The article examines how Nvidia’s dominance in artificial intelligence hardware and software has attracted extensive antitrust scrutiny while simultaneously becoming central to national strategies to expand computing infrastructure. It outlines how competition regulators across multiple jurisdictions are probing whether Nvidia’s conduct in graphics processing units and related tools undermines competition, particularly in the emerging foundation model ecosystem. At the same time, governments that view artificial intelligence as a geopolitical priority are partnering with Nvidia to accelerate domestic deployments, raising questions about whether reliance on the company will blunt the appetite for strong enforcement.
The piece maps a series of investigations and inquiries into Nvidia’s business practices. In the United States, the Department of Justice sent questionnaires and in September 2024 escalated its scrutiny by issuing a subpoena to examine whether Nvidia violated antitrust laws, including concerns that it makes switching suppliers difficult, penalizes buyers that do not use its GPUs exclusively, and offers preferential allocation and pricing to customers that commit to its full systems. The United Kingdom’s Competition and Markets Authority reported an “interconnected web” of over 90 partnerships and strategic investments linking Nvidia with major technology firms and warned that firms controlling critical inputs for developing foundation models may restrict access, exploit their positions in downstream markets, and use partnerships to entrench market power. French authorities raided Nvidia’s offices on September 28, 2023, and a June 2024 market study by the Autorité de la Concurrence found likely abuses of dominance involving “price fixing, production restrictions, unfair contractual conditions and discriminatory behavior,” as well as concerns about dependence on CUDA and Nvidia’s investments in artificial intelligence focused cloud providers such as CoreWeave. The European Commission began informally collecting views on potential anticompetitive practices in GPUs, while China’s antitrust authority initially challenged commitments related to Nvidia’s Mellanox acquisition before dropping its investigation following trade agreements with the United States.
The article argues that these regulatory efforts are increasingly intertwined with national ambitions to lead in artificial intelligence, which may jeopardize robust enforcement. In the United States, President Donald Trump’s AI Action Plan aims to promote national champions to “accelerate American leadership,” explicitly positioning Nvidia as integral to that goal. The European Union is investing in “AI factories” and a so-called “Eurostack” of regional infrastructure, but still relies on offerings from US firms, including Nvidia, to supply “sovereign” systems. The United Kingdom similarly seeks to “mainline[…] AI into the veins” of the country through partnerships with US technology companies. Given Nvidia’s near monopolist position in artificial intelligence focused chips, its CUDA software, and the widespread adoption of its GPUs for “sovereign” applications, governments face a dilemma: as they go “all in” on artificial intelligence, their dependence may undermine the political will to restrain the company’s market power.
The analysis warns that this dependence creates a gatekeeper dynamic in which Nvidia accrues economic, informational, and political power. By controlling key infrastructure and allocation of resources, the company can shape which technologies and applications are prioritized, reinforcing chief executive Jensen Huang’s vision of a highly automated world defined by “agentic AI” and very large language models that demand intensive computation and therefore more Nvidia GPUs. The article notes that Nvidia has already leveraged its position in geopolitical negotiations, with Huang pressing President Trump to relax export controls and the administration announcing that Nvidia can sell its H200 chips to China in return for the United States government receiving a 25% cut of those sales. Citing scholars from the AI Now Institute and Tech Policy Press, the piece argues that embedding artificial intelligence systems from dominant firms into public services risks eroding democratic accountability, concentrating decision-making in opaque automated systems and in the hands of technology providers. As governments outsource more functions to big technology companies, those firms become further embedded in critical infrastructure, which allows them to set the policy agenda, protect their markets, and build a durable shield against regulatory scrutiny.
