Would breaking up Google really do any good?

We need to consider consumer welfare, but also technological progress and national security.

Photo by Maarten van den Heuvel on Unsplash

The Department of Justice has just unleashed a sweeping lawsuit against the internet giant, alleging uncompetitive practices and demanding that Google’s ad tech platform be spun off into a separate company. In general, I’ve been a supporter of increased antitrust action in the U.S. When it comes to the frequent calls to break up Big Tech, though, I’ve always been pretty ambivalent.

For years, the harshest criticism, and the most strident calls for a breakup, focused on Facebook/Meta. It was always questionable whether forcing Meta to, say, spin off Instagram would lead to anything on the internet being meaningfully cheaper (or workers paid better, etc.). And in the past year or two, Meta has looked anything like a monopoly — losing market share to TikTok, having its platform revenue appropriated by Apple, and seeing its stock plummet by almost two-thirds. Two months ago I wrote a post arguing that Apple, rather than Facebook, had the most market power in the world of social media.

Did Big Tech antitrust pick the wrong target?

Why did antitrust action focus on Facebook instead of Apple for years? One reason might be that antitrust lawyers and economists don’t have a great grasp of where pricing power comes from in the tech ecosystem. But I also suspect that antitrust, in general, is about political power as much as it is about economics.

The state wants to assert its primacy over private companies, and people want the state to do this, because it lets them feel like companies still answer to the people. So when a company gets very big and profitable and seems powerful, there is a tendency to want to smack it down. So it probably was with Microsoft in the 90s and early 00s, when people thought — wrongly, it turned out — that control of desktop operating systems made that company an all-conquering empire. And so it probably was with Facebook, which some people blamed for the outcome of the 2016 presidential election. The economic harm caused by these companies’ (ultimately ephemeral) market dominance seemed less relevant than the general impression that they were growing too big for their britches.

Given these past experiences, I naturally start from a place of skepticism when thinking about the new antitrust action against Google. Maybe I’m biased by my econ training, but I think that antitrust should be first and foremost about consumer and worker welfare, rather than about institutional power struggles. But unlike in the case of Facebook, there are some pretty clear reasons to think that Google’s practices in the digital ad market might be restraining competition, and that curbing those practices might result in a cheaper, fairer internet.

The real question is whether a breakup is the right solution. And when making that decision, we should think not just about consumer welfare, but also about Google’s importance for scientific progress and national security.

What a breakup of Google might hope to achieve

The basic complaint against Google is that it controls too many pieces of the digital ad market, unfairly raising the prices that advertisers pay and reducing the prices that websites receive. So the idea is to make internet advertising cheaper for advertisers and more lucrative for websites by forcing Google to spin off parts of its business.

Basically, in order for an ad to go on a website, three things happen:

1. The website uses a tool to sell ad space.

2. The advertiser uses a tool to buy ad space.

3. The website’s tool and the advertiser’s tool come to an agreement about price, using an online exchange.

Currently, Google has a major presence in all of these three pieces of the process. It owns Doubleclick, the tool that almost all websites use to sell ad space. It owns several tools for advertisers to buy ads. And it also owns an ad exchange where buyer the buyer and seller tools come to an agreement. Here’s the DOJ’s diagram:

Source: Department of Justice

The DOJ complaint alleges that Google uses its control of each of these segments to bolster its control of the other segments. Basically, the DOJ alleges that Google will restrict your access to the ad exchange if you don’t use Google’s other ad tools, and that it’ll restrict access to the ad tools if you don’t go through Google’s ad exchange. That seems like the main complaint, along with the acquisition of the tools in the first place. (There’s also an allegation that the ad exchange results are manipulated, but I can’t say much about that one.)

I am not a legal expert, so I can’t speak to the legality of these various practices; that’s for the courts to decide. And I’m not a tech expert, so I can’t say how easy it is to use alternative tools and/or alternative exchanges. I will say that while the >90% market share for the Publisher Ad Server looks especially egregious, the ~50% market share for the ad exchange doesn’t look as dominant as you might expect. If Publisher Ad Server has most of the publisher-side market, but only 50% of ads trade through Google’s exchange, it suggests that publishers have options to trade elsewhere.

But at the end of the day, what matters isn’t market share, it’s market power — the ability to extract unearned rents from a market. And Google certainly extracts a lot of value from this ad market — about 30% of all the dollars that pass through the company’s ecosystem of products get pocketed by Google. That’s a pretty fat cut — equivalent to the fee that Apple extracts from in-app purchases through its App Store.

Why would advertisers and websites together pay 30% of ad dollars to Google? There are two possible reasons. Possibility 1 is that Google’s tools and exchange are just really high-quality, so the 30% is a fee for services rendered. Possibility 2 is that Google’s alleged anticompetitive practices make it difficult for companies to use alternate tools and/or exchanges, so that the 30% is a nuisance fee for a nuisance Google itself created.

To the degree that it’s the latter, a corporate breakup would shrink those rents without compromising quality, allowing more dollars to flow from the same advertisers directly to the same website publishers without a middleman taking a big fat cut. If Google spun out its ad exchange and/or its publisher ad tools and/or its advertiser tools, then other companies might develop more exchanges and more tools, and these spaces might become more competitive, reducing those middleman fees. That, in short, is the hope of the DOJ’s lawsuit.

Non-tech businesses and their shareholders would clearly benefit if online ads got cheaper. Websites tend to be smaller businesses, and a majority of advertising dollars probably come from companies outside the tech sector, meaning that profits would get distributed from the big to the small and from new industries to old. Altogether that could produce a broader distribution of capital income — the business world would catch a break from being eaten by Big Software. Hopefully, this would also lead to lower consumer prices and higher wages, because companies would save money on ads and pass along some of those savings to their customers and workers.

This idea is not clearly preposterous. Compare the idea of forcing Google to spin off some of its ad businesses to the idea of forcing Meta to spin off Instagram.

In the case of Meta, there was a network effect at work — social media tends to consolidate into a few platforms because people want to be on the same network as their friends and family and colleagues, so breaking up these networks is a temporary measure. And it was always highly dubious that making Facebook compete against Instagram for ad dollars would have a meaningful effect on the price of online ads.

But in Google’s case, there’s not obviously a network effect at work. Yes, buyers and sellers want to trade on the same exchange, but with modern technology it’s very easy to trade on multiple exchanges — stocks do this all the time. And if ad exchanges had a strong network effect, then Google’s would probably have significantly more than a 50% share of the market to begin with. And tools for ad buyers and sellers don’t clearly have a network effect either. So it doesn’t seem likely that a breakup would just result in the web ad market coalescing again in the hands of a single player and leaving us right back where we started.

Thus, if the courts find that Google was anticompetitive — and that is still very much an open question — then a breakup could yield real economic dividends. But that doesn’t mean a breakup is the optimal strategy. Instead, the courts could simply order Google to cease certain of its business practices, like they did with Microsoft in 2001.

Does America need a smaller Google, or a smarter Google?

In the event of a court ruling against Google, forcing the company to modify its business practices might simply be more legally expedient than ordering a breakup. But there are also some other considerations that might take into account.

A big one is national security. Google is very dominant in the field of AI research, publishing more top-tier papers than any university:

Source: MacroPolo

The U.S. is currently locked in a struggle with China for dominance of the emerging AI field. It’s very hard to tell, but currently the competition looks roughly neck-and-neck. AI dominance will likely mean dominance in the precision weaponry that determine success in modern warfare. It could also mean dominance in cyberwarfare, intelligence gathering, and other areas. So it’s very important that the U.S. stay ahead in AI research, and currently Google is probably the single most important engine of excellence in U.S. AI research. This is, of course, on top of the benefits to science itself from efforts such as DeepMind’s AlphaFold.

Would breaking up Google help or hurt these AI efforts? There are two relevant historical examples to consider. First is the case of Bell Telephone Company, the government-sanctioned telecom monopoly that the U.S. government forced to break up in the 1980s. The breakup effectively put an end to Bell Labs, the corporate laboratory that was responsible for so many breakthrough innovations in the mid 20th century:

The dream of bringing back Bell Labs

Bell’s telecom monopoly subsidized Bell Labs, and its organizational stability allowed researchers to embark on long-term risky speculative projects. So once those profits and that stability were gone, Bell Labs couldn’t survive. The stream of innovations that had come out of the labs — which had been freely licensed at low prices, due to a condition of the 1956 deal between Bell and the U.S. government that had allowed Bell to retain its monopoly — dried up. Eventually, U.S. telecom innovation fell behind the Chinese state-supported giant Huawei.

In a similar way, breaking up Google might result in moderately cheaper internet ads, at the expense of U.S. leadership in one of the most crucial technologies of the 21st century. That would not be a great trade, in my view.

On the other hand, we should also consider the historical example of Intel. As Chris Miller explains in his excellent book Chip Wars, Intel’s comfy dominance of the lucrative server microprocessor market caused it to ignore developing opportunities in low-power chips, foundry manufacturing, and GPUs for AI. As a result, America’s once-great semiconductor champion now looks like a company in steep decline, leaving the country more dependent on companies like TSMC that are vulnerable to Chinese blockade.

It’s possible that Google is sinking into a similar inertia. Even as other companies like Amazon and Microsoft have diversified aggressively into cloud computing and other business lines, Google’s revenues remain overwhelmingly dependent on the digital ad market. Whereas Microsoft is aggressively working to develop its partner OpenAI into a viable business, Google’s AI efforts remain fairly academic. Breaking off Google’s ad business might force the company to invest more aggressively in alternative business models and sources of profit, including AI.

So it would be nice to preserve Google’s research efforts while also forcing it to treat that research as a top priority rather than as a fun side project. This suggests that the optimal outcome is one where Google retains the ability to make big R&D investments but gains added incentive to do so.

And as I see it, the best historical example here is Microsoft. The Justice Department initially wanted to break up Microsoft, but ultimately settled for clipping the company’s wings, forcing it to extract fewer profits from its desktop operating system monopoly. In the years after the ruling, Microsoft seemed like it had been tipped into decline, fumbling its attempts to enter the web search and mp3 player markets. But in the long run, Microsoft recovered its footing, pouring money into R&D, becoming a major player in cloud computing, and ultimately assisting in the development of OpenAI.

My best guess is that the U.S. should aim for something similar with Google. If the courts do identify anticompetitive practices by which Google keeps competitors out of the market for online advertising tools and exchanges, then instead of breaking the company up, they should probably just order Google to cease those practices. Such an order, of course, would carry an implicit threat — if Google eventually developed new and different anticompetitive practices to replace the old ones, the next lawsuit would order a breakup. This solution would still improve the ad market, and it would leave Google chastened but still with plenty of money to develop new lines of business and new cutting-edge technologies.

---