31 October 2024

Contents

• Contents
• Introduction
• US Search & Distribution
• Overview
• Ruling
• Appeals
• Remedies
• Impact on Google
• Search & Distribution Footnotes
• Search & Distribution Useful Resources
• AdTech
• Overview
• The Trial
• Trial Outcome and Remedies
• Impact on Google
• AdTech Footnotes
• AdTech Useful Resources
• Epic vs. Google
• Overview
• The Appeal
• Impact on Google
• Epic vs. Google Footnotes
• Epic vs. Google Useful Resources

Introduction

Owning well-established businesses with strong competitive advantages often attracts increased regulatory scrutiny. This has affected various areas of our portfolio, with big tech companies being a particular focus for regulators. Google, a large holding for us, currently faces several legal challenges. We’ve owned Google since strategy inception (1 Jan 2019; plus a long time prior) and regulation was the primary risk identified in our original underwriting. We have closely monitored these cases over the years, reviewed many of the primary case materials and trial exhibits, and have recently consulted multiple US antitrust legal experts to assist our analysis. The trial exhibits are often treasure-troves of disclosures that we do not normally get as investors. In this piece, we outline our views on these cases and the potential impact for Google.

Google has an effective global monopoly in general search (Google), mobile operating systems (Android), browsers (Chrome), general search text ads, supply-side ad platforms (DoubleClick for Publishers and AdSense), ad exchanges (Google AdX), and ad networks for open web display advertising (Google Display Network/Google Ads). The same is true for the US market except for mobile operating systems where Android has only 42% share versus 72% globally. A jury also found that Google monopolizes app stores within the Android market, though we contend this finding. Under US law, there is nothing inherently wrong with being a monopolist if this position has been reached through fair competition. Sometimes a monopoly is the best outcome for consumers (i.e. natural monopolies). The issue arises when the monopolist abuses their position and engages in anti-competitive practices (such as exclusive contracts, bundling the monopoly product/service with those in other markets, predatory pricing, etc) and forecloses a market. European law is different and places obligations on monopolies, even if achieved legally.

Google has indeed engaged in anti-competitive practices over the years and is rightfully being held accountable. There are three key ongoing cases:

1. US Search and Distribution: This case, brought by the DOJ, focuses on Google's abuse of its monopoly in general text search, its control over the search distribution funnel, and the resulting harm through higher ad prices and reduced search competition/innovation. Judge Mehta recently found Google guilty, and the DOJ has presented proposed remedies. The remedy trial is scheduled for 2025, with Google having two chances for appeal—the final opportunity being the Supreme Court, should it choose to hear the case.
2. US AdTech: This case was also brought by the DOJ and alleges that Google abused its monopoly position in the supply-side ad platform, ad exchange, and open web display advertising markets. The trial has concluded and a decision is expected in November after closing arguments are given. The remedies are not likely to be separated in this case and we expect a Judge Brinkema to make a ruling on those in January 2025. Google will have similar appeal avenues here.
3. Epic vs Google (Play Store): This is an antitrust suit brought by Epic for Google’s abuse of its monopoly position in the Android app store market. Google lost this case in a jury trial and the judge imposed various remedies. Almost all have been put on hold pending the appeal outcome.

US Search & Distribution

Overview

Google has fortified its stronghold on search by controlling the search distribution funnel (see chart above) through acquisitions and contracts, maximizing flows to its search engine. Android, used on 72% of mobile phones globally, typically comes with Chrome as the default browser. Chrome also dominates desktop browsers with a 65% share. On all Chrome installations, Google is the default search engine. For Android phones, the Google Play Store is indispensable. Google leverages this by tying Chrome, Maps, YouTube, Gmail, and other Google apps to the Play Store, while mandating the use of Google-approved Android versions. In most cases, Google sweetens the deal with revenue sharing. This strategy extends beyond Android—Google pays Apple, mobile carriers, and various other browser owners for default search engine status on their platforms. As a result, Google is the default search engine on nearly 100% of mobile devices and about 85% of desktops. Around 65% of these (52% in the US) are covered by a contractual agreement for default status (”covered queries”).

Google's obsession with controlling the search funnel dates back to the company's early days. Ironically, the 2001 antitrust case against Microsoft partly helped Google dominate search. Without this case, Microsoft would have continued tying Internet Explorer to Windows, potentially helping MSN Search become the dominant engine. Google recognized its good fortune.

This focus on control became ingrained in Google's DNA during the rise of smartphones. Initially, US mobile carriers controlled default apps on the phones they distributed. Fearing interference between Google and its users, Google approached the FCC, offering to bid $5b in a spectrum auction if the winning bidder allowed device manufacturers to choose the default apps. Though Google didn't win the bid, it shifted choice away from the carriers.

Despite this, Google initially struggled, falling behind in contracts to be the default on Nokia phones, while Android-based phones were shunned by most telcos except T-Mobile. However, Google secured default status on BlackBerry and iPhone[1].

The development of the Android Open Source Project also stemmed from Google's desire to control the funnel. Fearing iPhone dominance could one day threaten Google's default status, they released an open-source version of Android to proliferate an alternative smartphone operating system. Once Android achieved scale, Google reasserted control by developing its proprietary version and requiring manufacturers to use it in exchange for access to the Play Store.

Google obsesses over funnel control for good reason — being the default matters. For instance, Google's share of desktop Edge users (where Bing is the default) is ~23% compared to 76% on Windows desktop overall. While this difference is stark, we shouldn't overestimate its impact based on this alone. Edge users might simply prefer Microsoft products rather than just sticking with the default.

Real-world examples exist where the default has been changed from Google to other engines. Mozilla Firefox has conducted various experiments over the years to analyze the impact of moving away from Google. They've trialled both Yahoo and Bing as the default for a random selection of users. The results? Bing retained 30-60% of users after a few months, and Yahoo 20% , with the rest switching mostly back to Google. Given that Bing already had a ~10% share of Firefox users and Yahoo 3%, this represents a 20-50% share loss for Google when Bing became the default and 17% when Yahoo became the default. The trial noted that the share loss would be greater on mobile (70%) than on desktop (30%) given it is slightly easier to change the default engine on desktop (though the reference exhibit is unfortunately sealed). The issue here is that Firefox users also have a selection bias. They are typically more tech-savvy and are incrementally more likely to understand how to change the default engine.

Another illustrative example occurred when Apple changed the default from Google Maps to Apple Maps. Two years after this switch, Google Maps had reclaimed 40% of its prior peak[2]. That said, over time, we think Google Maps’ share of iPhone users has climbed significantly since then. Some estimates have Google closer to 65% today.

Ruling

Unfortunately for Google, exerting control over the search funnel in this manner is illegal when the company holds a monopoly in search. Judge Mehta ruled that Google has:

• A monopoly in general search services (a market that excludes vertical search like Amazon or social media)
• A monopoly in general search text advertising (sponsored search links), but not in the broader search ads market (which includes vertical search but not display ads)
• Engaged in anti-competitive practices by controlling the search funnel, resulting in over 50% market foreclosure, depriving competitors of scale, reducing investment and innovation by rivals

Notably, the judgment was silent on how these findings translate into consumer harm—a necessary component under US antitrust law. Presumably, the harm is oriented around reduced consumer choice.

Google did achieve some victories. The ruling that it lacks a monopoly in general search advertising safeguards certain actions beyond text (such as maps and shopping, which have previously faced regulatory scrutiny). Judge Mehta also determined that Google initially attained its position through superior innovation — a finding crucial to the remedies outcome (discussed later). Moreover, the judge reiterated that Google has no obligation to assist competitors, thus legitimizing certain actions to exclude Bing. Importantly, the trial didn't examine whether Chrome or the Play Store is a monopoly. Although indeed it is, this omission will restrict certain potential remedies (explored below). In our view, the DOJ missed an opportunity here.

Appeals

We believe Google's appeals are unlikely to succeed, though they may delay the implementation of remedies. The plaintiff's case was robust, and Google can only appeal conclusions of law, not findings of fact. Google will likely challenge market definitions, attempting to include vertical search, and argue that its agreements result from superior quality, aligning with Judge Mehta's ruling on how they initially achieved their monopoly. We observed some minor flaws in the market definition arguments. For instance, the monopoly in general search services was based on all queries, both non-commercial (80%) and commercial (20%). However, a key piece of evidence proving monopolization was Google's supra-normal profits, which are only relevant to commercial queries. This inconsistency in market definitions between the accusation and the proof wasn't addressed by the defence.

Google is likely to appeal the foreclosure level calculation. During the trial, Google contended that while over 50% of US queries are covered queries, this should be evaluated against a hypothetical scenario where such agreements didn't exist. Google argued that because it would reclaim some market share in this scenario, the actual foreclosure level should be lower than the covered query percentage. However, Judge Mehta rejected this argument.

Another potential avenue for appeal, as mentioned earlier, is the trial and ruling's lack of detail on consumer harm. Surprisingly, in Judge Mehta's 286-page ruling, the phrase "consumer harm" appears only once, and that's while referencing another case:"Even though monopolistic conduct requires proof of actual or threatened consumer harm, the proof need not invariably be elaborate.” Rewind the clock, and you'll find much of the investment community believed it impossible for Google to be found guilty under US antitrust law because its products are free for consumers. We disagreed then and disagree now. Although the trial was light on this aspect, we don't believe Google will succeed in an appeal based on this point.

“One might yearn to suggest that there is a market unjust here that should be investigated by some government entity, but let us not forget that the consumer is not harmed here — in fact far from it. The consumer is getting great software at the cheapest price possible. Free. The consumer might be harmed if this activity were prevented.” — Bill Gurley, “The Freight Train That Is Android”, 2011 (to be fair, he was right for over a decade!)

We also noticed some minor data discrepancies. For instance, we believe some of the "% of queries" figures presented by the DOJ and accepted by Judge Mehta as findings of fact are inaccurate[3a]. Additionally, we suspect the court misinterpreted Google's internal estimates of potential share loss[3b]. Lastly, much of the data relied on query share tools like GS Stat Counter. While these tools provide useful ballpark estimates, they're prone to inaccuracies due to evolving rules on what data browsers allow to be seen. Google's decision not to contest this suggests the actual numbers might be even higher.

Separately, Google may succeed in appealing the remedies, depending on their severity and appropriateness. The remedies trial could also result in a settlement if Google and the DOJ reach an agreement. This outcome becomes more likely if Trump wins office, as he has recently stated that he doesn't believe Google should be broken up and that it's important for Google to remain competitive with China.

Remedies

The remedies proposed by the DOJ are extreme. The DOJ is “considering remedies to address four categories of harms related to Google’s (1) search distribution and revenue sharing, (2) generation and display of search results, (3) advertising scale and monetization, and (4) accumulation and use of data[5].”

Remedies should be designed to right the wrongs, including helping competitors reach the position they might have achieved without these violations. Under US law, remedies shouldn't focus on removing a monopoly gained through legal means (as Judge Mehta ruled was the case for Google). They must consider whether the same goal can be achieved with less drastic action, whether there are potential unintended consequences, and the viability of any new standalone businesses created. Remedies shouldn't undermine the defendant's core business — Google can't be forced to give away its "secret sauce." Lastly, remedies should be based on the trial's legal conclusions, and the plaintiff must prove their reasonable effectiveness.

(1) Search distribution and revenue sharing

The DOJ will aim to severely restrict Google's control over the search funnel. They propose banning all contractual arrangements and revenue sharing agreements that make Google the default search engine, including the tying of Google's apps to the Play Store on Android phones. The DOJ has suggested both behavioural remedies (restricting contracts and payments) and structural remedies (spin-offs).

Unlike in the AdTech case, structural remedies in this case are improbable. The two main structural remedies would be spinning off Chrome and/or Android. Spinning off Chrome is illogical. Browsers are poor standalone businesses, and a standalone Chrome would require the very default payments the DOJ aims to eliminate. Moreover, Chrome and Google Search are intrinsically linked — it's impossible to distribute search without a browser. The same objective could also be achieved through behavioural remedies — namely, preventing the tying of Chrome to the Play Store's distribution while Google Search remains the default, and/or forcing Chrome to use a choice screen. Similarly, for an Android (or even a combined Android-Play Store) spin-off, the same goal could be accomplished by prohibiting the tying of Google Play to any additional app where Google Search is the default. Google would have a strong argument that structural remedies are unnecessary in this case.

The simplest and most effective solution is behavioural remedies. These will likely involve significant restrictions on all contracts and revenue sharing agreements related to Google Search. For Android, we believe the restrictions will be limited to preventing Google Search from being tied to the Play Store (via Chrome). Although Chrome and Google Maps are monopolies, the trial did not address or conclude this. Similarly, the trial did not determine that the Play Store is a monopoly (only that it's a "must have"), despite it being deemed one in the Epic-Google case (discussed later). Consequently, it should remain legal to tie Google Maps, YouTube, and Chrome (with a choice screen), as well as other apps like Gmail, to the Play Store.

The potential remedies could range from prohibiting Google from contracting or paying for default status while allowing competitors to do so — a practice only illegal due to Google Search's monopoly status — to implementing a choice screen. This compromise, similar to the one introduced in Europe following the European Commission's Google Android case (and later reinforced by the EU Digital Markets Act's gatekeeper provisions), was ironically advocated by Google itself in 2005 for Microsoft's Internet Explorer. However, the US legal system differs from Europe's, and remedies can only be imposed on the guilty party. This raises questions about how the court could compel Apple or other distribution points to implement a choice screen or prevent them from accepting default payments from non-monopoly search engines. A potential solution to incentivize these parties might be allowing Google to continue revenue-sharing agreements for inclusion on the choice screen, possibly even securing a top position.

“We proposed instead that users be prompted to select the default search provider the first time they use the inline search feature. This approach eliminates any company's own self interests and places control in the hands of the end user, where it belongs.” — David Drummond, Former Senior VP at Google, emailing the GC of Microsoft in 2005. Source: DOJ Trial Exhibit UPX0172

If the DOJ can't force Chrome's divestiture, they may push for a choice screen on all Chrome versions (including desktop Chrome, Pixel phones etc) even for queries not covered by default agreements. This argument is more challenging, though not impossible, as Chrome was primarily built to distribute Google Search. Moreover, neither the DOJ's arguments nor Judge Mehta's ruling declared Google Search's default status on self-distributed Chrome illegal. However, the judge did conclude that "it is a market reality that significantly narrows the available channels of distribution and thus disincentivizes the emergence of new competition." The DOJ sought to include user-downloaded Chrome numbers in the foreclosure calculation (and Google argued the opposite), but the judge left this unanswered as the threshold was met without it. One argument for including self-distributed Chrome in the remedies is the court's conclusion that Chrome is a separate product from search. It will come down to a judgement whether tying Google Search to self-distributed Chrome is aimed at foreclosing competition. We think this is unlikely.

(2) Generation and display of search results

The DOJ has indicated its desire to impose remedies preventing Google from using data gained through its search monopoly to cement advantages in other markets. For instance, they aim to restrict Google's use of its web index and accessed publisher content for training AI models. Websites typically prefer to allow only a few search web crawlers to avoid performance degradation, which naturally advantages large search engines like Google. However, the trial did not address this issue, nor did the DOJ demonstrate that AI constitutes a separate market. As such, this proposal appears to be a red herring. At best, it's likely a negotiation tactic by the DOJ — something they're willing to concede during potential settlement talks.

(3) Advertising scale and monetization

The trial concluded that Google is a monopoly in search text advertising and that its related anti-competitive conduct was charging supra-normal ad pricing (by influencing auctions) and impairing competitors text ad revenue by foreclosing the general search services market.

The DOJ has proposed some odd remedies here, like dampening Google’s ability to use artificial intelligence to improve ad performance (e.g. Performance Max). They even suggest licensing Google’s ad feed independently to its search results. Degrading ad performance is a sub-optimal solution for Google’s customers so its unlikely to be considered by the court. Forcing Google to license its ad feed would also undermine its core business (its part of the “secret sauce”). Search and advertising are intrinsically linked as a unified business model and it represents one of the advantages that, as ruled by Judge Mehta, Google won legally. As such, we can’t see this being considered by the court.

Given most of the advertising harm discussed in Judge Mehta’s ruling was centred around the impact of the distribution agreements, the same distribution remedies should suffice here. The only addition, and strangely not suggested by the DOJ, would be to ban Google from artificially raising text ad auction prices.

(4) Accumulation and use of data

The DOJ has suggested that Google be forced to licence its search index and potentially its search results and ranking systems. These elements also underpin Google Search’s core advantages that it won legally. As above, we think these remedies would unjustly undermine Google’s business model and will therefore be rejected by the court. There is potential for a watered down version where rivals get some sort of access for a limited period as compensation for the harm done (akin to the Play Store remedies discussed below). However, it would be very complicated to execute in practice given privacy and competitive considerations. Providing it in a form that is both compliant and useful might be impossible. The burden will be on the DOJ will to come up with a solution that lowers barriers to entry while solving for these issues.

Impact on Google

The most likely outcome is a severe restriction on Google's control over the search funnel through a ban on its ability to contract or pay for default status. We estimate that 65% of global search queries (52% in the US) are Google searches via devices subject to a contract or revenue share agreement (termed "covered queries" by the court; see chart below). We evaluate the potential share loss and financial impact of this below. However, more crucial is how this affects the quality of the business. Ultimately, if Google Search maintains its strong value proposition backed by structural competitive advantages, it will continue to generate attractive economics. As seen with payment networks facing various regulations over the years, robust companies always find new ways to create and capture value. We remain confident in Google's prospects, given the strength of its business today and our belief that the potential share losses shouldn't materially alter its fundamental strength.

The Mozilla real world experiments (discussed above), are probably the best representation (and least biased) of what would could occur if the defaults were switched to a competitor (i.e. average 35% share loss). However, we can’t just assume a 35% loss for all defaults as it would overestimate the impact.

First, Google is a superior search engine than its competitors. Bing is comparable in core desktop search but doesn’t match Google in long-tail search or on mobile. As such, companies may not want to switch their defaults away from Google, even if it meant losing short-term profit. Further, Android smart phone users have become accustomed to and expect Google’s products. The OEMs may continue to install these apps as the default in fear of losing customers, especially if the contracts continue outside of the US/Europe anyway. Similarly, Apple wants to offer its customers the best user experience, which means the best search engine. Browsers like Firefox would have to switch as they depend on such revenue to survive, but they are a small share of covered queries compared with Apple and Android.

Second, it may be that choice screens become the norm. If used by all access points, this could actually be a good thing for Google. It could be similar to the ban on advertising for tobacco companies that enabled them to maintain market share while saving significant S&M costs.

In the scenario where defaults are awarded to a competitor, we estimate Google retains 55-60% (about a 35% share loss as not all were using Google despite it being the default). As mentioned though, not all providers would choose a default other than Google. If 50% did select another default, this would equate to an 18% share loss of covered US queries and 9% of total US queries (double if 100% selected a default other than Google). Professor Michael Whinston, an expert witness for the DOJ, estimates a 66% loss of covered queries (33% loss of all US queries) but he assumes 100% of covered queries change to a default other than Google plus to estimate the loss, he uses the average of Google’s share of Internet Explorer and Edge queries (44% and 23% respectively[4]). As explained above, this is an overestimation, which is not surprising given the incentives driving the DOJ. Some of Google’s own estimates were released at trial. They use the Mozilla example as their base case on Mac and as their upside case on iOS. Their downside case on iOS was based on the Apple Maps default switch (i.e. ~60% loss of covered queries; discussed above). However, we don’t think the Apple Maps case is representative unless Apple launched its own search engine that was reasonable in comparison to Google. Further, as mentioned, Google Maps’ share of iPhone users has steadily climbed since then to something more like 65% (35% loss).

If a choice screen were implemented instead, we don't expect any significant change in query share. As discussed in the trial[5], this approach was implemented in Europe, where Google still wins about 90% of covered queries — roughly the same as its overall query share.

Overall, we estimate the US query share loss to range from 0-10% (or 0-3% as a percent of global traffic and 0-18% as a percent of US covered queries). It could be greater than this in the short-term, as was the case in the Firefox experiments and Apple Maps since it takes time for users to switch back. Our share loss estimates only assume the Chrome default will be removed for covered queries (i.e. where Chrome is tied to the Play Store on Android) not for self-distributed Chrome.

The potential share loss could be more than offset by savings from eliminating or reducing default payments. Traffic acquisition costs (TAC) amounted to $29 billion globally in 2023 (excluding display network revenue shares), representing 16.5% of total search revenue and 26% of covered search revenue. Revenue share agreements vary by provider. Previously estimated through regression analysis, we now have specific details from trial disclosures. The Epic-Google trial revealed Apple's revenue share as 36%. Judge Mehta disclosed that Google paid Apple $20 billion globally in 2022, which we estimate rose to ~$21.5 billion in 2023 (constituting the majority of search TAC). This implies Google earns about one-third of its global search revenue through Apple, with the average revenue share for other providers at ~11% of covered revenue. At the upper end of our share loss estimate (considering only the US market due to the case's jurisdiction), we project Google could gain $8 billion in profit by ending TAC payments to Apple[6]. For all other providers, they would lose $0.5 billion. This represents a net ~10% boost to operating income. Additional revenue opportunities may even arise if Google can charge licensing fees for Android and/or the Play Store.

An additional complication arises in this analysis. As Google's share loss increases, it moves further from a monopoly position. A key advantage of a monopoly in digital advertising is that when ad volume growth slows relative to demand, natural upward pressure on auction prices allows revenue to continue growing. Without this monopoly, the pricing safety net weakens, as it becomes dependent on competitor volumes. We believe this dynamic remains intact at a 10% US share loss, but it's a crucial consideration. Moreover, this analysis doesn't account for how revising US contracts might affect global agreements. Extending this globally, the additional gain from ending the Apple contract is offset by losses elsewhere, still resulting in a ~10% boost to operating income.

A final potential impact is the possibility that competitors will bring their own suits against Google for damages. This is very hard to quantify.

Search & Distribution Footnotes

[1] Richard L. Brandt, “The Google Guys: Inside the Brilliant Minds of Google Founders Larry Page and Sergey Brin”

[2] Email from Mike Roszak Google VP of Finance, trial exhibit UPX0097. We assume that this also equates to a 60% share loss as Google Maps was likely close to 100% share prior to the launch of Apple Maps.

[3] (a) When concluding some query share losses, they appeared to neglect that Google didn’t have 100% share to begin with. We also question their “user-downloaded Chrome” figures. They use Prof. Michael Whinston’s numbers of 20% in the US while we have it at 8%. We think they are incorrectly including Chrome on Samsung given Samsung also comes with its own default browser alongside Chrome.

[3] (b)Judge Mehta’s ruling states “In 2020, Google’s internal modeling projected that it would lose between 60–80% of its iOS query volume should it be replaced as the default GSE on Apple devices.” However, the (unfortunately redacted) trial exhibit UPX1050 sets the worst case equal to the Apple Maps loss, which was 60%. The best case was Mozilla (circa 40% loss). So we think Judge Mehta should have concluded 40-60% not 60-80%.

[4] Plaintiff’s Proposed Remedy Framework, Case No. 1:20-cv-03010-APM

[5] Trial Exhibit UPXD104, Prof. Michael Whinston Presentation “Monopoly Maintenance & Competitive Harm”

[6] This requires several assumptions on how query shares translate into revenue shares. We know the non-Apple TAC globally is ~11% of covered revenue. We assume it is 15% in the US, given US users are worth more than non-US users. We also assume that a US-iOS user is worth 1.2x more than a US-non-iOS user given their more attractive income profiles, and that a desktop user is worth 1.1x more than a mobile user given the higher propensity to monetize and higher CPC rates.

Search & Distribution Useful Resources

• Plaintiff’s case materials
• Trial materials, trial exhibits, original complaint, proposed remedy framework
• Note: Our favourite page of the many many materials is page 12 of the DOJ’s Complaint. A great antitrust pun for the avid reader!
• Trial transcripts
• Judge Mehta’s Ruling
• Detailed Trial Summary
• Various Court Filings

AdTech

Overview

Google's dominance extends beyond the search funnel into the AdTech ecosystem. As previously mentioned, they hold monopolies in general search text ads, supply-side ad platforms (DoubleClick for Publishers and AdSense), ad exchanges (Google AdX), and ad networks for open web display advertising (Google Display Network/Google Ads). Supply-side platforms consolidate ad space from website publishers for sale through real-time auctions on ad exchanges. Demand-side platforms perform a similar function for advertisers (see chart below). This case focuses on Google monopolization of this ecosystem and its associated anti-competitive practices. The complaint is limited to display advertising rather than search advertising.

Google launched its current demand-side platform, Google Ads, in 2000, followed by its supply-side platform, Google AdSense, in 2003. Initially a platform for search advertising bids, it expanded to include a network of publishers selling their own ad space. The 2007 acquisition of DoubleClick solidified Google's position, providing a leading publisher ad server and inventory (DoubleClick for Publishers “DFP”) along with an ad exchange (AdX) and a demand-side platform (DoubleClick Bid Manager, now known as Display & Video 360 “DV360”). While DFP caters to larger publishers, AdSense focuses on small and medium-sized enterprises. In 2010, Google further expanded its reach by finalizing the acquisition of AdMob, essentially creating a mobile app equivalent to AdSense.

Google’s dominance in search meant that access to its search ads through Google Ads was a “must have”. This meant that Google built up very strong advertiser demand in Google Ads and DV360. By limiting this large demand pool (especially in Google Ads) to AdX and DFP, these too became dominant businesses. Reinforcing this is a strong data advantage. Google is able to use data from its search and other businesses to help advertisers with targeting even when not on a Google property. It supplements this with a large data pool generated from its publisher network. Over time, various parts of the value-chain became more open (e.g. allowing DFP publishers to offer inventory to competing exchanges and allowing some non AdX inventory to Google Ads). However, Google often gave its own assets preferential treatment to ensure continued control of the ecosystem. Fast forward to today and Google has a monopoly or near monopoly in each segment (depending on how the market is defined).

Control of the AdTech ecosystem also reinforces Google Search. The additional data pool improves targeting and the performance of Google Search ads.

The Trial

The core part of the trial has concluded. Closing arguments and the final judgment are expected in November, with remedies in early 2025. Unlike the search case, a separate remedies trial is unlikely. The DOJ argued that Google engaged in anti-competitive practices to achieve monopolies in publisher ad servers (DFP with 90% share), exchanges (AdX with >50% share), and advertiser ad networks (Google Display Network, accessed through Google Ads, with 80% share). With the case focusing on illegal practices to build the monopoly, a guilty verdict would likely lead to remedies breaking down the monopoly, most likely through divestitures. This contrasts with the search case, which found illegal practices only after the legal monopoly was established. Notably, the case doesn't target AdSense, DV360, AdMob, or search advertising.

The alleged anti-competitive conduct includes:

• the acquisition of DoubleClick to pursue Google’s goal of dominance across the entire ad tech stack
• the restriction of Google Ads’ advertiser demand exclusively to AdX
• the restriction of effective real-time access to AdX exclusively to DFP
• various actions that favoured Google’s own ecosystem ahead of rivals at the detriment to its own customers

The market classification is crucial to the DOJ's argument. The case focuses on open web display advertising — display-based advertising on websites. The DOJ contends that this is distinct from other forms of advertising such as Google search ads, Facebook/Instagram ads, and Amazon ads. There's a strong case, now supported by legal precedent following the Google Search case verdict, that display ads are a separate market from search ads. This distinction also applies to Amazon ads, which essentially function as search ads limited to a specific vertical. The ad formats and the advertisers bidding on ad placements differ for each (e.g. only certain advertisers can bid on the keyword "hotel"). However, Facebook and Instagram also function as website publishers that auction display advertising space. Social media ads and open web display ads both occupy the top of the marketing funnel, serving the role of product/brand discovery and awareness. Judge Mehta in the Google Search case even concluded, "Social media advertisements are essentially display ads that are integrated into a social media feed." Indeed, the DOJ even made up the term “open web” and witnesses on both sides attested that they do not discuss the market like this in practice. They are not separate markets in our view.

Interestingly, the DOJ's argument for distinction hinged on the fact that Meta Platforms and others don't sell their inventory through an open auction, but rather through a "walled-garden" closed system controlled internally. We suspect the DOJ avoided arguing about differences in the ads themselves, as this would likely be a losing argument. Meta Platforms' choice to sell its inventory through its own closed ad ecosystem shouldn't affect the market definition, and Google certainly made this case, pointing out that open web display ads have been losing market share to social media display ads.

The DOJ argued that supply-side platforms, exchanges, and demand-side platforms constitute three distinct markets. Google countered by comparing their integrated offering to an eCommerce marketplace, asserting that these are simply different components of the same market. In our view, the DOJ's argument was more compelling, as each component serves a unique function and need not be exclusive.

For supply-side platforms, the DOJ argued that DFP differs from a supply network like AdSense, claiming the latter caters to less sophisticated advertisers. They applied the same logic to distinguish ad demand networks like Google Ads from general demand-side platforms. The DOJ emphasized that both supply and demand-side platforms serving larger businesses offer more openness and broader reach. However, this argument seems tenuous. In reality, large advertisers frequently use both Google Ads and AdSense, which essentially serve the same purpose.

On the anti-competitive conduct, the DOJ’s case was very strong. If Google is found to be a monopolist then this conduct is very clearly anti-competitive and aided Google to form the monopolies in the first place. Google really needs to win the market definition arguments.

Various commentators, including Google's defence, argue that under US antitrust law, Google has no duty to deal with competitors, and therefore the exclusivity of Google Ads and DFP to AdX isn't anti-competitive. This argument is too simplistic. The "no duty to deal" concept in US antitrust law isn't absolute (as demonstrated in Aspen Skiing Co. v. Aspen Highlands Skiing Corp, 1984). Courts evaluate whether the refusal to deal stems from legitimate competition or an intent to foreclose a market. If Google is deemed a monopoly in DFP and Google Ads, and AdX is considered a separate market, Google's refusal to deal with other exchanges becomes problematic for two reasons. First, other exchanges aren't competitors to DFP or Google Ads — it's a refusal to deal with a customer (in DFP's case) or a supplier (in Google Ads' case). While the "no duty to deal" concept isn't limited to competitors, it's harder to justify when it affects the supply chain. Second, if Google holds a monopoly in these segments, it represents using monopoly power in one market to stifle competition in the ad exchange market. The court would need to weigh this against any legitimate competitive rationale Google might offer.

Trial Outcome and Remedies

We anticipate Judge Brinkema will likely concur with the DOJ's argument that supply-side platforms, exchanges, and demand-side platforms constitute separate markets. Google should succeed in arguing that DFP and AdSense belong to the same market; similarly for Google Ads and DV360. However, this alone won't be enough for Google to win the case. The trial's outcome hinges on whether Judge Brinkema determines that open web display ads and social media ads are distinct markets. We side with Google on this point, as does Judge Mehta's ruling. If Judge Brinkema sides with Google here then it would win the case for them. Without this separation, Google does not have a monopoly in display advertising (see charts below).

“In trying to show that Google has market power, the DOJ had to hypothesize a market limited to banner ads on websites. But that blinkered definition misses many other types of display ads, let alone all the other places people see ads – in apps, on social media and on connected TV. Benneaser John, CTO at Xandr after it was bought by Microsoft, testified that its integrated ad tech platform transacts across a whole range of digital advertising surfaces and formats. To try to win its case, the DOJ labored to come up with a market for “open web display advertising.” But many of its witnesses acknowledged that they had heard the term only in the context of this lawsuit, never in the real world. Even the DOJ’s own expert testified: “That particular combination of four words is not used commonly.” Andrew Casale testified that putting the four words together “is just not a common way we speak in ad tech.”” — Google, "Key takeaways from the ad tech trial”

A guilty verdict would necessitate structural rather than behavioural remedies. Firstly, if the monopoly was illegally obtained, breaking it up is justified. Secondly, the complexity and opacity of the AdTech market would make monitoring compliance with behavioural remedies (such as bans on certain practices) challenging. Moreover, behavioural remedies wouldn't dismantle the already established monopoly.

The DOJ has requested the divestiture of DFP and AdX. While divesting DFP alone could address most of the issues by ensuring Google represents only one side of the bidding process, maintaining AdX would still allow Google to extract rents from publishers for access to Google Ads demand. This demand largely stems from Google's monopoly in text search advertising, which the company has been found guilty of illegally maintaining in the search case. However, a complicating factor is that Google Search remains entitled to sell its demand through a company-owned exchange, as is the case for Meta Platforms and Amazon. On balance, we think a guilty verdict would mean a forced divestiture of DFP but not AdX.

This case has potential implications for Meta Platforms and other tech companies. If Judge Brinkema determines that social media ads are distinct from display ads, it could pave the way for a monopoly case against Meta Platforms and its ownership of its adtech ecosystem (should there also be anti-competitive conduct). Moreover, if the justification hinges on the "walled-garden" concept, any closed system might be considered a separate market (which does not make sense to us).

Impact on Google

This case primarily affects Google's display ad business, which accounted for only 4% of net revenue in 2023 but ~9% of operating income[7]. The actual impact would likely be smaller, as Google wouldn't lose all its display revenue. Google Ads would remain an attractive platform for advertisers to purchase display advertising, and it would still retain a share — albeit potentially at a lower take rate due to the same inventory being offered elsewhere. Depending on the value achieved through divestitures, it might even result in a short-term value gain. However, the more significant impact lies in the loss of control over the ad ecosystem and the data that publisher websites feed into the broader targeting pool. In essence, it's the degradation of overall business quality that matters most. We're less concerned about the impact here than in the search or Epic cases.

AdTech Footnotes

[7] Revenue less all revenue sharing

AdTech Useful Resources

• Plaintiff’s case materials
• Trial materials, trial exhibits, original complaint, proposed remedy framework

Epic vs. Google

Overview

As discussed in the Google Search case, the Google Play Store is essential for Android-based smartphones. Similar to Windows on desktop, combined mobile operating systems and app stores exhibit strong network effects. Developing apps for multiple operating systems is costly, so developers focus on platforms with the highest monetization potential (iOS App Store and Android Play Store). In the Epic vs. Google case, the Play Store was ruled a monopoly in the Android app store market. This monopoly status made Google's practice of forcing developers to monetize solely through the Play Store illegal. Judge Donato has determined remedies, though most are on hold pending appeal. Some remedies aim to prevent Google from mandating its own billing system for in-app purchases (where it takes a cut). Others target the monopoly itself: Google would be barred from forcing OEMs to set the Play Store as the default and would have to allow its app catalog to be sold through third-party app stores. All remedies would be in effect for 3 years, giving competing app stores an opportunity to achieve the scale necessary to compete effectively.

The Appeal

Epic brought an almost equivalent lawsuit against Apple. However, in that case, Apple won. In the Apple case, the App Store and the Play Store were deemed to be competing in the same markets, while in the Google case, they were not. The Play Store is only a monopoly if Android is a distinct market. Ben Thompson at Stratechery rightly points out that there are some key differences. Principally, Apple is and always has been a fully closed system and therefore there are no illegal agreements with distributors of iOS that force them to exclude the Play Store. Further, because it has always been this way, developers are not harmed by subsequent foreclosure of the market. In contrast, Google started as an open system and progressively became more closed as Google’s power grew. However, this only makes Google’s conduct illegal if it is also a monopoly while doing it. The fact that this was a jury trial likely also helped. We think Google has a strong chance of overturning this case on appeal as the true reality is that the primary area of competition is winning consumers at the point of device purchase, where the app stores matter. The Play Store does not exist in a separate market to the App Store.

Impact on Google

Play Store revenue is significant. We learned in the trial[8] that the Play Store earned $11.2b of revenue in 2019 with $7b in operating profit (63% margin). This includes both app sales revenue and advertising revenue, which hit different line items in Google’s reporting. Epic alleged that by 2021 its operating profit had grown to $12b with an operating margin of 70%. This was broadly in line with our estimates prior to the trial. $12b represented 15% of Alphabet’s 2021 operating income. We estimate it will be closer to $19b of operating profit in 2024 (still 15%).

Assuming Google loses the appeal, the first order effect is that it would lose in-app revenue on key apps. There are various market estimates for in-app purchases, but most point to in-app revenue representing 30-40% of total Play Store operating income. Not all app developers would utilise their own billing and Google may lower its take-rate for in-app purchases to further encourage this. A 20% loss of Play Store operating income would mean a 3% drag on Google’s overall operating income.

The second order question is whether the Play Store would be out-competed by an alternative. We think this is unlikely. 3 years is a short-time to overcome the existing network effects and any app that does not offer in-app purchases has no incentive to deal with a new store. However, the disincentive would be a lot lower than it is today given the remedies also undo the tie between Google’s version of Android and the Play Store. With that tie, the app developer would also have to redesign the app to be compatible with a slightly different operating system (i.e. a different Android fork). The remedies proposed means this would no longer be required so the network effects of the Play Store are reduced, even if still significant.

Taking this into account, in addition to the strong grounds for appeal, the overall impact to us appears limited.

Epic vs. Google Footnotes

[8] Epic vs. Google Complaint p60

Epic vs. Google Useful Resources

• Plaintiff’s Complaint
• Google’s Response to Epic’s Complaint
• Epic’s Response to Google’s Response
• Trial Materials
• Judge Donato’s Explanation of Remedies