· Enterprise AI · 3 min read
Why Google Won't Monopolize the LLM Market
Even if Gemini keeps pulling ahead, spillover science, commoditized data, and rival coalitions will keep the market competitive.
With a big jump in the quality of Gemini 3 compared to previous models and competitors, one scenario is that in the next few years, Google LLM models become so good that they are way ahead of competitors. This means every reasonable customer would pick Gemini over all other options, and Google takes virtually all of the market share.
That is unlikely to happen, not only for Google but for any other AI lab. Here are a couple of reasons:
LLM training techniques have high spillover: Science is built layer by layer; a breakthrough is built based on other foundational research. So it is hard to find a “secret sauce” technique out of nowhere that other AI labs can’t find out. Besides, talent moves freely between labs, and the knowledge moves with them too.
Data is the key for good models, but is a commodity now: LLMs mainly rely on text data, and the world’s text data is mostly used up by frontier models. Google has good sources of proprietary data (YouTube, user search intent), but other AI labs (Microsoft, Meta, Apple, and even OpenAI now) have proprietary data too. It is also unclear how much gain non-text data could contribute to LLM performance.
TPUs are an advantage, not a moat: Designing their own AI chip (TPUs) is a big advantage in terms of cost and performance for Google. But rivals can design their own chips too. Other AI labs entered the chip designing race late, and so have to resort to GPUs (which were originally built for graphics and gaming, not for AI). New AI-specialized chips will spawn very soon and catch up with TPUs. The situation is similar to AWS being ahead for a decade in Cloud Computing, but Azure and GCP caught up in a couple of years.
Teaming up between rivals: If Google starts to dominate, other AI labs will start joining forces. Some have deep pockets and vast amounts of data (Microsoft, Meta), and some have good model momentum (OpenAI, Anthropic, xAI). Any coalition could pose a serious challenge to Google.
Attracting antitrust scrutiny: Laying low to avoid an antitrust breakup is already in the playbook of Google, as earlier this year they avoided the forced selling of Chrome in an antitrust case. Google doesn’t want another summons to Washington DC for their Gemini monopoly.
I still keep my prediction from early this year that Google will get the biggest pie in the LLM provider market, but only up to 60%. They can’t and don’t want to become the monopoly in this game.
