AdC Warns of Competition Risks in AI Chip Market
The Portuguese Competition Authority has released a paper highlighting competition risks in the AI chip value chain, emphasizing structural concentration and vertical integration issues.
19.02.2026 | Portuguese competition authority
The Portuguese Competition Authority (AdC) has published a paper examining the competition issues related to access to chips necessary for training and running Artificial Intelligence (AI) models. The report indicates that the AI chip value chain may exhibit structural characteristics that could lead to concentrated market outcomes.
One of the key findings is that the production of AI chips is dependent on a complex global value chain, which is marked by significant levels of concentration in both chip design and manufacturing. This concentration is attributed to factors such as substantial economies of scale, high capital expenditure requirements, and technological complexity. These elements can create bottlenecks and supply rigidity, potentially driving up AI chip prices and increasing barriers to entry for start-ups and AI developers.
The AdC also points out a trend of increasing vertical integration within the AI sector, where leading chip designers and cloud service providers are expanding their operations across various layers of the AI value chain. This has led to the development of integrated ecosystems that combine hardware, software, and services. Such developments may raise competition concerns by providing privileged access to sensitive information, aligning incentives across the ecosystem, and increasing the risk of market power being leveraged into adjacent markets.
As an example, the report highlights NVIDIA’s CUDA software stack, which, due to its widespread adoption and associated network effects, may reinforce NVIDIA’s position in the AI chip market. The AdC also notes that access to public high-performance computing (HPC) infrastructure could help mitigate some of the identified constraints, provided that access conditions are transparent and non-discriminatory.
This publication is part of the AdC’s ongoing efforts to assess competition issues related to generative AI, following previous papers on data access, AI model openness, and AI labor markets. The series aims to identify emerging competition risks in digital markets and support the AdC’s proactive enforcement strategy.
