AI’s Next Monopoly May Not Look Like a Monopoly – Is India’s Competition Law Up for the Challenge?

Artificial intelligence is no longer a futuristic buzz‑word; it’s becoming the backbone of every sector from finance to health‑care. The unsettling part? The very foundations of AI – high‑end semiconductors, massive cloud servers, proprietary data sets and the so‑called "foundation models" – are increasingly held by a handful of global players.

When those inputs sit in the hands of a few, the whole downstream market starts to feel the pressure. Imagine trying to launch a new fintech app only to discover that the only cloud provider that can run your AI models also happens to be a competitor in the same space. That’s the new reality many businesses are staring at, and it raises a big question: are we heading toward a monopoly that looks nothing like the classic, single‑company dominance we’ve seen before?

What makes AI different from earlier digital waves is that the gate‑keeping power isn’t always about owning a platform. It can be about who gets preferential access to compute power, who signs exclusive data‑sharing deals, or who can offer cheaper cloud credits. Those levers, once pulled, can tilt the competitive landscape across every industry – banking, education, e‑commerce, you name it.

Take the foundation models that most developers now plug into via APIs – ChatGPT, Gemini, Claude, and the like. Companies are no longer building massive models from scratch; they’re layering their products on top of these ready‑made engines. The competition, therefore, shifts from the app layer to the owners of the underlying models and the hardware that runs them. If you control the chips, the cloud, and the data, you essentially set the rules of the game.

Think about NVIDIA’s dominance in the high‑performance GPUs that train today’s large models, or how Google, Microsoft, Amazon and Meta have a foot in virtually every step of the AI value chain. They sell chips, run clouds, develop foundation models, and push consumer apps. This vertical integration creates a potent mix of ability and incentive to squeeze out rivals – be it by limiting access to computing resources, locking away valuable datasets, or simply favoring their own AI products over third‑party services.

There’s also a feedback loop that’s worth mentioning. Better compute resources lead to better models, which attract more users, which generate more data, which in turn refines the models further. It’s a virtuous cycle for the incumbents, but a pretty steep hill for any newcomer trying to break in.

From a competition‑law perspective, three major worries surface.

1. Algorithmic collusion. Modern pricing engines can learn to set supra‑competitive prices without any explicit agreement between firms. In the U.S., the Justice Department sued RealPage in 2024 for enabling landlords to share sensitive pricing data via its software, leading to coordinated rent hikes. While that case showed existing laws can tackle overt data sharing, it doesn’t fully address a scenario where algorithms independently converge on similar pricing strategies without any communication.

India’s Competition Act, particularly Section 3, focuses on agreements, decisions and concerted practices. The recent amendment broadening the scope to “hub‑and‑spoke” arrangements might capture some forms of algorithm‑mediated coordination, but a self‑learning algorithm that decides “higher price = higher profit” on its own sits uncomfortably outside the classic definition of collusion.

2. Algorithmic self‑preferencing. When a firm controls both the model and the infrastructure, it can nudge the system to favor its own services. Think of a cloud provider that subtly pushes its own AI APIs higher in the marketplace, making it harder for independent developers to compete. This isn’t just a pricing issue; it’s about shaping the very terms of innovation.

3. Vertical foreclosure. A company that supplies chips, hosts clouds, and runs a foundation model can potentially block rivals at multiple points – denying compute, restricting data, or even offering bundled services that lock customers into an ecosystem. The result is a choke‑point that stifles upstream and downstream competition alike.

So, where does India stand? The country has a robust competition framework on paper, and recent amendments show a willingness to adapt. Yet, the speed at which AI infrastructure is consolidating may outpace the legislative process. Enforcement agencies will need technical expertise, real‑time market monitoring, and perhaps a fresh set of guidelines that consider non‑human actors – the algorithms themselves.

In practical terms, policymakers could explore a few avenues: encouraging open‑source AI models, mandating data‑portability standards, or even setting caps on the concentration of cloud‑compute capacity among a single entity. International cooperation will also matter, because the same tech giants operate across borders, and unilateral action can only go so far.

The window to act is narrow. Once the AI supply chain solidifies around a few dominant players, trying to untangle it later will be akin to rewiring a building that’s already been occupied for years. The goal isn’t to stifle innovation – it’s to ensure the playing field stays open enough for fresh ideas to flourish, for startups to experiment, and for consumers to benefit from genuine competition.

Bottom line: AI’s next monopoly may look more like an invisible wall than a single corporate behemoth. Whether India’s competition law can climb that wall before it becomes too high remains the pressing question for regulators, businesses, and anyone who cares about a vibrant, inclusive digital economy.