The $100 billion race: who dominates generative AI?
A simple question: when you use ChatGPT, Gemini or Claude, what exactly are you interacting with?
Gen AI market by the numbers
Before 2022, generative AI wasn't a meaningful commercial market. However, ChatGPT's launch in November of that year changed everything. Since then, investment has surged, models have multiplied and valuations have reached new heights.
The global market size is estimated to reach $104 billion in 2025. In 2023, this figure was estimated to reach $8 billion.
The three layers of the market
As of today, the generative AI market is divided into three interconnected layers, each of which needs the support of the other two to function properly.
The infrastructure layer is dominated by a single company: NVIDIA, whose GPUs power virtually every major AI model in existence. At the cloud level, Amazon, Microsoft, and Google have built the data centres where models are trained and run. And at the application layer, the race is wide open — with thousands of companies building on top of foundation models.
Who's winning the model race?
At the foundation model layer — the most visible and contested — OpenAI maintains a significant lead, but the gap is closing.
These numbers have evolved consistently as new leaders have emerged in recent years and entered the race.
In 2025, OpenAI holds roughly 23.6% of the market — a commanding position built on ChatGPT's massive user base (700 million users as of mid-2025) and its enterprise API dominance. The top five players collectively hold 58.1% of the market, meaning the remaining 42% is fragmented across hundreds of smaller players.
Open models have become serious competitors. Mistral's models, for instance, consistently punch above their weight relative to their parameter count — and the company has positioned itself explicitly as a European, privacy-respecting alternative to American giants. Meta's Llama 3 has been downloaded tens of millions of times by developers worldwide.
What this means for businesses (and for you)
For any organisation considering adopting generative AI tools, the market concentration raises a real strategic question: you're likely building on infrastructure controlled by a handful of American companies. That creates dependency risks — on pricing, on data handling, on geopolitical decisions you can't influence.
The practical takeaway? The market is consolidating fast at the top, but the application layer remains wide open. The real competitive advantage won't come from picking the "best" model — it'll come from knowing how to use any of them well.