Document · 121 blocks

Bought or Earned?

R E S E A R C H B R I E F I N G · A L P H A B E T D U E D I L I G E N C E · C O M P A N I O N T O " H O W G O O G L E M A K E S M O N E Y "

Bought or Earned?

Google’s default payments, its core strength, and what would happen if the money stopped

?

Google pays roughly $42 billion a year to be the search engine that is already switched on. This briefing answers one question: is Google's dominance a product of core strength — better algorithms, more user data, a stronger ecosystem — or a product of those payments? The evidence comes from the US search trial record, Google's own internal models, and every natural experiment where the default changed hands.

Prepared for: Bobby

Date: 9 August 2026

Sources: the liability and remedies opinions in US v. Google (D.D.C., Judge Mehta, August 2024 and September 2025), Alphabet’s SEC filings, academic field experiments (Allcott et al., NBER 2025; Decarolis et al., 2023), and measured share data from every market where the default changed. Figures described as estimates are estimates.

One-line answer: the floor is earned; the top ten to twenty points are bought; and the money that buys them is itself a product of the earned part. The full argument is in Chapter 6.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 1 / 11

1. The question

Google’s search share is about 89–90% worldwide. Two stories claim to explain it.

Story one: core strength. Google is simply the best. Best algorithms, most user data, best monetization, strongest brand. The default payments are just efficient marketing by a superior product.

Story two: bought monopoly. Google’s quality lead is real but not decisive. What keeps rivals out is $40+ billion a year paid to Apple, Samsung, Mozilla and the phone carriers so that no user ever has to choose Google — it is simply already there.

The distinction matters for the investment. If story one is right, the payments are optional: Google could stop paying tomorrow, keep most of its share, and add tens of billions to profit. If story two is right, the payments are the moat itself, the moat has a market price, and anything that disturbs the payment system — a court, a regulator, a richer bidder — disturbs the business.

A federal court spent four years on exactly this question. The trial record contains Google’s internal models, Apple’s internal deliberations, Microsoft’s rejected offers, and measured data from every market where a default actually changed hands. This briefing works through that evidence and ends with the counterfactual: what happens to Google’s economics if the TAC stops.

2. What the money buys

2.1 The payments

Google’s traffic acquisition cost (TAC) was $62.9 billion over the last twelve months. About $21 billion of that is passthrough to website owners in the ad business — irrelevant here. The relevant part is the other ~$42 billion (estimate): payments for default placement.

profit

All partners $26.3bn in 2021 — four times Google’s entire search R&D and other search costs combined Court finding

Mozilla Over $400m/year — about 80% of Mozilla’s operating budget Court finding

2.2 What the money produces

The court’s arithmetic: 70% of all US search queries flow through access points where Google is the default. Half of all US queries flowed through the specific contracts challenged at trial. Only 30% of queries happen anywhere Google’s presence wasn’t pre-arranged.

And the queries are the revenue. Google’s own documents: in 2017, defaults drove 54% of Google’s search revenue. On Samsung devices, 80% of search revenue came through pre-placed access points.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 2 / 11

2.3 Why defaults work — the court’s findings

The mechanism is not persuasion. It is inertia, and Google knows it precisely:

Google’s own behavioral team, 2021: “Inertia is the path of the least resistance. People tend to stick with the status quo.” A 2015 Google study covertly switched iPhone users from Google to Bing. About half never noticed. A 2020 study: over half of US iPhone users were unsure which search engine powered Safari at all. Replacing the Google search widget with Bing’s on Android was a 10-step process. The effect is strongest on phones, where the friction is highest — and phones are where the money went.

One more finding worth holding onto. Google ran an experiment deliberately degrading its own search quality by an amount equivalent to removing twice all of Wikipedia’s information. Revenue fell only 0.66–0.99%. Users barely left. Read that both ways: it shows how strong Google’s habit-and-brand hold is, and it shows that in the short run, users do not discipline quality. The default decides; the user follows.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 3 / 11

3. The case for core strength

The evidence that Google’s position is earned is strong, and it comes mostly from Google’s adversaries and partners — not from Google.

3.1 The quality gap is real, and everyone in the industry says so

From the trial record:

Apple’s Eddy Cue: “Google still has the best search engine by far.” Mozilla, internally: “Google is the clear winner when it comes to product experience and what users want.” Apple’s own 2021 study found Google superior to Bing on relevance across every access point except desktop Safari, with “a much larger lead on Mobile,” especially on rare, specific queries. The court itself: “The result is the industry’s highest quality search engine, which has earned Google the trust of hundreds of millions of daily users.”

One measured exception: Bing has reached rough parity with Google on desktop. Hold that thought — it matters in Chapter 4.

3.2 Apple’s own reasoning was quality first, money second

Microsoft offered Apple a 90% revenue share for Bing. Then 100%. Then offered to sell Apple the whole company. Apple said no every time. Cue’s testimony: there was “no price that Microsoft could ever offer” — because Bing’s quality and monetization were inferior, and “if you have an inferior search engine, customers wouldn’t use it.” Staying with Google was “as close to a sure thing as can be.”

Money mattered too — Apple’s analysis showed ~$40bn from Google over five years versus ~$20bn from Microsoft’s offer. But the point stands: the best-funded company on earth looked at the alternative and concluded it could not afford free.

3.3 The monetization gap: rivals cannot buy the defaults even at 100%

This is the single most important fact in the whole record. Google’s internal “Alice” study calculated that for Microsoft to match Google’s payment to Apple — then just 33.75% of revenue — Microsoft would have to pay 122% of Bing’s revenue. More than everything Bing makes. Google’s conclusion: “it will not be possible for Alice to match our payments profitably.” Events proved it exactly right: Microsoft offered 100% and lost.

Why this matters: the default slot is an auction, and the auction is won by whoever earns most per query. Google earns roughly three times more per search than Bing. That monetization edge is core strength — better ad systems, more advertisers, better targeting. So the payments are not an external subsidy propping up a weak product. They are the strongest bidder winning an auction it structurally cannot lose, for as long as its per-query economics stay superior.

3.4 When Google loses the default, users claw it back

Every natural experiment shows the same pattern: a meaningful minority follows the default; the majority finds its way back to Google.

Firefox/Yahoo (2014–17): Mozilla made Yahoo the default. Google’s share of Firefox queries fell from 80–90% to 60–70% — and then partially recovered while Yahoo was still the default. Two-thirds of the volume went to Google

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 4 / 11

anyway. Yahoo stuffed the results with ads to cover its guaranteed payments, users complained, and Mozilla switched back in 2017. Google’s share snapped back immediately. Windows desktop: Bing is the pre-installed default on every Windows PC. Google’s share on Windows is 80%, and has been since Chrome launched. Users defeat the default by downloading Chrome. Mozilla’s own experiments: when Firefox users were silently defaulted to Bing, Bing kept only 20–52% of the search volume depending on the test; Yahoo kept 16.5%. Most users switched back — 65% of those who switched did so within the first day. EU choice screens: since 2020, every Android phone in Europe has shown a search choice screen. Google’s mobile share moved by less than 2 points (measured, difference-in-differences). The court: “when users are given a choice today, they will overwhelmingly choose Google.” South Korea: the reverse case. Naver holds ~63% of Korean search against Google’s ~30%, despite Google being the default on most Korean phones. Preference can beat a default — when a genuinely preferred alternative exists.

3.5 Summary of the core-strength case

Google without any default would not look like Bing. The floor, visible in every experiment, is roughly 60–80% share in Western markets — carried by brand (“to search is to Google”), habit, quality, and a monetization engine nobody can outbid. On this evidence, most of Google’s position is earned.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 5 / 11

4. The case for bought position

Now the other side — which is why Google lost the case.

4.1 The contested layer is worth tens of billions

The 10–20 points of share that defaults control are not decoration. Google’s own 2020 internal model: losing just the Safari default would cost 60–80% of its iOS query volume, worth $28.2–32.7 billion in net revenue — double that gross. A 2015 internal presentation: Google’s position with Apple users “is still very vulnerable if defaults were to change.” Companies do not pay $26 billion a year, four times their search R&D, for something cosmetic.

4.2 The default holds whichever way it points

The Windows story from Chapter 3 has a mirror image. On Edge itself — where Bing is the default — Bing gets ~80% of queries and Google only 20%. Same users, same quality gap, opposite default, opposite outcome. The default is sticky within every surface; users escape it on desktop only because installing Chrome is easy. On phones, where the friction is high, there is no escape hatch — which is why Google’s mobile share (94.9% at trial) exceeds its desktop share, and why the payments are concentrated on mobile.

And where a real rival existed and the screen was well designed, the default moved real share: Russia’s forced choice screen took Google down ~7 points and made Yandex the #1 engine on Android in Russia within 18 months. Turkey: ~12 points. The EU screens failed because there was no credible rival on them — not because defaults don’t work.

4.3 The flywheel: the bought layer feeds the earned layer

This is the deepest finding in the case, and it partly dissolves the question itself.

Search quality is built from user data. The court found user interaction data is “a critical input that directly improves quality” at every stage — and the scale gap is staggering:

Google receives 9x more queries per day than all rivals combined — 19x on mobile. Of millions of unique query phrases, 93% were seen only by Google; on mobile, 99.8% of rare queries never reached Bing at all. Thirteen months of Google user data equals 17 years of Bing data.

So the flywheel runs: defaults → queries → data → quality → users and monetization → more money to spend on defaults. The court described exactly this loop and concluded the contracts “ensured that Google would reap the greatest benefit” from it while rivals “were starved of scale.”

The proof that distribution builds quality is Bing itself: the court attributed Bing’s desktop parity to its Windows default. “That added search volume has allowed Microsoft to improve its search quality on desktop… Yet… Google guaranteed default placement on all mobile devices [means] Microsoft has never achieved the mobile distribution that it needs to improve on that platform.” Same company, same engineers: parity where it has distribution, permanently behind where it doesn’t.

The consequence for our question: today’s “core strength” is partly the accumulated product of twenty years of bought distribution. The two stories are not alternatives. The payments bought the data that built the strength that funds the payments.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 6 / 11

4.4 The market never gets to test the question

Two more findings complete the picture:

Entry died. Only two notable search startups appeared in fifteen years. Neeva — founded by Google’s own former ads chief — shut down, citing inability to get “even a default provider” slot. His verdict on Google’s “competition is a click away” line: defaults are “enormously powerful,” despite the “pious prose.” Venture capital called search a “no fly zone.” Apple was paid not to compete. The court found the ~$20bn payment “disincentivizes Apple from launching its own search engine when it otherwise has built the capacity to do so.” Apple projected it would lose over $12bn in five years by separating from Google. The biggest potential entrant is on the payroll.

4.5 What a fresh look at users shows

A 2025 randomized experiment (Allcott et al., NBER) paid Google users to try Bing for two weeks. 33% kept using Bing afterward. 64% said it was better than expected. The authors’ conclusion: Google’s default works “because it ensures that users are never exposed to Bing, and hence never learn about it.” Search is an experience good; the default decides who gets experienced. Removing every friction at once, they estimate, would roughly double Bing’s share — from a very low base, so Google stays dominant, but the habit layer is softer than it looks.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 7 / 11

5. Every natural experiment, one table

act

licensing

The consistent reading: defaults move roughly 5–20 points of share where a plausible alternative exists, and nothing where one doesn’t. Google’s earned floor is high. The bought layer sits on top of it and is worth, at Google’s scale, tens of billions a year.

6. The counterfactual: what if the TAC stopped?

Now the economics. Distribution TAC ≈ $42bn/year (estimate) against $243bn of Google-owned search & other ad revenue and $148bn of Alphabet operating profit. Three scenarios.

6.1 Scenario A: nobody may pay for defaults (a universal ban)

Distributors must choose a default without payment. They choose on quality — and every witness agreed what happens next. Cue: “we have to pick what’s best for our customers, and today, that is still Google.” The court, rejecting exactly this remedy:

“On ‘day one’ post-judgment… because Google is the best search provider, they likely will maintain it as the default GSE… Google thus would continue to receive a disproportionate volume of search queries for a fraction of the cost. Freed of having to pay billions in revenue share, Google’s profits would increase. Not paying Apple alone would result in a windfall worth tens of billions of dollars.”

Effect on Google: roughly +$42bn pre-tax — about +28% to operating profit — with little share loss. This is the cleanest possible answer to the core question: the judge who studied it longest concluded that if payments vanished for

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 8 / 11

everyone, Google keeps the position free, because the position ultimately rests on quality. He refused to ban payments partly because it would enrich Google (and impoverish Apple, Mozilla and the carriers, who’d lose the subsidy).

6.2 Scenario B: Google alone stops paying; rivals may still bid

The dangerous version. Microsoft can offer Apple 100% of Bing’s revenue — recall it already did. Apple takes the money; Google loses the Safari default.

Google’s own model: it loses 60–80% of iOS queries, $28–33bn of net revenue, against ~$20–26bn of payments saved. Net negative, but only by single-digit billions in year one. Then the claw-back begins (Firefox pattern): Google recovers a substantial fraction as users switch back, while the rival — paying out 100%+ of revenue — bleeds cash on every query, exactly as Yahoo did before the deal collapsed. The real cost is not year-one revenue. It is the flywheel: years of lost mobile query data while a rival finally gets scale. That is the one asset the claw-back does not recover.

Conclusion: the payments are worth roughly what they cost — market-priced insurance, not pure waste. Google pays ~$20bn to Apple to protect ~$30bn of revenue and, more importantly, to keep the data pump running and the strongest potential entrant (Apple) on the payroll.

6.3 Scenario C: the status quo — which is what actually happened

Mehta’s September 2025 remedies left the payments legal: no exclusivity, one-year renewable terms, data sharing to rivals. As of August 2026: Apple still defaults to Google, no payment terms have publicly changed, and the first big new deal extends the relationship to AI — Apple chose a custom Gemini model to power the new Siri in January 2026, reportedly paying Google ~$1bn/year. Note the direction: for search defaults Google pays Apple ~$20bn; for AI, Apple pays Google. In search, Apple owns the shelf and Google rents it. In AI, Google owns the product and Apple rents it. The DOJ has appealed, arguing “a competitive window may close if Google can continue using its monopoly profits… to squelch the threat” from AI companies. Google’s reply is due 29 September 2026.

6.4 The sensitivity, in one line

Distribution TAC is ~17% of Google’s own search revenue. So dropping all payments is profit-accretive as long as Google keeps more than ~83% of that revenue — and every experiment in Chapter 5 says it would keep far more than that in a world where rivals can’t pay either. Only Scenario B — a rich rival buying the slots Google vacates — breaks the math, and only until the rival’s own economics break first.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 9 / 11

7. Verdict

7.1 The decomposition

Google’s ~90% share splits into three layers:

and all of the data moat

So the honest answer to the core question is: it is core strength, financed and reinforced by monopolized distribution — and the two are no longer separable. The payments did not create the quality gap, but they decided who got the data to widen it, and they denied every rival the scale to close it. Bing on desktop — parity where it has distribution — is the proof that the “quality gap” is partly just the distribution gap wearing a lab coat.

7.2 The three facts an investor should keep

1. The auction for defaults is won by monetization, and monetization is core strength. Microsoft offering 122% of Bing’s revenue still couldn’t match Google’s 33.75%. As long as Google earns ~3x per query, no rival can profitably outbid it, and the $42bn is not a vulnerability — it is the visible price of an unassailable position. The TAC line is best read not as a cost of weakness but as the dividend the moat pays to its gatekeepers.

2. The payments are rationally priced, not excessive. Pay ~$20bn to Apple; protect ~$30bn of revenue, the mobile data pump, and Apple’s neutrality. Mehta’s “windfall” finding confirms the flip side: in any world where rivals also cannot pay, Google’s profits rise ~28%. The genuine downside case requires a competitor who can outbid Google per query — which requires beating Google’s ad monetization first — which no one has done in twenty years.

3. The vulnerable layer is the bought one, and AI is the first credible bidder. Everything above holds while the monetization gap holds. An AI product that answers commercial queries and monetizes them better than a search ads auction would flip the default auction’s winner. That is why the Siri-Gemini deal matters more than its $1bn size: Google won the most valuable AI distribution slot before any rival’s economics matured — and this time the money flows the other way: Apple pays Google. The court explicitly kept jurisdiction to intervene if Google re-runs the old defaults game in AI. Watch the per-query economics of AI assistants, not their user counts.

7.3 What to watch

WHAT WHY

1 D.C. Circuit appeal — DOJ’s cross-appeal on the payment ban (Google reply The only live path to a forced Scenario A/B

due 29 Sept 2026)

2 Apple–Google annual renewals (now one-year terms) Repricing of the moat’s toll, every year, in public view

3 Siri-Gemini rollout and any OEM AI-assistant deals The defaults playbook migrating to AI

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 10 / 11

WHAT WHY

4 Safari query volume Cue already testified it fell in 2025 for the first time in 22 years

5 Any rival’s ad monetization per query (ChatGPT ads, Perplexity) The variable that decides who can bid for defaults

Bottom line. The question “core strength or bought monopoly?” has a measured answer. Take the defaults away from everyone, and Google keeps roughly its position and gains ~$40bn a year — that is core strength, and it is why the judge refused to ban the payments. Let a rival buy the defaults instead, and Google loses 10–20 points and its data pump — that is what the payments actually purchase. The moat is real; the toll is real; and the toll is affordable precisely because the moat is real. The only force that changes this arithmetic is a product that monetizes attention better than search ads do. Nothing has yet — and in the first big AI distribution deal, the money already flows to Google: Apple pays ~$1bn a year for Gemini in Siri.

Written 9 August 2026. Court findings are from the liability opinion (August 2024) and remedies opinion (September 2025) in US v. Google (D.D.C.). Google internal figures (query-loss models, the “Alice” study, revenue-by-default shares) are trial exhibits as characterized in those opinions. Natural-experiment data: trial record, StatCounter, Decarolis, Li & Paternollo (2023), Allcott et al. (NBER w33410, 2025). TAC split between distribution and network payouts is an estimate; Google stopped disclosing it in 2019. This is analysis, not investment advice.

Bought or Earned? — Google’s defaults and core strength 9 August 2026 · 11 / 11