Crossroads Capital's letter to partners for the month of September 2024, discussing the beginning of the end of Google's ad tech monopoly.
Dear Partners and Colleagues,
Last September, prior to the start of the DOJ’s trial against Google’s alleged monopolization of the search market, we laid out a thesis that pointed out investors’ complacency about Google’s antitrust risks in both the search and ad tech trials. While the outcomes of the search trial were hard to handicap (until June 4th, more on that later), we saw a very compelling case and a definable set of outcomes in the ad tech trial – and most importantly, wildly mispriced publicly traded beneficiaries in the event of a Google loss. In both cases, the dominoes are starting to fall, and our thesis is coming to fruition.
I. Search Trial – DOJ Wins – Potential Implications
On August 5th, 2024, the judge in the search trial ruled that Google illegally held a monopoly in search and text ad markets. While this was a surprise to many, our conviction that Google would lose the search trial had grown following Google’s loss to Epic in December 2023. The unanimous verdict in the Epic trial was supported by the fact that Google had deleted more than 20,000 documents and internal chats. The judge also ruled that any time there was mention of intentionally deleted chats by Google, this lack of evidence could be interpreted by jurors as negative to the company. So any time jurors encountered missing chats, or the phrase “let’s turn off the chat history,” they were free to assume Google was hiding potentially illegal monopolist behavior. Even worse for Google, its own lawyers joked in internal emails about putting any competition- or market-related communications “under fake privilege,” – i.e. claiming attorney-client confidentiality to hide evidence of anticompetitive behavior.
The Epic verdict proved that 12 American citizens could accurately understand a dense, technology-based case – which had been a supposed sticking point for the ad tech trials. Concerning deletion of evidence, the DOJ filed the relevant documents from the Epic ruling in both its search and ad tech trials, and the Eastern District of Texas (EDTX) did the same for its ad tech suit as well.
These actions made a huge difference in the search trial. The DOJ requested some form of sanction against Google given that “[t]he sheer volume of destroyed documents is remarkable.” Critically, the judge said during concluding remarks that “Google learned from what happened to Microsoft and had ‘employee after employee after employee’ use means of hiding documents.” Ouch. Given that this judge had been extremely deferential to Google during the trial, this comment was quite a tell.
All that leads us to the last piece in our search trial mosaic: Google’s abrupt firing of its Head of Competition Law (who had been at Google for over 15 years and was one of the creators of the company’s antitrust defense), on June 4th, not long after the trial’s concluding remarks. This move was cleverly addressed, and somewhat hidden, in a whitewashed press release on privacy personnel changes. Our conversations with a few antitrust lawyers familiar with the case and people involved aligned with our initial thoughts: You don’t fire a longtime senior executive unless he really messes up. So, when the verdict came, we were thrilled, but not surprised.
The search ruling is a clear win for a broader application of antitrust enforcement beyond the old consumer welfare standard, a principle in antitrust law under which a monopoly or business practice is determined to be illegal by considering whether it harms consumers – typically by leading to higher prices or reduced quality of the product in question. In the search trial, the DOJ successfully argued that even though Google Search is free to consumers, the elevated prices paid by advertisers, the restricted competition and innovation in the industry through exclusionary contracts, and the limitations on user choice resulting from its market dominance were all ultimately harmful to consumers even if they didn’t immediately result in higher prices for consumers or lower quality of the search “product.” But regardless of what standard of antitrust enforcement is applied, it’s hard for a defendant to win any case on the merits if it’s proven to have deleted or hidden evidence.
With the more difficult trial of the two having gone its way, the DOJ has a “hot hand,” and is already pressing its position. Search is the focus of the trial, but Android and Chrome (Google’s smartphone OS and browser app, respectively) are the gateways to Google’s control of almost 98% of mobile queries, so they are likely to be included in a remedy. Chrome is particularly important, as it is used for Search and Google Networks (i.e., the ad tech business). Google anticipated that third-party cookie deprecation would lead to a potential 70% revenue loss at Google Networks without a replacement. However, its proposed solution, the “Privacy Sandbox,” generates antitrust concerns of its own.
Given the search ruling, we believe that default payments (like the $20+ billion to Apple) are likely to be banned or severely curtailed, and the divestiture of Android is a high-probability shot on goal. Divestiture of Chrome is a lower-probability event, but likely will still be included in the list of the DOJ’s remedies as a negotiation tactic. In addition, there could be licensing and/or data sharing arrangements with competing search engines and even AI developers to help level the field in the search market.
Moreover, there has never been a real market for search-based ads, Google priced its inventory at its whim, and by what it thought was “fair.” As a Google executive stated in the search trial, “I would describe it less as raising prices and more coming up with better prices or more fair prices, where those new prices are higher than previous ones.” So what happens to the CPMs of Google’s ad inventory in an actual competitive market?
Another key variable to consider is Apple’s response. What does Apple do when it’s no longer paid to go along with Google, but instead is incentivized to compete for the same pool of ad revenue? Safari is roughly 30% of global search market revenue, and Apple is going to want to make up as much of that $20+ billion in default search engine payments as it can if and when those payments go away. On top of that, we think there’s plenty of demand for a competing search engine once it’s allowed to scale. Data for Progress, an American Economic Liberties Project, surveyed over 1,000 voters in September 2023 asking, “If Apple launched a search engine to compete with Google, how likely or unlikely would you be to try it?” 58% responded “Likely.”
This is no longer just us voicing concerns, as we have for almost a year now. Some people on the Street are also starting to take Google’s situation seriously. Recent notes published by some large banks have, correctly in our view, identified structural remedies such as the Chrome divestiture as a 20% hit to Google search queries and ad revenue, and Android divestiture as another 20% hit. So, Google’s Search division could be 40% smaller due to these structural remedies (notably excluding anything with Apple), or by elevated ad prices reverting to a lower norm via actual market forces. Even if only behavioral remedies get enacted, screen choices and non-exclusive capped revenue sharing default payments (fixed dollar amount, vs. percentage of revenue as with Apple) could be just as harmful to Google.
It should also be noted that with Google’s future being increasingly tied to AI advances, remedies could limit its unique access to inputs/queries while giving competitors inroads with their new AI “answer engines” – a double-whammy on an important technological growth vector.
Either way, we think the era of fines and slaps on the wrist is over. The hearings on search remedies start September 4th, and the DOJ apparently thinks it has a winning hand, as it’s openly considering breaking up the entire company. Remedies are likely to be decided by 2025; after presumably being appealed and delayed by Google, they should take effect in the next year or two. However, the market will price in these actions ahead of their impact, once there’s enough visibility.
Make no mistake, the timing of the remedies is nothing compared to their impact. They’re intended to disrupt Google’s network effect. Courts do not care about shareholder returns; they care about competition and creating a real market. But they’ve never broken up a network effect company before, so what happens when they neuter the effect? It’s too soon to tell, but that’s the DOJ’s focus, and these developments don’t sound positive to us. Of course, though, we’d be happy to pick up many smaller parts of Google at the right price after the dust settles.
Our caution today on investing in Google stems from what we see as a misapplication of precedent antitrust actions by many managers, leading to far too much confidence that a breakup or forced divesture would be a positive for the share price. The most discussed antitrust breakup analogy is the AT&T case, because both AT&T and Google were network-based businesses. While that’s partly true, we’d say that AT&T didn’t listen in to every call and monetize that data multiple times over. In addition, a smaller, post-breakup network would have less data to target consumers with, which would lower currently elevated ad prices/fees.
What about Standard Oil? Didn’t that breakup unlock plenty of value? Well, remember that Rockefeller spent most of his time obscuring each subsidiary’s earnings. When the breakup came, the market could see for the first time that the earnings potential of each business was far greater than had previously been thought. Google, while losing $4 billion per annum in its Other Bets division and saving $20+ billion in default payments, is far from the same level of using opacity and losses to cover up profits.
At a high level, we simply advise caution in comparing past precedents without any thought to the key contextual differences. Everyone likes to talk about flywheels, and “1+1=3” on the way up – but what happens when that’s undone, or going in reverse? The past may not be precedent, and that’s why we think it may be best to hedge a Google investment with positions in open internet ad tech players. After all, the ad tech trial is the one Google trial with publicly traded beneficiaries that aren’t multi-trillion dollar companies unlikely to have large changes in their earnings prospects from these trials.
Finally, in a potentially wild tail event on a tail event, the DOJ ruling has bolstered the Ohio AG’s position in its lawsuit arguing that Google should be considered a common carrier – i.e. a utility provider for the public. If this view gains traction, we imagine Google would much prefer separating into smaller, profitable parts than being mandated to maintain low ROEs like a public utility. However, while this outcome is interesting to consider, we believe it is more academic and unlikely to occur.
Ultimately, the search trial win has given the DOJ a stronger standing in the court’s eyes for remedies. A potential Chrome divestment could finally allow for industry-wide third-party cookie deprecation. That would represent a further benefit to publishers and their ad tech intermediaries, Supply Side Platforms (SSPs). So now, with the search case out of the way, let’s revisit our ad tech trial thesis. It’s changed only slightly since our original publication.
II. Ad Tech Trial – Sell-Side Focused – Starting September 9th, Ending in October
Remarkably, only two variables have changed since our original thesis: the start date of the trial, and the shift from a jury to a bench trial. The ad tech trial in the Eastern District of Virginia (EDVA) begins on September 9th. It’s mandated by the court to last no more than 6 weeks, and might take just 4 weeks. Amusingly, in early 2024 the judge opted for a later start date to avoid any summer vacation issues for herself and a potential jury.
During the summer, Google argued that the damages listed in the lawsuit could be paid off, thereby eliminating the need for a jury. The judge agreed, and Google ended up paying $2.3M to strike the jury from the ad tech trial.
Ironically, the company had requested a jury to begin with back in 2023, but when the Epic jury was not fooled by its technical jargon, it changed course. However, Google is not out of the woods yet.
The presiding EDVA Judge, Leonie Brinkema, has shown she has a firm grasp of the technical issues and little patience for any obfuscation. She also has no problem handling high-profile cases: Her CV includes presiding over CIA leak trials, a $3.5 billion fraud case, and the trial of Zacarias Moussaoui, the only person convicted in the US in connection with the 9/11 terror attacks. To point out just how tough she is, when Moussaoui’s sentence was formally issued, she coldly told him, “You came here to be a martyr and to die in a great big bang of glory, but to paraphrase the poet T. S. Eliot, instead, you will die with a whimper.” and when he tried to interrupt, she replied sternly “You will never get a chance to speak again, and that’s an appropriate and fair ending.” We think Judge Brinkema is unlikely to be cowed or confused by anything from Google’s playbook of trial shenanigans.
On that note, the company is attempting to argue that the testifying executives and former employees in the ad tech case are not experts and “lack personal knowledge” of the anticompetitive events detailed in the suit. A high-profile lawyer to digital media companies described the move “as an example of Google scrambling after its recent loss” and added that “it doesn’t seem likely they’re going to prevail.”
In addition, in late August, the DOJ asked the court to apply an “adverse inference” against Google on document destruction during a litigation hold. Basically, it wanted the judge to assume that Google hid documents in bad faith. Yet again, Google had been asked to keep all documents and communications from October 2019 to January 2023, but did not. The exhibits filed included the following near-perfect (for the DOJ at least) chat message from a Google employee:
That’s just one of many examples suggesting Google’s failure to properly preserve, if not actively destroy, internal documents and communications, stretching from regular employees all the way up to CEO Sundar Pichai.20 The judge said they collectively represented “a clear abuse of the process” and were “totally unacceptable,” adding that she would “draw inferences from that.” Score another point for the DOJ.
In keeping with its “Rocket Docket” nickname, the court expects to have all supporting documents, findings of fact, and conclusions of law filed prior to the trial, so a verdict should come soon afterwards. This process is in almost complete opposition to that of the drawn-out search trial. There, supporting documents were filed at the end, delaying concluding arguments and the verdict by many months. Additionally, as we’ve seen, Judge Brinkema shows a masterful grasp of the case and all the other Google cases across the country, so there’s a possibility she might rule from the bench after closing arguments and write her decision later. At the latest, we expect closing arguments in early October and a ruling by the end of October.
Additionally, the ad tech trial is structured differently than the search trial, and in such a way that remedies should be determined much more quickly. The search trial was split into two parts, or bifurcated: one to determine liability, and the other to determine penalties. However, in the ad tech trial, liability and penalties will be decided in a single proceeding. Therefore, we wouldn’t be surprised to see remedies announced together with the verdict in October. However, given the importance of the case, we think remedies are more likely to be announced after the verdict, but most likely before year end.
Another reason why the DOJ’s ad tech trial is likely to have remedies or a visible conclusion in the near term is that it’s one of two suits brought against Google’s ad tech division. The State AG’s ad tech antitrust trial in the Eastern District of Texas is starting in March of 2025, and it’s been far more aggressive in discovery and deposing executives. That court is also fast-moving, and the suit may possibly loop in Meta (the parent company of Facebook) and its “Jedi Blue” deal with Google, the one big ad tech market manipulation scheme that the DOJ left out of its suit (but we included in our original thought piece). If the DOJ trial goes as expected, by late 2024 Google might have yet another loss on the books. It might therefore be inclined to play ball and settle the ad tech suits quickly, rather than incurring two adverse rulings for ad tech that would all but kill its chances of winning an appeal.
On that note, a June 2024 poll in Digiday, an online trade magazine for online media, found that 70% of industry participants believe Google will cut a deal ahead of the ad tech case. While that’s a bit too aggressive given that the trial is now days away, it shows that the industry expects Google to wheel and deal with the DOJ on its ad tech business. So, while search remedies may take a while to come into effect, with Google fighting hard to hold onto its original business, we wouldn’t be surprised if the ad tech business ends up being quickly divested or used as a bargaining chip to help Google keep as much as it can of its search business.
Further, antitrust lawyers we’re in contact with have noted a conspicuous silence from Google proxies regarding the ad tech trial, suggesting that Google is more focused on search, and a divestiture of ad tech would not hurt it as much.
But while ad tech may be small for Google, it’s certainly big to its competitors, who have suffered under the company's reign.
Given the above events, we’re confident in the outcome of the ad tech trial and we believe the beneficiaries of it are still considerably mispriced – even as our thesis has slid to the right for the start of the trial from early/mid 2024 to late 2024. Moreover, remedy impacts in the ad tech trial appear likely to arrive sooner and hit harder than when we originally wrote our thesis, courtesy of a strong start in the Eastern District of Texas that may force Google’s hand after the DOJ trial. Indeed, the deletion of evidence is so egregious in all these cases that the DOJ can argue mere behavioral remedies are insufficient.
Finally, any political change of hands shouldn’t really matter much: The DOJ lawsuits were started under a Republican administration and were obviously followed through on by a Democratic administration. We think a second Democratic term would probably increase the level of antitrust enforcement,23 while a reversion back to a similar Republican administration would most likely not change the trial process or outcome. In fact, the Republican vice-presidential candidate has stated his desire to breakup Google numerous times and has even praised FTC Chair Lina Khan (who was nominated to that position by President Biden). In addition, some Republican congressmen have stated that “they plan to hold Google accountable should it try to avoid a court-ordered remedy to address its monopolistic status.”
With all that in mind, let’s update the key points of our original thesis and take another look at the trial beneficiaries (which are still massively mispriced).
The feedback on our original thesis was of two kinds: Surprise, even shock, from fund managers – but excitement and satisfaction from industry experts, a handful of antitrust lawyers, and ad tech executives thrilled that someone in the investment community finally wrote about what they’d experienced over the past decade. The difference in perception between capital market participants and industry experts was stark. Additionally, there was almost no pushback on our industry/ad tech market analysis or our interpretation of Google’s actions by industry experts. The widest range of feedback was on what remedy the court would order, with many wondering whether SSPs or DSPs would be the true beneficiaries of a Google ad tech breakup.
The DOJ is focused on remedying the competitive environment on the sell side, where Google is most dominant, just as we originally surmised. The proof is in the trial witness list: Of the open internet ad tech executives set to testify, 12 are from SSPs and exchanges (the sell side), nine are from publishers (still technically part of the sell side), and only two are from DSPs (the buy side). If your plan is to show anticompetitive harm and the necessity for remedy, you pile on the evidence to prove the point. And that’s what the DOJ appears to be doing, with sell-side witnesses outnumbering buy-siders by a factor of 10 to 1.
Google’s market share was a question mark for some investors (and even some industry practitioners) following our initial publication. Thanks to recent court filings (and to Jason Kint, CEO of DCN, who went through the documents on social media), we now know that GAM (Google Ad Manager) sees roughly 85% of all ad auctions. If GAM is the ad server of record for the internet, it’s not surprising that spending by Google’s AdX ad exchange on GAM is 55x bigger than the next competitor’s, and 10x bigger than all others combined.
We also finally have a better estimate of revenue (before TAC) of Google’s sell-side businesses AdMob, AdSense, and of course GAM, which in 2020 was approximately $15 billion. From 2020 to 2023, according to SEC filings, Google Networks grew revenues by about 35%. Applying that growth figure to the disclosed 2020 revenue (before TAC) in the court documents yields roughly $20 billion in sell-side revenue (before TAC) for Google in 2023 – not far from our original $19.3 billion estimate.
Interestingly, the same court document lets us determine Google Network’s gross margins, which previously were completely unknown. The company’s sell-side business generates an obscene ~95% gross margin, well above the 75% of other open internet SSPs. On the other end, Google’s buy-side products (DV360, and CM360) generate a ~55% gross margin, below The TradeDesk’s 75%+. The lopsided disposition of gross margins across Google’s ad tech stack reinforces our assertion that competition is almost entirely stamped out on the sell side.
However, these financial disclosures may still not be the whole truth. Conversations with industry practitioners lead us to believe that something is being left out or obscured, possibly through unconventional line item names and/or clever re-routing of TAC to under-report the ad tech business’s net revenue and absolute gross profit. This is why we discuss and analyze Google's sell-side business based on revenue before TAC, or rather what should be called net revenue before any obfuscation (as best we can determine from the outside). Google certainly has multiple motives to reveal as little as possible, whether to hide potential opportunities from competitors, appeal to the court by portraying the business as small as possible (and not as dominant over its competitors), or to create a narrative for proxies that this business is financially insignificant to Google and its shareholders. Notably, the only element of the financial disclosures that industry experts unanimously agreed upon was that the gross profit margins, which are monopolistically high on the sell side, look correct.
As for the next part of our thesis, we originally stated that the 20% take rate Google earns on each transaction was likely to come under fire if the company were forced to compete with everyone else (who collectively earn only around 10%). Recent court documents show that Google’s own employees believed that the 20% take rate “wasn’t justified by value,” and that it was instead due to Google coupling the ad tech stack with access to AdWords,30 the monopolistic search business. In other words, it was an instance of “tying,” a practice that got Microsoft in trouble back in 2000 when it integrated its Internet Explorer web browser into its Windows operating system. Microsoft argued the two products were inseparable and won on appeal, resulting in only behavioral remedies instead of a full breakup.
This outcome was partly due to procedural errors on the DOJ’s part, as well as its failure to provide sufficient legal evidence for a separation. The DOJ has learned from experience and is certainly providing a sufficient legal basis for a breakup in Google’s cases. So, what happens when Google Search and Google Networks are no longer connected? Might that action alone bring down Google’s sell-side take rate, even before we consider ad tech market remedies?
With a large revenue opportunity for competitors, credible pressure on take rates from multiple fronts, and a strong case for structural remedies to enhance competition, share loss for Google should be viewed as a real possibility – so let’s take another look at our expected beneficiaries.
We believe the biggest beneficiary will be Magnite (MGNI), with other open-internet ad tech players competing for second place. While Magnite shares are certainly mispriced on their own, the asymmetry of the opportunity when taking the trial into account is even more disproportionate in the options market. Frequently, and inappropriately, options markets assign normal probability distributions to situations that have bimodal outcomes. As such, we continue to advocate an equity position, but we believe the LEAPs offer a compelling opportunity on their own, or as a low-cost hedge to bimodal antitrust outcomes for those who already own Google.
Also informing this positioning is our experience witnessing the market’s tendency to overly discount identified risks and under-discount risks that have not been explicitly identified yet. While the outcome here is still not specifically identified, we believe our knowledge and research have distinguished a reasonable range of outcomes that skew to the negative for Google, in opposition to what is priced into the company and the potential antitrust beneficiaries.
Our estimates on the opportunity are largely unchanged, and while the share price of MGNI is almost double what it was when we originally published our thought piece, the increase has come from the core investment thesis. Simply put, MGNI is the key enabler of Connected TV advertising for streaming platforms, an increasingly necessary revenue source for their media parent companies. Magnite’s contract win with Netflix is proof of its differentiation in the space and was something we expected after hearing that Microsoft’s Xandr ad tech stack wasn’t capable of true CTV ad delivery back in early 2023. From here, we anticipate Magnite to capitalize not only on the Netflix opportunity (with impressively high incremental margins), but also on other media customers’ accelerating adoption of the same type of programmatic infrastructure/services. Meanwhile, Magnite DV+, the company’s open internet SSP division, should continue its modest growth, even without a Google antitrust shockwave.
With a strong standalone investment case, the antitrust impacts to DV+ and Magnite as a whole are completely gratis, not priced in for the equity or the LEAPs. However, the options allow for some accelerant to returns, and we believe there’s two ways to play the trial: (1) a market re-rate on the trial verdict in late October 2024 and remedy announcement in late 2024 (Jan 2025 options as in our original thesis), and (2) the financial impacts to the ad tech marketplaces when remedies get enacted or a deal is struck sometime in 2025 (Jan 2026 options).
As for us, our position remains the same: We own Magnite equity and rolled31 our Jan 2025 options into the Jan 2026 LEAPs, as the antitrust opportunity was still there. Below is an updated table from our original thought piece, with illustrative returns for the leading SSPs’ 2026 LEAPs:
As our antitrust thesis is arguably even stronger than it was before, we still see the LEAPs opportunities as a 50- to 100-bagger as a re-rate occurs in the share prices of those who’ve been under Google’s thumb for close to a decade. Visibility on trial remedies – a possible trigger for all this action – could come much sooner than most expect.
III. Summary – DOJ’s Hot Hand, Ad Tech Win Likely, Era of Fines Over
In the year or so since we first published our thoughts on Google, a remarkable set of events has transpired, almost all of which contravened pundits’ steadfast predictions that the status quo would hold. Google’s court losses thus far have been historic, and yet there’s almost certainly more to come, with the odds increasingly moving against the company. The Epic jury verdict against Google’s Play Store got the snowball rolling, the deletion of evidence has negatively impacted all the cases against the company, a strong start to the EDVA and EDTX ad tech cases has added to the pressure and losing the search trial was the coup de grâce. With the most important trial of all going the DOJ’s way, the government has a strong case for remedies that should have a real impact.
As we've mentioned, the ad tech trial is expected to begin next week and take less than a month, with a verdict following shortly afterward. Remedies are likely to be announced before the end of the year, and Google appears more willing to divest its ad tech business than its search business.
Regardless of the timing of remedies in any of the cases, once there’s clarity on the outcomes, Google’s competitive standing is likely to take a hit, and shares of beneficiaries are likely to reflect a fairer competitive environment. Nevertheless, we still believe that many fund managers are woefully underappreciating the risks to Google and the benefits that would accrue to its minuscule competitors. Those who look to previous antitrust breakups, such as those of AT&T and Standard Oil, as templates risk overlooking key contextual differences that give us pause on investing in Google now.
Google’s economic engine is powered by data that is connected to such an extent that it becomes almost singular. Antitrust court losses and subsequent remedies could result in that data pool being split apart and, in the case of Search, possibly shared with rivals. So, what does it mean for one of the greatest network-effect businesses ever created when its network shrinks or loses its exclusivity? There may be a price for a “baby Google” that we’d be willing to pay, but we’re not sure how much pain might be inflicted on Google longs before we get to that point.
The equity valuations of the ad tech trial beneficiaries still fail to reflect the growing likelihood of these outcomes, which now seem not just possible but probable. Moreover, MGNI looks like an even stronger standalone investment than last year, thanks to its Netflix win, balance sheet deleveraging, and increasing free cash flow that should fuel share buybacks. It offers a potential hedge against adverse outcomes for Google shareholders, while also presenting significant upside on its own – 5x-10x in equity and 50x-100x in LEAPs.
As always, feel free to reach out to discuss this piece, as well as other opportunities we’re seeing at Crossroads.
Sincerely,
Ryan O’Connor
Founder and Portfolio Manager
Daniel Prather, CFA
Director of Research