Too Big To Fail? NVIDIA’s Dominance and Rethinking Antitrust Enforcement in the AI Era
In an era defined by rapid advances in artificial intelligence, the rise of leading chip manufacturers has reshaped both economic and regulatory landscapes. One such company that has established its dominance is NVIDIA. NVIDIA also holds an overwhelming market share in the AI and graphics processing unit (GPU) markets, capturing 92% in the add-in-board GPU market as of Q1 2025. [1] Becoming the first company to reach $5 trillion in market value, its power and wealth are undeniable. [2] However, NVIDIA’s rapid ascent from a graphics-card company to the world’s leading supplier of AI-accelerator chips has prompted new antitrust questions. While ensuring competition is important, the strategic importance of AI hardware for national security and competing globally means that antitrust enforcement cannot operate in a vacuum. By striking a delicate balance between preserving vigorous competition and sustaining U.S. leadership in a critical industry, we can find the best path forward to regulate NVIDIA to safeguard competition, foster innovation, and protect national-security interests in the age of AI.
Historically known for GPUs in gaming and visualization, NVIDIA has leveraged its hardware platforms, software frameworks, and relationships with data-centre and cloud firms to become the dominant vendor for AI training and inference workloads. One such prominent software is CUDA, NVIDIA’s proprietary parallel computing platform and programming model that allows developers to speed up their applications by harnessing the power of GPU accelerators, making it easier to build and optimize AI applications. [3] Modern AI workloads, like training large language models or running generative-AI systems, demand huge computing resources, fast memory access, efficient communication between chips, and robust software support. NVIDIA meets all these demands because it combines advanced GPU chip design, efficient memory architectures, high-speed interconnects between chips, and a widely used software infrastructure. For example, its use of High Bandwidth Memory (HBM) gives GPUs the ability to move data between memory and processing units extremely quickly, which matters because if data isn’t fed to the GPU cores fast enough, performance can suffer. [4] Meanwhile, its interconnect technology, such as NVLink, allows multiple GPUs to communicate at high speed, which is crucial when AI workloads must use many GPUs in concert.
Multiple regulatory authorities have turned their gaze to NVIDIA’s market position and business practices. In the United States, the United States Department of Justice (DOJ) antitrust division opened investigations into NVIDIA’s conduct in the AI-chip market, including issuing subpoenas to NVIDIA and third-party firms regarding alleged unfair supply and pricing practices. [5] Additionally, NVIDIA’s market grip has drawn public calls for enforcement, such as the U.S. Sen. Elizabeth Warren and other progressive groups pressing the DOJ to investigate NVIDIA’s practices. [6] But does NVIDIA’s conduct meet the standard for monopolization?
The Sherman Antitrust Act, passed in 1890, outlaws any person “who shall monopolize, or attempt to monopolize, or combine or conspire with any other person or persons, to monopolize any part of the trade or commerce”. [7] To establish a violation of Section 2 of the Sherman Act, whether the firm has “monopoly power” in any market must be proven first. Then, it must be proven whether that leading position was gained or maintained through improper conduct. [8] First, NVIDIA’s market share figures strongly suggest that it has monopoly power (or near-monopoly) in the relevant market. Proving NVIDIA conducted exclusionary or anticompetitive conduct, on the other hand, would require more thorough investigation. Improper conduct could be proven if, for example, NVIDIA required or induced customers to purchase bundled software with hardware or if it employed contracts that foreclosed rival chips. Merely having a high market share or obtaining a monopoly by superior products, innovation, or business acumen is legal, but antitrust concerns arise when there is exclusionary or predatory conduct beyond lawful competition.
However, should there be an antitrust investigation into NVIDIA at all? The semiconductor sector, especially AI accelerators, sits at the convergence of commercial competition, global supply-chain resilience, and national security imperatives. The passage of the CHIPS and Science Act in 2022, authorising over $50 billion in subsidies to boost domestic chip production, underscores the strategic importance of this supply chain. [9] While such an entrenched dominance raises concerns about a brittle supply chain or dampening innovation once competitive pressure disappears, having a dominant U.S. firm such as NVIDIA serves national security interests by enabling the U.S. to lead in critical AI hardware, reducing our dependence on foreign suppliers. Having a rigorous antitrust enforcement against NVIDIA can inadvertently undermine the U.S. objective of maintaining leadership in AI hardware and supply-chain security.
On the other hand, the AI ecosystem thrives on rapid iteration of chips, accelerators, interconnects, and software frameworks. If one firm secures overwhelming control of hardware, software, and supply, the risk is that innovation may converge around that firm’s architecture, like NVIDIA’s CUDA ecosystem, which deters alternative designs or open‐source frameworks that could challenge the dominant paradigm. If rivals cannot credibly challenge NVIDIA, the risk of complacency, lower investment in rival architectures or software ecosystems, and reduced developer choice may weaken long-term AI competitiveness and inadvertently harm national security resilience.
To bridge the gap between preserving competition and sustaining U.S. leadership in a critical industry, we should rethink traditional analytical tools of antitrust law that may be poorly matched to the realities of high-tech, innovation-driven markets. For example, antitrust agencies should continue their investigation of NVIDIA but tailor their approach to the unique characteristics of the AI-accelerator market. To mitigate lock-in and promote competition, policy intervention could encourage or require interoperability standards, open APIs, and decoupling of software ecosystems from hardware platforms, like enabling models to run on non-NVIDIA accelerators. Additionally, antitrust policy and national security for semiconductors and AI should be aligned by weighing a dominant firm’s potentially exclusionary conduct against its effect on supply-chain resilience. Finally, regulators should ensure that antitrust enforcement does not penalise firms winning due to superior design or innovation, but only when innovation is harmed due to exclusionary conduct.
As the competitive and regulatory spotlight increasingly turns to the architecture of the AI-accelerator market, the challenges posed by a dominant firm like NVIDIA become impossible to ignore. As probes into whether NVIDIA has used exclusionary practices to cement its lead increase, the unique setting of NVIDIA and AI hardware in the global economy and geopolitics requires a recalibration of antitrust enforcement to ensure competition while preserving technological leadership and supply-chain resilience. A balanced approach that updates antitrust enforcement to the AI market, promotes interoperability and open ecosystems, and aligns national security with antitrust policy can offer a promising way to preserve competition, encourage innovation, and protect national security interests as AI continues to transform critical sectors.
Edited by Vincent Hovsepian
Endnotes
[1] Faizan Farooque, Nvidia Secures 92% GPU Market Share in Q1 2025, Yahoo Finance (Yahoo Finance June 6, 2025), online at https://finance.yahoo.com/news/nvidia-secures-92-gpu-market-150444612.html (visited Nov 26 2025).
[2] Niket Nishant & Rashika Singh, Nvidia Nears Record $5 Trillion Valuation as AI Boom Powers Meteoric Rise, Reuters (Reuters Oct 29, 2025), online at https://www.reuters.com/business/nvidia-poised-record-5-trillion-market-valuation-2025-10-29/ (visited Nov 26, 2025).
[3] Fred Oh, What Is CUDA, NVIDIA Blog (NVIDIA Blog Sept 10, 2012), online at https://blogs.nvidia.com/blog/what-is-cuda-2/ (visited Nov 26, 2025).
[4] NVIDIA Corporation, NVIDIA H200 GPU, NVIDIA Cloud & Data Center (NVIDIA Nov 13, 2023), online at https://www.nvidia.com/en-us/data-center/h200/ (visited Nov 26, 2025)
[5] Angela Luna, The DOJ and Nvidia: AI Market Dominance and Antitrust Concerns, AAF (American Action Forum Oct 7, 2024), online at https://www.americanactionforum.org/insight/the-doj-and-nvidia-ai-market-dominance-and-antitrust-concerns/ (visited Nov 26, 2025).
[6] Elizabeth Warren, Warren Throws Support behind Department of Justice Probe into AI Chipmaker Nvidia, Underscores Need for Comprehensive Investigation, Warren.senate.gov (US Senate Sept 6, 2024), online at https://www.warren.senate.gov/newsroom/press-releases/warren-throws-support-behind-department-of-justice-probe-into-ai-chipmaker-nvidia-underscores-need-for-comprehensive-investigation (visited Nov 26, 2025).
[7] Sherman Act, 15 USC §§ 1–7 (1890).
[8] Federal Trade Commission, Monopolization Defined, FTC.gov (Federal Trade Commission June 11, 2013), online at https://www.ftc.gov/advice-guidance/competition-guidance/guide-antitrust-laws/single-firm-conduct/monopolization-defined (visited Nov 26, 2025).
[9] H.R. 4346, 117th Cong., 1st Sess. (7/1/2021).