After The Money's Gone: AI, Antitrust, and Big Tech

For Google, Holy Week became an unholy tsuris. Just days before hearings were set to begin in federal court in the remedies phase of the antitrust case it lost last year over its search monopoly, a second federal court ruled it was also guilty of maintaining an illegal monopoly in online advertising. It now faces the prospect of being forcibly dismembered through court-ordered sales of Chrome, possibly Android, and at some point perhaps either its buy-side or sell-side ad platform if the Justice Department gets its way in its proposed remedies in the two cases.

Nor is Google the only tech giant currently on the antitrust hotseat. Down the hall in the same court house in Washington, DC, the Federal Trade Commission is seeking to force Meta to divest Instagram and WhatsApp over what the government now claims were the Facebook parent’s competition-quashing intent in acquiring the two startups over a decade ago, even though the agency had cleared the two deals at the time.

Microsoft’s partnership and $13 billion investment in OpenAI has also come under scrutiny from antitrust regulators, on both sides of the Atlantic. The FTC opened an investigation into the partnership last year but did not take action before the change in administrations. The agency issued a report in the waning days of the Biden administration warning that exclusive or proprietary partnerships between cloud service providers and AI companies could raise antitrust concerns but did not single out the Microsoft-OpenAI tie up. Competition authorities in the European Union and the U.K. also examined the partnership, although the U.K.’s Competition and Markets Authority dropped its probe without taking action last month.

This week, the EU slapped Apple and Meta with a combined €700 million ($800 million) in fines for violating competition rules under the Digital Markets Act.

With the exception of the Microsoft-OpenAI deal, none of those cases are directly related to AI (although the DOJ has raised the specter of Google’s use of AI tools to extend its monopoly in search). But the growing antitrust scrutiny of Big Tech, let alone any forcible restructuring on their operations, could nonetheless have a significant impact on the future development of the technology.

The issue is money. It takes insane amounts of money to develop, train, and operate, generative AI models. That’s the main reason the leading model developers (outside of China) are all either Big Tech incumbents, or, like OpenAI and Anthropic, startups that are backed by those incumbents. What’s at stake in the stepped up scrutiny they’re all under, apart from its direct impact on their entanglements and operations, is their ability to continue to fund the continued development of AI models.

According to an estimate by market research firm eMarketer, Instagram is expected to account for more than half of Meta’s total revenue this year. It, along with WhatsApp, are also critical to keeping users within the Meta ecosystem, helping boost overall engagement, generate additional data on users, and thus increasing advertising revenue. A forced divestiture of the two apps as the government is requesting, could, at least initially, have a devastating impact on Meta’s bottom line, and with it the cash needed to keep iterating and operating its Llama AI models.

Google derives nearly all of its revenue from advertising, starting with search-based advertising. The loss of Chrome, and perhaps Android, as the government is seeking in the search-monopoly case, would be a major blow both to Google’s top-line revenue and its ability to harvest data from billions of browser and mobile device users. A forced sale of one or both of its digital ad platforms as a result of the online advertising case, could deal another major blow to its revenue and its ability to harvest data from advertisers.

Microsoft’s cash cows are its Windows monopoly and its Azure cloud computing platform. A forced split from OpenAI would not significantly affect the former. But Microsoft derives a great deal of AI technical knowledge and expertise from the partnership. Much of its cash investment in OpenAI also come back to it in the form of cloud computing contracts that help defray the costs of building and operating the massive data centers that underly Azure.

All that is looming, moreover, as investors are growing increasingly uneasy over the enormous capital spending on AI infrastructure and operations by the tech companies and its potential impact on earnings. In January, OpenAI announced plans to spend as much as $500 billion as part of a multiyear build out. Meta said it intends to spend as much as $65 billion on new AI infrastructure this year alone. The fact that a compelling business model that might justify all that investment is yet to emerge for generative AI has only added to investors’ unease, as has the surprise emergence of highly capable but apparently low-cost models like China’s DeepSeek.

In the face of investor concerns, both Amazon and Microsoft, the two largest cloud computing providers, have recently slowed their roll on data center construction, an early sign of a possible retrenchment in spending.

We’re a long way from the end of the Meta and Google cases, or from any resolution to the ongoing scrutiny of ties between cloud service providers and AI companies, like OpenAI and Microsoft, or Amazon and Anthropic.

But the loss of the cash cows that have fed Big Tech’s orgy of spending on AI could open space for startups and would-be competitors to get a toe-hold in the market. If so, that could ultimately work to the advantage of creators and rights owners as they engage with AI developers over the use of their content.

More competitors would mean more potential buyers in the market vying to secure access to the most high quality training data. That would help alleviate the danger of monopsony that former FTC chair Lina Kahn warned about as early as 2023, in which a small number of very large buyers are able to dictate prices to suppliers.

“Conduct that may violate the copyright laws––such as training an AI tool on protected expression without the creator’s consent or selling output generated from such an AI tool, including by mimicking the creator’s writing style, vocal or instrumental performance, or likeness—may also constitute an unfair method of competition or an unfair or deceptive practice,” the FTC said in comments to the U.S. Copyright Office. “In addition, conduct that may be consistent with the copyright laws nevertheless may violate Section 5.17 [of the FTC Act]. Many large technology firms possess vast financial resources that enable them to indemnify the users of their generative AI tools or obtain exclusive licenses to copyrighted (or otherwise proprietary) training data, potentially further entrenching the market power of these dominant firms.”

All rise.