How to Remedy Google's Search Monopoly

Audio of this conversation is available via your favorite podcast service.

This close to the end of 2024, it’s clear that one of the most significant tech stories of the year was the outcome of the Google search antitrust case. It will also make headlines next year and beyond as the remedies phase gets worked out in the courts.

For this episode, I turned the host duties over to someone who has looked closely at this issue: Alissa Cooper, the Executive Director of the Knight-Georgetown Institute (KGI). Alissa hosted a conversation with three individuals who are following the remedies phase with an expert eye, including:

• Cristina Caffarra is a competition economist and an honorary Professor at University College London, and cofounder of the Competition Research Policy Network at CEPR (Centre for Economic Policy Research), London.
• Kate Brennan is associate director at the AI Now Institute; and
• David Dinielli is an attorney and a visiting clinical lecturer and senior research scholar at Yale Law School.

What follows is a lightly edited transcript of the conversation. Note: Tech Policy Press is a recipient of a charitable donation from DuckDuckGo, a company mentioned in this discussion.

Alissa Cooper:

Hi there, podcast listeners. I'm Alissa Cooper, the executive director of the Knight-Georgetown Institute, KGI, a new center at Georgetown University dedicated to connecting independent research with technology policy and design. I'm excited to be hosting three experts in competition policy to discuss the US government's antitrust case against Google over the company's monopoly in online search.

In August 2024, US Federal Judge Amit Mehta ruled that Google broke federal antitrust laws by monopolizing the markets for general search text advertising. Now the case has entered the remedies phase and both sides are offering their ideas for how to remedy Google's illegal conduct. In November, the Department of Justice proposed a wide-ranging set of remedies. Google will counter with its proposal by December 20th. A remedies trial is scheduled for next year before the judge issues a final remedies order in August. The outcome of this case could affect how millions of people search for information online, so all eyes will be on this case and its potential to restore competition in a highly concentrated market.

I'm really pleased to be joined today by three leading experts to discuss the potential remedies in the case and their potential to reinvigorate competition in this key market. We'll start with Cristina Caffarra. Cristina, can you introduce yourself?

Cristina Caffarra:

So I'm a competition economist that practiced most of my life in Europe, but also across the Atlantic in the US. I do no longer consult, I'm now much more involved in policy and general discussion. But it is I think incumbent to people who have been very much in the private sector to make an initial disclosure. If I may, I have been in the past working adverse to Google and math is indeed around search and around Android. I worked in the European Commission cases on this, both the search effectively case and the Android case and more. I've been assisting the attorneys general in the US in the initial phases of their investigation. So my position is being consistently on the other side of Google.

Alissa Cooper:

Thank you. I appreciate that disclosure. Next we'll turn to Kate Brennan.

Kate Brennan:

Hi everyone, my name is Kate Brennan. I am the associate director at the AI Now Institute, which is a policy and research institute shaping AI and the public interest and really specifically focusing on concentration in the AI industry.

Alissa Cooper:

And last I'll turn to David Dinielli. David, can you introduce yourself?

David Dinielli:

Sure, thanks Alissa. My name is David Dinielli. I'm an attorney and I currently am a visiting clinical lecturer and senior research scholar at the Yale Law School where I run a clinical project called the Tech Accountability and Competition Project. My background is in litigation both in of antitrust, civil rights work and movement lawyering. I also should disclose, I've co-authored a number of papers that are presumably adverse to Google and that they, for example, lay out a roadmap for an antitrust against Google, both in connection with its search business and its advertising business. And I'm a co-author, along with other members of this group today, of a paper outlining some proposed remedies in the search case.

Alissa Cooper:

Great. Thank you David. And yes, indeed, David and I have co-authored a paper about the remedies in the Google case. So David, I'm actually going to kick off our conversation with you. I would love it if you could give us some background on the judge's ruling and how exactly Google broke the law. If you could tell us what the ruling's key findings were and why the judge assessed Google's conduct as being anti-competitive .

David Dinielli:

Sure, and on the presumption that there will be some non-lawyers or non-antitrust experts who will be listening to this, I want to give a little bit of background about the antitrust laws and specifically make certain that everyone knows that being big is not itself a violation of the antitrust laws. In fact, having monopoly power is not a violation of the antitrust laws. Rather what's made illegal under section two of the Sherman Act is using anti-competitive means to acquire or maintain monopoly power.

The other background piece of information that's important in this market is a feature whereby the more searches that a company does, the better the quality of the results. And so as you get bigger, your quality gets better. That makes it more and more difficult for rivals to catch up with the market leader and more and more important for the market leader to try to keep customers away from their rivals.

So with that background, it's important then to look at how Google did violate the law. And what the court found was that simply being big wasn't enough. Simply having a good product wasn't enough. But rather what Google had done was it had taken steps to try to deprive rivals of access to customers. Specifically it reached agreements with, for example, Apple whereby it obtained by, paying quite a bit of money, the exclusive default positions at all search access points on Apple devices. Similarly, with respect to Android devices, the audience members presumably know that Google owns Android. It used a series of contractual arrangements to ensure that on all Android devices, Google search also was the exclusive default at the search access points on almost all devices.

And so what we ended up with was a situation in which for many, many years, whenever anyone got a handset and they opened it up out of the box and they tried to do a search, no matter where they tried to do it or how they tried to do it, it would be a Google search unless the customer made affirmative efforts to try to change the default to use, for example, Bing or DuckDuckGo. The court relied heavily on behavioral economics in acknowledging that these default positions, even though there's a possibility of changing them, have great power in that most people simply leave the default.

And so Google might be saying that competition is just a click away. In truth, it's more than just one click. But even if we were just a couple of clicks, the reality is that very few people change the defaults. So this network of arrangements with Apple and by virtue of its ownership of Android essentially locked up the market and Google was able to pay tremendous amounts for those positions, amounts that no rival to possibly possibly match. The result we have is that the market has been very stable for about 10 years or more with Google conducting well over 90% of the searches, at least in the United States. And the court found specifically that the efforts to maintain that monopoly through these arrangements was anti-competitive in that it wasn't simply competition on the merits, but an arrangement in order to keep others out and prevent access to the customers that those rivals might need to gain scale and quality.

Alissa Cooper:

Thank you, and I think we can give those payment figures that have been in the news. In a single year there's a $20 billion payment to Apple and the $26 billion paid overall to all of these ecosystem partners. You can tell how important these defaults were to Google and to Google's business.

Now the DOJ has proposed a set of remedies to try to restore competition in this market. Can you tell us what's in the DOJ's proposal? Give us a quick summary of what is a very comprehensive proposal, but what are the she elements?

David Dinielli:

It is comprehensive, it's 35 pages long with about a 23-page summary of it. So it's a dense read, it's an important read. But I think that some of the most important provisions are the following. First, the DOJ is proposing that there be a total prohibition on payments by Google for preferential placement or default status or pre-installment. This will be a big change in terms of the way the marketplace works. Alissa, you gave the numbers of how much Google currently pays in order to get those preferential placements. The idea here is that if Google doesn't pay, then perhaps these channel partners will go to Bing or some other rival and say, "Well, can you pay me?" And they'll be acting in their interest by getting someone else other than Google to be placed on their platform.

There also is a proposal that Google be required to divest Chrome, its browser, to a buyer to be approved by plaintiffs. The reasoning here is that Chrome, which defaults to Google search, is one of the key ways in which Google maintains a pipeline of customers to its search product.

With respect to Android, you'll recall that I mentioned before that currently Google uses Android to essentially require through a web of contracts that Google search B default on Android devices as well. The DOJ is not proposing that Android be divested immediately. Instead, it's proposing that Android be subject to a whole host of behavioral remedies. And then if it turns out that either Google doesn't comply with those remedies or those remedies aren't effective in restoring competition, then at that point Google could be required to divest Android as well.

There are other provisions including for example, a requirement that Google be required to license its index at marginal cost. It also is required to syndicate at marginal cost. Then in addition, there are some additional protections for content creators and publishers. For example, there is a provision that is going to prohibit Google from entering explicit agreements with publishers and content creators for access to their data that Google might use in training its AI models. It also is going to allow publishers and content creators to opt out of having their content used in AI content on the search result page.

Those are things that probably Kate is going to give us a little more insight into, but that indicates that the DOJ is looking at the future and not just the market as it exists. But recognizes that the role of AI is going to be important here and is trying to make certain that Google does not simply do what it did in search, only with respect to AI. There is a lot more in that document go. Why don't I stop there because I think those are the two features.

Alissa Cooper:

I want to turn to Cristina. Cristina, earlier this year you wrote an article for Tech Policy Press urging the DOJ not to request what you called a Rube Goldberg machine of remedies that tries to address all of Google Search's monopolistic behavior at once. I'm curious if you could tell us more about the remedies that you were hoping to see here or that you still are hoping to see and give us your take on the remedies package that DOJ has proposed. At a high level, how does it stack up to what you recommended?

Cristina Caffarra:

Thank you. The article which I wrote with Robin Berjon of course, known to many of you, was immediately after the opinion of Judge Mehta came out. And our immediate sense was that it was important to somehow flag that of course we are looking at an ecosystem of great complexity here which connects multiple parties, multi-sided in multiple ways, and so remedies need to approach it as such. That said, the main piece of anti-competitive conduct that was majored on during the trial, as David has also discussed, was the body of exclusive contract that secured the position of Google and Google search as default and pre-installed at every entry point in many of these devices.

So our sense was that on the one hand it would be important to certainly not think that the remedy that just addresses those exclusive contract, the big piece of the anti-competitive conduct would be sufficient. There would be possibly a temptation to pick what we call the big obvious remedy, which needs to be of course part of the picture, but we need to go beyond that and we need to do more. And at the same time the Goldberg machine analogy came because the thought there was, while we want to go beyond the one big obvious remedy, we also shouldn't be thinking about designing a complex baroque construction that tries to address every single possible piece of conduct and is interdependent in ways that is likely not to work.

So where we ended up was essentially suggesting an approach that is looking at the main control points and was focused on the main areas, in our view, of additional intervention relative to the big obvious remedy, which again, David has described. Essentially forego any form of exclusivity and payment for exclusivity that essentially restrict the ability of others to bid against you and compete.

Where in our view, these additional areas, I'll go quickly through them. We had a notion that it would be natural given possibly the experience in Europe and given that this had been discussed to think of a choice green additional piece of the remedy. And so we mentioned it, although we were not particularly enthusiastic that would be a particularly effective solution, again, given the experience in Europe, but it is a natural place for I think the judge to go to.

Another area we were concerned of course, and so is everybody, about the position of publishers in this. And so we were thinking about various ways in which one should focus on interoperability and purpose limitation in the relationship between Google and publishers' work on that.

The other area was certainly data. We spent quite a bit of time talking about data disgorgement because of course of the implications not only for competition today in search, but for the future.

We also spent quite a bit of time talking about browsers. Browsers that we described as critical infrastructure. They are an incredibly important part of the ecosystem and we were thinking that any remedy would have to deal with the relationship between search and browsers. And given the dependency of browsers out there from funding from search, how does one think about how that funding for browsers that are key pieces of infrastructure should be sold of?

And finally, we also had of course suggestions for structural divestments and we were thinking about separating, you can use the scissors in different ways, but it was very obvious that on the one hand you had to keep search separate from Chrome and Android.

But so our effort was very much not necessarily prescriptive. None of us can pretend to have the perfect formula for remedy. It was more an effort to identify key areas for intervention and what we thought was something we absolutely need to see without at the same time going too far into arcane and Rube Goldberg-ish.

So how does it stack up? I'm very favorable to the proposal that the DOJ has put forward. I'm not saying that I was pleasantly surprised because that would imply that I didn't think they would do a good job. I know that they were thinking hard about it and all of these inputs that they got from various people certainly did play in their thinking. There was a surge of suggestions. But I think it is significant and a testimony to the work that's been done that I think the areas for intervention that they call out in their proposal are very much mapping into the kind of areas, a real solution or attempts a solution that I called out because of course the big obvious remedy is there. There is no question that something needs to be done about the contracts that are exclusive.

And what I also like about the way in which it is formulated by the Department of Justice is that the actual formulation of it is that there should be a prevention prohibition to Google from providing third parties something of value in exchange for this exclusivity. This is a clever formulation because one of the concerns that one could have was okay, you can stop them from making large upfront payments, but there's a million other ways in which value can be shared between two parties which do not take the form of an upfront large payment. So the formulation of preventing them from providing third parties with something of value is quite expensive and it seems to me should be capable of covering various ways in which this rent sharing could be ultimately structured. So there is this big piece of course.

And then as David said, there were multiple other subclauses preventing exclusive contract with content publishers and so on. The second big piece of it is divestment effectively, which comes under an umbrella, which is broader preventing ownership and control that enables self-preferencing. And that covers both the divestment of Chrome and the potential contingent divestment of Android if the behavioral remedies don't work. I think it's significant that Android behavioral remedy, there is also the mention of a choice screen, which is something that I had mentioned earlier.

Then there is the third piece, which is both the syndication of search index and the whole piece around data access, which is very much about the future. And if I may share, I know that from a discussion I had personally with Jonathan Kanter, he personally felt that this is almost the most innovative and significant part of the remedy piece for the future.

So I'll stop there, but to me this proposal is very comprehensive. It maps well into, not that it should, but from my perspective, into the kind of areas that Robin and I had identified as being important. And it does it economically and very precisely, so I think it is a very good proposal.

Alissa Cooper:

Yeah, appreciate that mapping. At KGI, the center that I run, we had also put out a report close in time with when the DOJ remedies were proposed and our top line was try to make it comprehensive. That you really need to attack this from multiple different vantage points. And so had a very similar reaction, that the comprehensiveness both in terms of which elements were included, and as you say, the formulation of some of the provisions, really you could tell that they put a lot of thought into trying to cover all the bases. Kate, I want to bring you in on this. Would love to hear your high level reactions to the DOJ's remedy proposals. We'll talk specifically about AI in a moment, but curious for your observations about what they went ahead and did with their proposal.

Kate Brennan:

So for a bit of context, I'm coming at this case thinking about how Google search monopoly has created unfair advantages in the generative AI market. So for me, effective remedies have to tackle Google's emerging dominance in that market more broadly. And so we'll talk about specifics later, but upfront I think it's incredibly important to drive home that Google is at where they are right now, as a massive AI power player, because of the market and data advantages that it gained through its illegally maintained search monopoly, full stop.

So there are quite a few remedies that address this dynamic, but I'll call out the big one. The DOJ proposed that Google has to divest their stake in rival AI companies who are building competitive search engine technologies or AI products and is prohibited from investing in those companies. So a clear cut example of this is Google's investment of $2 billion in Anthropic, which is one of two companies who are building a viable competitor to Google.

So at a high level I was thrilled to see this, it's bold, it goes directly to the heart of Google's structural advantages in AI. And I am bringing some caution into this case writ large, especially because there is an undercurrent of the memo. Just assuming that AI can be useful to generate competition in search and we know AI is not going to be a magical fix for competition in search markets, it's not going to be a magical fix for every problem in our economy. And so ending the structural advantages that big tech firms like Google holds in the AI market are going to require structural interventions way beyond the scope of search competition.

Alissa Cooper:

I can't wait to explore this a little bit more with you. I've stared at that definition of AI products in this proposal at great length and really, really curious to unpack that with you.

But before we go there, I definitely want to get reactions from David as well. As we had mentioned, David, you and I and some others co-authored a paper where we proposed, if I do say so myself, some innovative remedies related to revenue sharing and also had a pretty intense focus on Android and the role of Android in Google's search monopoly. So a different set than what Cristina and Kate have talked about, and I'm curious, when you look at the DOJ's proposal in light of that work, what do you think? How did they do?

David Dinielli:

I'll start with my reaction to the total prohibition on payments for preferential placement and fall status pre-installment. One of our concerns was simply looking at the economics of this market. If Google is currently paying its channel partners $26 billion in a single year for this preferential treatment, what happens when it's prohibited from making those payments? So one result will be perhaps Google simply pockets $26 billion, its share price goes lay up, and any pass-through that might have led to lower handset prices, for example, the siphon will be cut off. What'll that mean for handset prices? It seems as if it's fairly logical that handset prices might go up. And I'm concerned about public reactions to antitrust cases is what we end up seeing is Google with a windfall of $26 billion in a year and handset prices going up. That won't necessarily be the recipe for continued support from a populist standpoint, from an economic justice standpoint or from a democracy standpoint towards enhanced or continued enhanced enforcement of the antitrust laws.

Our proposal was innovative and it was not adopted, but we tried to get the baby on this by saying that perhaps Google should be able to pay for only about half of the searches. And we had various ways that that could be figured out. But that after Google was paying for half, that the channel partners would then have an incentive to try to find someone else because they can't get paid. For example, if all Verizon handsets had Google, it would get paid for half of the customers or half of the users or half of the searches, but then it simply wouldn't be paid for the rest.

So the economic incentive under our proposal for a Verizon, for example, would be to find some other search partner to service those other customers so they would be making some money. In our view, this would then give different search engines, say it Bing, or someone else if they were to come along, to get a foothold and start to build both brand recognition, build scale and thus quality, and at the same time blunt the possible price effects on handsets that we think are likely to happen if Google is prohibited from making any payments whatsoever.

So that was an interesting, I think way to balance what we saw as competing interests with the idea that the partners would then have an incentive to act like agents for other search partners and try to incentivize their own customers to adopt something other than Google. And this would be a way to harness economic principles to try to get other search engines in front of other people and get them used to using it. When we had a conversation, Alissa, you'll remember it, in Washington DC, there were some people who were very concerned that nothing could overcome the brand loyalty that people have for Google. And in fact, just this morning I was listening to public radio and heard it public, I know whose [inaudible 00:25:22], Google was sponsoring public radio, some show, and of course they had some bit on to make everyone excited about the most important searches of the year, right? So they're obviously already trying to counter any effects that might come about as a result of the remedies are. So that was our proposal with respect to that.

Alissa Cooper:

So Cristina, from what you said, I would love to get your take on this. This has been quite a hot button issue in terms of the revenue sharing. The DOJ proposal is clear that, as you say, nothing of value can be exchanged in order for Google to gain a preference with these partners and that leaves tens of billions of dollars that Google was spending. Now they get to keep it. What do you make of that? Was that the right choice or is there a better way to think about how these dollars could flow and how that might have an effect on the future of competition in search?

Cristina Caffarra:

That is precisely why I think further [inaudible 00:26:22] needs to go into how one ultimately formulates this specifically, for two reasons. One is because my skepticism to this proposal, which is in part addressed by this kind of formulation of something of value, is that the touch point between these companies, these commercial partners are so many that it is possible to make up these payments in some other form. It isn't just a simple question of I'll write you a check for 20 billion. There are multiple ways in which these companies encounter during multiple activities. And my original skepticism was really very much about, okay, but how are you going to even make sure that this kind of exchange or the sharing of rent is really stopped?

Because in fact, let's face it, the judge can impose something on Google, can tell Google not to make a payment, but it cannot tell Apple what to do as part of this judgment or this opinion. So if Apple, just for argument's sake, was to say, "I hear it but I still think that this is the best engine there is around and I will place it in a prominent position because I don't think anything else is good enough. I think everything else sucks." Then I don't think there's much that can be done in that sense.

So the question, and I like very much the shape of the remedy that you and David came up with because if one is trying to incentivize the use of other channels, then one needs to just do so by proactively trying to push. Perhaps your solution is a clever one, to just say you can take payment for half the searches, but for the rest, if you want payment, you need to go somewhere else. There is much to, I think reflect on this. It is the obvious elephant in the room. That's why I call it the big obvious remedy. And I think a judge will obviously be very much induced to go in that direction. But whether it will actually generate or resuscitate competition in and of its own? Absolutely not. And there's a lot of thinking that needs to go further into how to really calibrate the details of it.

Alissa Cooper:

So maybe a little more nuance needed for thinking about the impact of the whole ecosystem of these contractual provisions?

Why don't we talk about the divestiture pieces. Also quite the hot button issue. So DOJ is proposing a divestiture of Chrome and the potential future divestiture of Android. Today a significant amount of browser and web technology development is funded through search and search advertising, which means that it's funded in large part by Google itself, in Chrome and through these contracts we've been discussing that are at the center of this case. David, I'm wondering if you think the evidentiary record supports the DOJ's proposal of a Chrome divestiture? And if the Chrome divestiture does come to pass, how do you see the competitive landscape shaking out in both the search and the browser markets, which are clearly deeply intertwined?

David Dinielli:

I'll respond maybe by deflecting a little bit. I think that the evidentiary record absolutely supported the divestiture of Chrome. I think it also supported the divestiture of Android and the Play Store. And I was surprised frankly to see that these entities that all seemed critical to the ways in which Google has maintained its monopoly were not treated in the same way. I do think that there is a case to be made for divestiture of Android. It's not clear to me who the best buyer would be. It's not clear to me who the DOJ would approve as the buyer and what conditions might be placed on that purchase.

It seems to me that Android would have been the more obvious entity to say ought to be divested. The companies that use the Android operating system rely on Google for everything. They're utterly dependent. And to Cristina's points that there are so many ways in which these business relationships can be adjusted and fine-tuned in terms of pressure, in terms of payments, in terms of value. Using someone else's operating system and all that entails is incredibly complex and seems to be ripe with ways for Google, if it continues to own Android, to put pressure to advantage itself, to punish bad behavior to prevent people from doing anything that would disadvantage Google search in ways that would be very difficult to police given the complexities of its arrangements.

So I would have been more expecting to see a single requirement that there be a divestiture of Android. Chrome might have come along with it or not, the Play Store might have come along with it or not. But as it is, I don't know if there's simply conservatism or a concern that the court is unlikely to require such a significant divestiture that we've come to know and be accustomed to as such critical part of this marketplace. But Android seems obvious. Chrome seems less obvious, but certainly well within the bounds of what the evidence showed. And I'll leave it at that and have others comment on that.

Alissa Cooper:

So I guess if they do go forward with these behavioral remedies for Android, you're going to be watching those closely to see if it's actually [inaudible 00:32:08].

David Dinielli:

I'll be watching closely, but I think the difficulty will be it will be very difficult for anyone to be watching closely. I'm not going to be in the meetings, I'm not going to be in the financial reviews that the companies are doing. I won't be looking at whether there are performance issues that could be imposed that might be under the radar that I would never know about and that even a real regulator might not even know about. So I'm just very concerned about the ways in which Google will continue to exercise the threat of retribution through Android if it continues to own it.

Alissa Cooper:

Cristina, give us your thoughts. I think the combined potential impact of the prohibition on revenue sharing in the Chrome divestiture has some folks worried in the web world, the browser community. What's your take on these divestiture proposals

Cristina Caffarra:

Close to David's position in that I would've thought, perhaps again, coming from a European experience in which we focus very strongly on Android, we didn't solve the problem and we in fact failed spectacularly to make any headway on that. But the focus of the investigation was very much on Android and the criticality of Android in the ecosystem. I think there was an expectation that Android was going to be very much at the front of any kind of divestment. But of course the case for Chrome is also there. We are talking about two critical assets that are essential to entrenching the moat and the position by making self-preferencing so easy for Google. So these are two ways and they're mutually reinforcing in which these conducts lead us to where we are.

The divestment is perceived of course as the ultimate sanction, the ultimate solution because we have developed a belief that behavioral remedies are very difficult to police and they are very easy to game. And that I think is difficult to counter. That said, even a divestment, who is it going to go to? Who is the buyer? How is that structure going to be designed? And how many years will it take before we get there is I think the question that should be obviously in front of us. We're not talking about something that's going to happen at the end of 2025. We're talking about something that maybe, with wind behind, is going to happen in a few years after multiple appeals.

So it's a bit hypothetical, but it is of course, as many of us have said, the significance of it is almost symbolic, that in the space of five years, and let me say this, as someone who has been an observer of the journey of antitrust in the US in the last 20 years, it's remarkable that this has come to pass. You never six years ago would've told me that in five years we're going to face a decision, in fact a proposal by the DOJ to break up Google. This was unthinkable in 2019. So this is the significance of it.

Alissa Cooper:

And a very useful piece of public education you just did about the timelines that we're talking about here. I do think that's not understood by the general public, that through the process of appeals, it's likely that some of these remedies, even if approved by the judge initially, might take years to come into effect.

Let's come back to the AI topic. I want to bring you back in, Kate. The elephant in the room has already been mentioned, but you actually wrote a piece for a Tech Policy Press where you talked about the elephant in the room in the Google search case being GenAI, and you've already I think set up well our audience to understand that there's both an implication for search itself when it comes to generative AI. Would love to hear you give us an explainer on that, how search and GenAI are related, but also why it's important for the case to address GenAI, both in the context of search and also how Google's conduct outside the context of search when it comes to GenAI investments and development can be and should be implicated in this case.

Kate Brennan:

First and foremost, it's always nice to clarify what we mean when we're talking about AI since it's a very effective marketing term that big tech companies like Google have repackaged a whole host of technologies that have been around for decades. In the context of this case, I'm talking about the more recent development of large scale large language models, LLMs. They're trained on massive data sets to predict or generate text. So in Google's case, the best example is Google's own LLM, Gemini. So when you search, instead of an algorithm serving you a third party web page, you get Google's own generated text.

So the first point in teasing through the interrelatedness of these things is worth repeating what I said earlier, is that Google simply could not have built Gemini or achieve its dominance without the search monopoly. So first is the data component. LLMs like Gemini require massive amounts of data and Google has access to this. They have access to user real-time data from search, they have access to data from Chrome, they have access to data from Android. And this isn't a secret. Google is proud to tout around that they use this data to build their models.

But building generative AI goes beyond data. They're extraordinarily expensive and resource intensive to build and run. So running large scale AI models requires massive amounts of compute infrastructure, which is cloud, access to data centers on the ground, access to chips. And Google could really only do this by moving considerable capital from its search monopoly into this AI infrastructure. Google owns Google Cloud, right? Google owns their own data centers. Google has even built their own chips. So Google has vertically integrated the entire AI stack and who can compete with that? So the first really important point to drive home is that entire stack is built on the back of Monopoly.

But the second important connection to tease through is that AI is changing search. So even Google is seeing AI as the future of its search business. I think just a few days ago, Ruth Porat, who is Google's chief investment officer, was on stage saying that their biggest bet for AI is search. So Google knows these markets are related and so the court has to too. And you can enforce the most effective search remedies, but if Google rebrands their entire search business as AI and pulls it into a separate app and walks away, we have to be ahead of that.

Alissa Cooper:

So let's come back to the remedies. You had already teed up for us. The piece around an investment in AI, and as I had noted, this proposal uses a definition of AI products, which to my eye is quite broad. And so I'm interested in your take on the far-reaching potential effects of that. But as David had noted when talking about the proposal itself, there's also a number of other publisher and data related remedies that are proposed here that feel very central to the future of the AI conversation, including allowing publishers to opt out of having their content used for training for AI models, prohibiting exclusionary agreements between Google and publishers that restrict publishers from making their data available to other competitors, including competing search engines or competing AI products. So walk us through your take on these proposals, what you were hoping for versus what you've seen from the DOJ more specifically?

Kate Brennan:

Just to circle back on the point about the divestment and the acquisitions and to your point about the definition on AI products being very broad. It is, it's so broad. Google has done a lot of really fancy and tricky and clever, which I think is the word you've used, Cristina, way to avoid kind of acquisition and merger scrutiny. In terms of investment in AI companies. Google has made direct investments like its investment with Anthropic. It has made really clever talent poaching acquisitions by basically hiring everyone but the company. Cristina, you've written about how investments evade regulatory thresholds. And so by requiring that Google has to divest from any company that is essentially developing an AI product, that is really quite broad, it does not limit it to just AI companies that are developing AI search related products. So that's fantastic for addressing the broader market concentration problem in AI.

In terms of some of the behavioral changes that we've talked about, so the data sharing requirements, the enabling publisher opt out, banning the exclusionary agreements, these are great and really important pieces. One of the elements around exclusive contracting is that Google would not be able to exclusively access AI training data. And to give a super tangible example of that, earlier this year Google announced a $60 million annual deal with Reddit essentially contracting for exclusive access to its training data. And anyone who Googles knows the meme that you attach Reddit to the end of your Google search to get the best results. This is the most valuable search data. So the fact that Google is shutting out any potential competitor from this super valuable resource, it's a meaningful thing to block that.

But I also want to make a very important point about how AI relates to the future of search competition. So there is this current throughout the memo and the remedy framework that generative AI might be good for the future of search. So I think that the memo says something like AI may present an opportunity for fresh competition, and I'm very, very wary of this. I'm wary of propositions that generative AI is going to fix competition. And so the argument goes, ChatGPT is going to come and end Google's search dominance.

First, and maybe the less compelling argument is that it overlooks how dominant Google's position is. So chat GPT has come out and Google's market position has only improved. But I think the larger point to make is that the generative AI market largely is dominated by three companies, Google, Amazon, Microsoft, and every single generative AI startup is dependent on one of these three companies. Google has come out and said proudly that 70% of generative AI companies rely on Google for cloud. 70%. So whether it's companies trying to access cloud resources or literally license their LLMs, they rely on Google. So we should really be asking ourselves why we would want to replace one monopoly market problem with another.

And so yeah, I know the focus on the remedies on this case importantly we're tailored to search competition and not perfect competition in the AI market, but this paints a really important picture for how companies like Google are using their own advantage to achieve dominance more broadly. And I'm excited that the DOJ is paying attention to all of those elements.

Alissa Cooper:

I think all of us who exist in this technology space often feel like we're just in a land of giants, you move from adjacent market to adjacent market, and yet there you are again with the same slightly reconfigured set of players. Cristina, oh, you're shaking your head at me, so [inaudible 00:43:39].

Cristina Caffarra:

Well, no, not at you. Not at you. I like what I hear. I'm shaking my head at what actually is the broader sort of landscape. I was on an OECD panel today talking about AI and it was extremely difficult to make these arguments against a general consensus that in fact it's all different, that somehow what we've seen as the previous playbook will not play out here. The challenger is just born this year. All of these kind of very optimistic narratives that are suffocating us in every direction.

Going back to the remedy and then I want to open up again, like Kate, I think is very encouraging and very good that the department is paying a lot of attention to this and is putting on the table remedies that seek to really draw the judge's attention to this. And I think encouragingly the judges quite switched on to this. In his judgment, in his opinion, he does go on about the future in the AI market. So I think some of these suggestions may find fertile ground there.

And I think however, particularly where there is a discussion where there is a suggestion around, for example, the opportunity which is being given to publishers to opt out, for example, from having their content crawled for the index or training a large model, this opt out remedy's something that feels real and feels achievable in some sense. And I think it is something where the judge may well go.

And I think it's a sentiment we all share, Kate, it's fascinating that the department has entertained the possibility of divesting parent interest or relinquishing agreements. I just don't see that happening given the climate we are in, given the fact that Google isn't the only one doing this. I mean the reason we are where we are is because this is a pervasive practice in the industry. It is not just Google and they would have a strong argument to say, "But why just me? Look at all the others. They're doing exactly the same. They're teaming up, they're making agreements, they're entering into this kind of provisions with third parties, which absolutely is a ways of circumventing the merger laws, the requirements around acquisitions." I think it would be much harder there to argue that Google, on the back of this particular case, which is about its behavior in search, really needs to divest alone this interest or to walk away from this interest. They'd have a case that then, what about Microsoft? What about Amazon? Everybody's exactly doing the same.

And the fact that the general discourse, as I was seeing today, is still so far away from considering this problematic. It is quite scary to see how people who are academics, I was on a panel of academics, they were all saying, "Data is not an advantage. Machine learning costs are declining." All of these tech optimism that we know we are surrounded by, I am skeptical that kind of piece of remedy will be taken up. But it is significant that it is mentioned.

Alissa Cooper:

That's a great segue actually to what will be my final question for all of you, which is about the change of political administration that we are about to embark on in the US. Obviously we have a new president elect who is coming in and bringing in a new team at the DOJ and the rest of the federal agencies, many of whom are a little, I think techno-optimist about AI at least. But of course this case was started under the first Trump administration and then the Biden administration DOJ carried it forward and now we'll have new folks coming in. So if you look into your crystal ball, I'm curious to hear what do you expect from the changing of the political winds vis-a-vis this case and antitrust enforcement more broadly? And I will start with you, Kate.

Kate Brennan:

I'll speak a little bit to antitrust more broadly. We're seeing a contradiction. I think we're seeing a little bit of a positioning of excitement about strong enforcement actions against big tech companies, and at the same time we are seeing massive AI boosterism across the board. We're seeing an AI czar who was just named. We're seeing an all above the ground energy approach to accelerating AI energy usage. So what I really think is going to happen is that maybe we're going to retain some of the corporate power arguments, but we're going to be missing anything that is taking meaningful action against AI companies and anything that can slow down AI growth since slowing down AI growth is antithetical to the administration's policies.

Alissa Cooper:

Cristina?

Cristina Caffarra:

I'm a European and I reside in Europe, so with that provision, I'm an observer to what happens in the US. But my sentiment having been nonetheless quite a close observer and involved in some way is that yes, very much as Kate says there is part of the world out there, Wall Street in particular, which expects that this will be an absolute return to the past. We are going to see lots of deals. We're going to see, and I hear much of that, lots of deals coming through, deals that are being held back. So that is something which is being expected.

On tech, there is no question that I think the current nominees, those we know about, both Department of Justice and the Federal Trade Commission, are far less hawkish on returning to the past and really very much have a sense for big tech remaining a problem. What will happen to this case? The Google case is unclear. We are very advanced in this case, right? There is a judgment that is due next year. There is another one on AppTech which is very imminent. We have the Texas case, which it's very significant. The case is starting in Texas in three months.

So it's not clear to me how the current political change in climate will affect this particular case, but the question remains for the other cases that are in the pipeline, what will happen to the Apple case? What will happen to the Amazon case? There is a preference that this administration appears to have for little tech. We hear it all the time, which is very much this narrative that open source is all we need when we know that OpenAI is certainly not a complete solution.

But perhaps on an optimist, at least for me being a progressive when it comes to antitrust, that on an optimist note, I was in DC a couple of weeks ago and I was speaking to a senior Republican thought leader in antitrust. And what this person told me was, which I like to repeat because I think it's a beautiful analogy, they said, "Lena Kahn and Jonathan Kanter threw the sword all the way to the other end of the field. We are not going to walk it all the way back. We are going to take it up halfway through the field and that will be the new normal." So maybe that was early day enthusiasm and so on, but I have a sense that we're not returning to the time of President Bush by any means or shape or form. That's my two cents seen from London as it is.

Alissa Cooper:

David, you get the last word.

David Dinielli:

Thank you. And Cristina, I had not heard that analogy. It's an interesting one. I have a concern that although there is at a very high level bipartisan support for continued enforcement of the antitrust laws, that there could be enforcement with respect to favored or disfavored companies at varying levels based on, for example, how those companies are treating Elon Musk. Or who knows what? We've seen a lot of people showing up at Mar-a-Lago. Zuckerberg showed up at Mar-a-Lago within the past couple of weeks. And so I agreed that it will not be walked back entirely, although we'll have to wait and see for whom it has walked back all the way and for whom it has walked back not at all. I would hope that there's convenient even-handed enforcement, but we'll see.

I also will observe that in the last four years, it seems as if there's been at least a dual, if not multiple basis for antitrust enforcement. Antitrust and consumer protection often go hand in hand, and I think that consumer protection will no longer be, at least publicly, a motivating force. Many of us who have thought about these markets for a while have thought about, for example, problems with the dangers of, for example, social media and thought that one remedy is antitrust enforcement because then there will be more options and perhaps some of the companies will offer safer products. I think that'll no longer be the principle for enforcement. I think that the principle will be more purely economic or political and we'll have to watch and see how that plays out.

Cristina Caffarra:

Can I add one 30-second thing, because David is right to mention it? Of course it is the case that we are all aware and concerned about the more transactional nature of the possible enforcement that we are going to see. Let me say one last thing from my perspective from Europe, which is I also think that the uncertainty that this is creating is reverberating very strongly in Europe at the moment. The uncertainty in Europe is so complete, so extreme that you see a complete paralysis in enforcement, in particular on regulation, which is in suspended animation entirely until we see the white of the eyes.

And the reality is that, as we know, the bargaining chips that are being held are so many relative to Europe. If President Trump was to say one of the companies that currently under investigation in Europe that are currently going through the DMA, for example, "Do your worst. I have your back. And I will tell these Europeans not to discriminate in favor of these wonderful American companies." It is not clear to me what Europe can do. We can jump up and down and say, "We have the law. We need to apply the law." What exactly are we going to do? When you mention NATO, defense, are we going to send the troops to Mar-a-Lago? Let me just leave it there.

Alissa Cooper:

We have much that we will learn in 2025 both about this case as the events unfold and also about how these political dynamics play out domestically and internationally as everyone navigates the new world that we will be living in. I want to thank all three of you so much for a truly fascinating discussion and I can't wait to continue it online in Tech Policy Press and all the other places where you brilliant folks reside. So thank you very much.

Cristina Caffarra:

Thank you.

David Dinielli:

Thank you. Bye.

Kate Brennan:

Thank you, Alissa.

Related Reading

• Considerations for Effective Search Competition Remedies
• “Google is a Monopolist” – Wrong and Right Ways to Think About Remedies
• The Elephant in the Room in the Google Search Case: Generative AI
• Judicial Remedies To Restore Lost Competition in the Market for General Search