India's competition regulators are preparing to govern artificial intelligence using rulebooks designed for a different era, potentially allowing the same tech giants that dominate cloud computing and chip markets to entrench their power in AI. The Competition Commission of India (CCI) released a major report on AI and competition last year, but its recommendations rely on outdated enforcement strategies that experts warn won't work in fast-moving digital markets .

Why Is India's AI Regulation Approach Problematic?

India's CCI conducted an expansive market study on artificial intelligence and competition, examining how AI technologies are being developed and deployed across the Indian economy. The study revealed something alarming: the AI supply chain is already heavily concentrated in the hands of a few massive companies, creating barriers that make it nearly impossible for Indian startups to compete .

The data tells a striking story. NVIDIA controls 88% of the graphics processing unit (GPU) market share in India, a critical bottleneck since GPUs are the specialized chips needed to train and run large language models (LLMs), which are AI systems trained on vast amounts of text data. Meanwhile, just three companies, Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP), control nearly 65% of the compute market and about 46% of the data layer . These aren't just statistics; they represent the infrastructure that every AI company needs to build anything.

Despite identifying these dangerous levels of concentration, the CCI's report recommends an approach called "ex-post" regulation. This means allowing companies to compete freely until harm is proven, rather than preventing harmful practices before they happen. The recommendations include self-auditing by AI firms, voluntary transparency measures, and advocacy efforts by the CCI .

"The report's investigation into the core focus areas of Indian startups reveals that about two-thirds of Indian startups work in the area of developing AI applications models, while only 3% work in foundational model development, 10% in the compute infrastructure layer, and 20% in the data layer," noted Abhineet Nayyar, Senior Research and Policy Associate at IT for Change.

Abhineet Nayyar, Senior Research and Policy Associate at IT for Change

This concentration of effort at the application layer, rather than in foundational infrastructure, reveals a troubling reality: Indian startups are being pushed toward building on top of Big Tech's platforms rather than competing with them directly .

How Can Regulators Better Protect AI Competition?

• Adopt Ex-Ante Rules: Instead of waiting for anticompetitive harm to occur, regulators should establish clear rules upfront that dominant firms must follow, similar to the approach the European Union has taken with its Digital Markets Act.
• Address Bundling Practices: Large tech companies are already engaging in mandatory bundling of AI services and deep discounting at the model layer, practices that lock in customers and prevent startups from gaining market share.
• Ensure Cloud Interoperability: Regulators should require that AI models and data can move between different cloud providers, preventing companies from being trapped on a single platform.
• Monitor GPU and Chip Markets: With NVIDIA's overwhelming dominance, regulators need to watch for anticompetitive practices like bundling GPUs with proprietary software that raises barriers for competitors.
• Protect Foundational Model Development: Policies should incentivize Indian companies to build foundational models and infrastructure, not just applications, to create genuine competition.

The problem with India's current approach is that it assumes digital markets work like traditional industries. In reality, digital platforms have unique characteristics that make ex-post enforcement ineffective. First, platforms can use network effects to lock in users quickly, giving first-movers an overwhelming advantage. Second, software scales at almost zero cost, so a successful company can expand into adjacent markets with ease, often by bundling services together. Third, anticompetitive conduct by dominant platforms, like strategic acquisitions of competitors, often looks legal under traditional competition law because regulators focus narrowly on whether consumers pay lower prices .

These dynamics are already playing out in India's AI market. Big Tech firms have invested billions in developing datasets, algorithms, and AI products on top of their existing cloud and software offerings. The financial requirements to build and scale large language models pose the biggest challenge for startups, including Indian ones, to enter the market. Emerging market evidence indicates that Big Tech firms are engaging in practices like mandatory bundling of AI services, deep discounting at the model layer, and restricting cloud interoperability, all of which are likely to further entrench their dominance and restrict startups from competing .

India's draft Digital Competition Bill, proposed in 2024 by the Committee on Digital Competition Law, was supposed to introduce ex-ante rules for digital markets. However, two years later, the bill has yet to be officially introduced in Parliament, while digital markets have seen rapid proliferation and expansion through the AI revolution .

The CCI's report does deserve credit for identifying the real anticompetitive risks in India's AI ecosystem, including high entry barriers, switching costs, self-preferencing practices, algorithmic cartelization, and price discrimination. But by recommending an outdated enforcement approach, the report risks allowing the same patterns of monopolization that have defined global tech markets to take root in India's emerging AI sector. For Indian startups and the broader goal of building a competitive AI economy, the stakes could hardly be higher.