as of August 1, 2026. What we should learn from the case is not a conclusion. It is an order. Purpose, then Learning, then Redefinition, then Creation, and Enterprise Value last. However large the Learning term, without Redefinition behind it the product does not grow. What this enterprise shows is how hard that connection is. And only the companies that attempt the connection get to move on to the next question.

In brief

What Google redefined is not search but the form in which the route to information is delivered.
Within what public information supports, business and capital resemble a Continuous Redefinition Enterprise; leadership cannot be asserted.
The value arose in Learning. The connection to Redefinition came late, and capture was missed.
If the conversion of the revenue unit does not arrive in time, the contraction of the old unit becomes the contraction of revenue.

Key concepts

Future Value Chain / Future Value Creation Capability / Future Capital / the Enterprise Redefinition Maturity Model / the Medium Pattern / the Layer Shift Pattern (→ Vol. VI, Ch. 059)

The chain of ideas

Learning → Redefinition → Capital Allocation → Creation → Enterprise Value

Related first principles

Principle 2 — Future Value Precedes Enterprise Value. Principle 4 — AI Optimizes. Humans Define. Principle 6 — Enterprise Exists to Redefine Itself. Principle 8 — Trust Compounds Faster Than Capital.

Related chapters

Vol. VI, Ch. 059 “Cases of Enterprise Redefinition” — the definitions of the Medium Pattern and the Layer Shift Pattern are there
Vol. VII, Ch. 069 “Does Innovation Become Enterprise Value?” — treats the separation of value creation from value capture
Vol. VI, Ch. 051 “What Does It Mean to Redefine Competitive Advantage?” — why management decides the order of selfdisruption
Vol. V, Ch. 046 “What Does It Mean to Redefine a Business Model?” — the procedure for rewriting the unit of revenue

Papers and companion volumes

Kadowaki, N. (2026). Future Value Theory: A Management Framework for Enterprise, Capital, and Society in the Age of AI. VURA Working stract=7120980 / Paper. Zenodo: SSRN: https://ssrn.com/abhttps://doi.org/10.5281/zenodo. 21255662
Kadowaki, N. (2026). Enterprise Redefinition: Toward an Enterprise Evolution Theory for the Age of AI. VURA Working Paper. (Published on Zenodo; under review at SSRN)
100 Questions on Management in the Age of AI, #069 “Where Does Innovation Come From in the Age of AI?”

Chapter 087 What Did Google Redefine?