Since the release of ChatGPT in November 2022, generative AI (GenAI) has advanced considerably, but scarce computational resources could constrain its future growth. In response, downstream labs that supply GenAI products to end users are increasingly forming vertical partnerships with upstream hyperscalers that provide the computing power needed to develop, train, and deploy GenAI models at scale.
Antitrust enforcers are scrutinizing these partnerships, with current attention on whether they operate as “de facto mergers” or enable unlawful coordination. Investors appear to expect a “winner-takes-all” GenAI market, however, so the more relevant question may be whether the partnerships foreclose competition. In this article published in the ABA Antitrust Law Section’s newsletter, author Arshia Hashemi applies the ability and incentive framework to two features, cloud investments (potential customer foreclosure) and margin squeezes (potential input foreclosure), and weighs the theoretical foreclosure risks against procompetitive efficiencies. Because those risks are fact specific, each partnership warrants case-by-case assessment.
Chicago
