AI startup Perplexity timed its recent $34.5 billion bid to buy Google’s Chrome web browser with a much-expected court ruling that could have facilitated such a deal.
But as that ruling by U.S. District Court Judge Amit P. Mehta came by on September 2, it fell short of Perplexity’s expectations: Google’s parent Alphabet doesn’t have to spin off Chrome, although it must share some of its search data with competitors, The New York Times reported. This follows an August 2024 ruling by the same court that Alphabet violated antitrust laws in striking deals with Apple and other device makers to make its search engine the default option.
Perplexity does benefit from Judge Mehta’s ruling; it could potentially have Apple include its search engine as a default on iPhones, along with other search providers like OpenAI. But as the ruling settles some aspects of the debate over the size of big tech, it raises new issues. Google, for one, has concerns about how the new data-sharing requirements could affect the privacy of its users, the Times report noted. Gen AI is a new face in the debate as it threatens to replace traditional search engines. In fact, Judge Mehta in his ruling pointed out that the rise of generative AI “changed the course of this case.”
The government’s antitrust case is that big tech firms use their size to exercise monopoly power. But firm size is the wrong measure to determine antitrust abuses, because market power relates to specific products, and not firms, according to Herbert Hovenkamp, Wharton professor of legal studies and business ethics, and law. “The power of a firm or a group of firms to raise prices across the market depends heavily on market share,” he wrote in his recent paper titled “Antitrust and eMarkets.”
In his paper, Hovenkamp also emphasized that e-commerce is an integral part of the marketplace and should not be treated as a separate market, as many do. For instance, many antitrust activists “describe the market shares of companies like Amazon strictly by reference to its online sales,” he said. “The fact is, Amazon competes in the greater market with Walmart and many other offline sellers.”
An Obsession With Size
“There are certain people in antitrust and many people who are outside of antitrust who think it’s all about the size of the firm, and that we should go after Microsoft or Google because they are very large,” Hovenkamp said.
“They want to break up big firms; they’re less interested in breaking up real monopolies,” Hovenkamp said. “An effective antitrust policy that has any chance of reducing prices or increasing output or stimulating innovation needs to break up monopolies like Google Search. Breaking up a firm is much less likely to do that, except occasionally by luck.” The exception is when a firm makes only one product.
“Tell the Yellow Cab drivers of Philadelphia that they do not compete with Uber, or the many small traditional retailers who have been injured or even ruined by Amazon that it is not their competitor.”— Herbert Hovenkamp
That is a flawed argument, he noted, citing Microsoft and Google parent Alphabet to show how firm size and market share don’t go hand-in-hand. Microsoft’s market cap values it at nearly $4 trillion, while Alphabet’s market cap values it at some $2.5 trillion. But in web search, Google has a market share of more than 90%, while Microsoft’s Bing has less than 5%, he said. “Microsoft could never monopolize search, notwithstanding that it’s a very large company.”
Determining Market Boundaries
“Any inquiry into the extent of the market must include questions about the competitive relationship among e-commerce firms with each other, as well as the extent of competition between e-commerce and offline firms,” Hovenkamp wrote in his paper. “Online markets have been part of the commercial culture for three decades. They can no longer be considered as anything other than an integral part of the marketplace.”
Hovenkamp explained why offline and online commerce are not mutually exclusive. “Consumer choice between online and offline commerce is largely unrestricted, and the boundaries are easily crossed,” he wrote. “Tell the Yellow Cab drivers of Philadelphia that they do not compete with Uber, or the many small traditional retailers who have been injured or even ruined by Amazon that it is not their competitor.”
The paper noted that Amazon, which sells mainly online, and Walmart, which is mainly offline, consider one another as their principal competitor across thousands of products, although each competes with several other online and offline stores, too. Drawing from that, “any antitrust market definition that includes Amazon but not Walmart, or vice versa, is at least presumptively nonsense,” Hovenkamp said.
Product-by-Product Analysis Is Key
Competition between offline and online commerce can be assessed only on a product-by-product basis, and multi-product firms often have widely differing market shares in different products. The implications are hard to overstate.
Most times, antitrust cases require proof of the defendant’s market power, or ability to control the market and raise prices above cost, Hovenkamp said. For example, Amazon has a roughly 40% share of “e-commerce,” but only 4% of “commerce,” which includes offline sales as well, he added. In order to make a case of an antitrust violation, “the firms involved have market shares in the 30% to 40% range, but not when the market share is 4%.”
E-commerce also rarely reduces competition and threatens innovation, Hovenkamp continued. “Collusion is not a big problem in e-commerce. Not that it doesn’t exist, but it could,” he said. “Most of e-commerce comprises new entry, which is output-expanding. E-commerce overall is also highly innovative.”
Another dimension of the antitrust debate is the so-called “reasonable interchangeability” test, which helps determine if one should place two products that one can substitute for another in the same relevant market.
“Reasonable interchangeability indicates that two goods are in the same market only when both are being priced at close to the competitive level,” the paper stated. But it is risky to place two goods with different inputs or cost structures into the same market simply because customers use both, the paper stated.
“You can use both wax paper and cellophane to wrap your peanut butter sandwich, just as you can either walk to work or take the subway,” the paper noted. “But these pairs of products are made with very different technologies and use very different inputs in their production.” Each set of those substitutable products has different fixed and variable costs, it pointed out.
“One important feature of market definition in antitrust cases is that it is a question of fact.”— Herbert Hovenkamp
Clarity With the “Hypothetical Monopolist Test”
According to Hovenkamp, the dominant and the most successful test for market definition is the so-called hypothetical monopolist test, or HMT. In this test, analysis of data on substitution helps identify the smallest grouping of sales capable of sustaining a monopoly price. The data tracks a firm’s price increase in the product or location to estimate how many sales the firm loses and the effects on profits, and to which alternative sellers they go, the paper explained. The HMT does not distinguish between offline and online sales.
Significantly, in order to identify the “relevant market” for antitrust purposes, the hypothetical monopolist test goes beyond considering the physical features of a product to observe consumer behavior or supplier behavior, Hovenkamp noted. This test could also help bring clarity to cases such as those against Amazon over price increases. “If Amazon is selling a product like, say, shampoo, its principal competitor will be Walmart. So, we want to question whether Amazon has market power in that product vis-a-vis the full range of sellers — and those will be both offline and online sellers.”
On the other hand, Amazon’s Prime Music or Prime Video are services deliverable only on the internet, Hovenkamp continued. But people who patronize that streaming market could alternatively buy physical CDs or go to concerts, he noted. “To what extent do these offline alternatives need to be incorporated into the market as well?” Those are all “fact questions” that a good economist with the right data could address, he added. “One important feature of market definition in antitrust cases is that it is a question of fact.”
How AI Affects Antitrust Issues
“As an antitrust matter, which is concerned with higher output, the development of AI so far has been extremely positive,” said Hovenkamp. “It has produced more product differentiation, greater output, and intense competition within a group of very large players to develop products for this market.”
Of course, AI could have negative effects, such as copyright infringement or making it easier for students to cheat in exams, but those are not antitrust problems, he added.
Also, with AI, consumers have greater access to services. “[Consequently,] there’s less fear of monopoly products from big tech,” Hovenkamp said. “And there are not that many products sold by big tech that are truly monopolized. Monopoly is no more common on the internet than it is in old commerce, and arguably even less common.”
