Skip to main content

The FTC’s Tech Summit on AI[1] highlighted three panels that reflect different layers of the AI tech stack – hardware and infrastructure, data and models, and front-end user applications. Today, we publish the first in a three-part series of “Quote Books” summarizing each of the three panels. This first quote book is focused on hardware and infrastructure, including semiconductor chips and cloud computing.

The voices of everyday Americans can sometimes be lost in discussions involving dense technical, policy, or legal language. While the benefits or risks of new technologies are being debated by policymakers, these individuals experience the effects of innovation in real-time. The FTC recognizes that this is not a representative sample of the entire population, and we strive to continue to listen and engage with a variety of perspectives. The goal of the “Quote Book” solely aims to reflect and compile quotes from the participants aggregated into common themes. This summary aims to be a resource, to quickly see various perspectives on topics – from ideas to enable competition and innovation to potential competition concerns like input foreclosure, tying, or other bottlenecks.

“At the chips layer, there's one company that is predominant in designing the most advanced chips. There's one company that is predominant in manufacturing those chips, and there's literally one company only that produces photolithography equipment, which is an essential input into manufacturing those chips,” said Ganesh Sitaraman, Director of the Vanderbilt Policy Accelerator

“Ensuring that our markets are open, competitive, and fair—and that start-ups are getting ahead or falling behind based on their own skill and ingenuity, rather than the dictates of an existing giant—is critical for ensuring that we continue to foster the breakthrough innovations that have kept America ahead,” FTC Chair Khan recently said in her remarks.[2]

Panelists in the Tech Summit on AI discussed ways that dominant firms may have control over key inputs such as cloud computing and access to hardware such as GPUs, and how this may be exacerbated by obstacles to migrating between offerings. This may in turn enable or enhance the exercise market power and protect incumbents. The panelists also discussed ways in which the structure of these markets may make it challenging for new players to compete even where their offerings may be better than the incumbents.

“As a startup founder, you are basically competing with the most valuable companies in the world […] and it includes fundamentally all of the major hyperscalers, all of the major platform companies. What we've seen over the last five years, through a series of acquisitions, is platform companies, hyperscalers, starting to make their own chips,” said Dave Rauchwerk, startup founder and technologist.

These topics are not new for the FTC. The agency has an existing track record on semiconductor chips and cloud computing. 

[Semiconductor chips] As Chair Khan explained in her recent remarks, “One of the first merger lawsuits we filed after I joined the agency blocked Nvidia’s attempted acquisition of Arm[3]—which would have been the largest semiconductor chip merger in history. Our team determined that giving one of the largest chip companies control over the computing technology and designs that rival firms rely on to develop their own competing chips would be bad for competition and would hamstring innovation of next-generation technologies. The trajectories of both companies in the wake of this action has illustrated how organic growth and competition can spur firms to further innovate in ways that benefit the business and the public alike. Not only has Nvidia remained the leading AI chip maker in the AI arms race, with a surging stock valuation—but Arm ended up going public and has a forward earnings multiple that is more than double Nvidia’s.”[4]

[Cloud computing] Additionally, the FTC is no stranger to security issues in the cloud, having brought numerous enforcement actions – including Drizly[5], Chegg[6], Skymed[7], Ascension[8], and Uber[9] – related to cloud security. Last year, the FTC launched a Request for Information on Cloud Computing business practices[10] outlining questions that cover single points of failure, security concerns, and market power and practices affecting competition. The FTC published an analysis[11] of the public comments – highlighting competition themes that relate to software licensing practices, egress fees, and minimum spend contracts. Tania Van den Brande, Chief Economist at Ofcom UK highlighted their study and report[12] along with themes including egress fees, switching costs, and discounting structures to shape incentives to put most, if not all of customers’ cloud needs with a single provider.

[Non-competes] Across both areas, staff have highlighted[13] that one of the essential building blocks of generative AI includes talent and expertise. To ensure a competitive and innovative marketplace, it is critical that talented individuals with innovative ideas be permitted to move freely, and crucially, not be hindered by non-competes.[14]

The rapid development of AI tools reminds us that “the technologies of tomorrow will be built on the inputs that companies have access to today.”[15] To that end, the FTC aims to ensure our skillsets and knowledge are keeping pace with evolving markets. We plan to continue to hear and learn from players across the AI ecosystem through forums like the recent FTC Tech Summit on AI and will continue to use our existing legal authorities to address harms.

“Inertia is such a powerful force it's hard to overcome. […] And it becomes this almost insurmountable series of obstacles to the corporate decision-making process where […] the big get bigger and the gulf grows wider. It becomes a bimodal distribution whether we want it to or not,” said Corey Quinn, Chief Cloud Economist at The Duckbill Group.














[14] The FTC issued a notice of proposed rulemaking that could potentially help to address such concerns. Non-Compete Clause Rule, 88 Fed. Reg. 3482 (proposed Jan. 19, 2023),


More from the Technology Blog

Get Business Blog updates