Author Name : Suvin Arora
Sundar Pichai, the CEO of Google, once said that AI is the most important thing humanity has ever worked on. Satya Nadella, the CEO of Microsoft, described AI as a platform shift. Even Jensen Huang, the CEO of Nvidia said that AI is no longer a single breakthrough or application – it is essential for infrastructure.
If there is one thing that all tech experts around the world agree on, it is the evolution of AI from a breakthrough technology, to a much more nuanced, intricate entity integrated into our daily lives. It has gone from answering daily queries to handling extensive arithmetic and cognitive tasks, but, before we crown the AI industry as the next big tech oligopoly, let us understand what it really means.
An oligopoly is a market dominated by a small number of powerful companies that have significant influence over prices, production, innovation, and the industry’s direction. High costs, control over resources, brand loyalty, and other barriers make it difficult for new competitors to enter. Thus, an oligopoly lies between a free market and a monopoly.
This connects directly to AI. While most people see AI simply as chatbots, image generators, or digital assistants, behind these tools lies a much larger system of chips, data centres, cloud infrastructure, data, and enormous economic power. Now that we have identified the scale of the AI industry, let us understand whether it resembles an oligopoly or not. Let us start with the positive markers:
Firstly, the structure: frontier AI is completely dominated by a handful of companies, namely OpenAI, Google, Anthropic, Meta etc. Their decisions influence the direction of the AI industry directly and indirectly.
Secondly, the fact that frontier models in AI, such as Gemini, ChatGPT, Claude etc., require enormous amounts of computing power, specialised chips, data, energy, and most importantly, money. This makes it extremely difficult for a standard AI startup to gain any momentum in such a capital-intensive industry.
Now, the most impactful point: dependence on scarce infrastructure. Nvidia is the dominant supplier of high-end AI GPU’s, while companies such as Microsoft, Apple and Google operate enormous cloud infrastructures used to train and run AI models. This leads to only a small amount of companies being able to realistically keep up with the chain of supply
However, the AI industry also has some features which challenge its claim to the throne as an oligopoly:
First, the very fact that AI has several competitors and a rapidly growing ecosystem contradicts the main criterion of being an oligopoly. Companies such as Mistral, Cohere, Deepseek develop AI models and applications
Next, open-source models, such as Meta’s Llama models have allowed developers and companies to build AI systems without developing a frontier model from scratch. An extension of this point is the very fact that the whole AI market in itself is unusually volatile. A company which leads in 2024 will almost certainly not do the same in 2025. This shows that the power doesn’t remain in the hands of one company for very long, switching hands almost annually.
Ultimately, the AI industry cannot be considered a true oligopoly. While a small number of companies may be shining today, advancements in efficiency, open source models and declining barriers to entry could pave the way for a freer market tomorrow. So, it is not a question of whether AI is an oligopoly or not- the question is, what do we want it to become?
