This article originally appeared in Law360 on 24 January 2025.
On Jan. 17, the Federal Trade Commission revealed the much-anticipated findings of its ongoing inquiry into the artificial intelligence sector, publishing its staff report on AI partnerships and investments. While the report was pushed out at the end of the Biden administration, it and the accompanying statements from members of the incoming FTC majority shed light on the direction of future antitrust enforcement in the AI segment.
The report's findings regarding the power and influence of large cloud service providers are particularly notable in light of the FTC transition under the new administration.
Both President Donald Trump and the sitting Republican FTC commissioners have advocated for increased enforcement against Big Tech, and those sentiments are echoed in the commissioners' concurring and dissenting statements regarding the report's release. This strongly suggests that the new administration is likely to focus on the report's observations regarding Big Tech and to more closely scrutinize the AI sector.
Observations on Investments and Partnerships
The FTC market study sought to shed light on the nature and competitive implications of certain partnerships involving three leading cloud service providers, Google, Amazon and Microsoft, and two leading developers of large language models, OpenAI and Anthropic.
While some information regarding these partnerships has been publicly available, the report provides insight based on the more granular information contained in internal documents companies provided at the FTC's demand.
The report identifies several key terms in these partnerships relevant to their potential effect on competition:
- Significant equity and revenue-sharing rights for the cloud service providers;
- Certain "consultation, control and exclusivity rights" for the providers;
- Requirements that the AI developers spend large sums on the cloud services;
- Sharing of "key resources and information" between partners, including computing resources at discounted rates, assets and intellectual property relating to developers' models, and certain financial and training data; and
- Opportunities for the providers to expand current offerings, including through integration of AI models into their products or through deployment of the models on their platforms.
The report noted the observation by one partner that information sharing through the partnership amounted to a "multiyear crystal ball into the future needs of AI infrastructure." It also described one party's belief that only a few players could scale past the capabilities of current frontier models, and that the partnerships were important for model developers to stay ahead of the pack.
Areas to Watch
In light of these observations, the report identified three areas for potential competition issues. These observations are largely consistent with those expressed by other competition regulators — including the European Commission and the U.K. Competition and Markets Authority — in connection with their own AI sector inquiries.
Access to Inputs
The partnerships could affect access to key components needed for AI product offerings, including compute resources and engineering talent. Cloud service providers might, for example, consider limiting the compute resources made available to AI developers.
As one provider employee acknowledged in a document, "Scarce GPU resources are being disproportionately used by a few large customers who are getting steep discounts ... driving hoarding behavior." Another noted that the "quality of results across a broad range of AI tasks is mostly constrained by the amount of AI training computing resources that can be brought to bear on the training task."
Increased Switching Costs
The partnerships could increase contractual and technical switching costs for AI developers by restricting their use of multiple service providers or otherwise making it more difficult for them to switch providers, even after the terms of the partnerships have ended.
All the partnerships, for example, included billions of dollars in spending commitments for cloud services, and multiple partnerships imposed "conditions or restrictions" on the developers' ability to operate with other companies.
Moreover, potentially lengthy migration times between specialized cloud services and AI chips create technical barriers to switching.
Sharing of Sensitive Information
Finally, the partnerships provide cloud service providers with access to "sensitive technical and business information" from the AI developers that may be unavailable to others. That information includes data related to generative AI models, AI development methods, confidential chip designs and customer and revenue numbers.
The information could be used by providers to develop their own internal models and apps, some of which may compete with those of their developers. The report noted one provider's acknowledgment that "if we wait for our own models to mature, we risk not participating in developing the necessary IP to build, operate and secure these applications" and that partnering would allow it to "more quickly learn the 'art' of refining and reinforcing these types of models."
The report goes on to note a number of questions for future exploration, including the partnerships' impact on nonpartner AI developers, specific switching costs and the competitive advantages obtained by service providers through their simultaneous role as investor, intermediary and market participant.
Context and Future Implications
Both the stated scope of the report and accompanying statements from the incoming FTC majority shed light on its implications for future enforcement.
FTC staffers acknowledge that the report it not a "formal legal or economic analysis of the partnerships or markets," that its observations are based on a narrow set of information provided by the five responding companies, and that even a "comprehensive analysis of the subject partnerships is beyond the scope of the report."
These significant qualifications underscore the report's limitations both for serving as a basis of any future enforcement action and for fully ascertaining the competitive dynamics of the AI ecosystem.
Perhaps equally important are the continued criticisms of Big Tech in the concurring and dissenting statements of FTC Chair Andrew Ferguson and Commissioner Melissa Holyoak, who will be members of the FTC majority once a new Republican commissioner is confirmed.
Ferguson and Holyoak objected to the "areas to watch" section of the report, with the former explaining that the quick conclusion and notably narrow scope of the report "should foreclose the drawing of broad conclusions about the AI industry and its future, or even about the partnerships themselves." Since that section is based on speculation about the future, the FTC should not have included it in the publication, and it should be skipped or read "with tremendous skepticism," Ferguson said.
Holyoak's statement — joined by Ferguson — also notes that the report "leaves no doubt about the significant roles Google, Amazon and Microsoft play in shaping how AI products and markets develop." And, Holyoak said, given both past censorship of conservatives and commitments made to the Biden-Harris administration, one area of future interest is how business-to-business agreements are intended to reshape consumer-facing products in ways that restrict free speech.
The statements by Ferguson and Holyoak echo concerns that they, along with Trump and other members of the new administration, have expressed regarding the power and influence of Big Tech companies. They have indicated a need to rein in this influence to protect startups and other so-called Little Tech firms.
These themes will thus likely inform antitrust enforcement in the AI segment as the new administration continues to take shape and articulate its priorities. In the meantime, despite its acknowledged limitations, the report serves as a strong indicator of how the agencies will view future competitive concerns in this rapidly evolving space.
