I've been thinking about this for a while and remembered a good analogy from the car industry. Remember electric cars? Tesla's stock hits $1,000, they capture the market, and BMW or Mercedes seem to "do nothing." In the AI world, OpenAI is Tesla, and Google is BMW.
Google actually invented the Transformer architecture (the "T" in GPT) back in 2017. They had the technology but no product. OpenAI had an imperfect product, but Altman could afford to disrupt the market by shipping a raw, hallucinating ChatGPT. Google (like BMW) couldn't do that — they need quality, safety, and integration, or they'll get sued and tank their stock. So they weren't in a rush to pour billions into their own technology, afraid of cannibalizing Search (even though there were internal champions who believed in the LLM idea).
ChatGPT's launch was Google's wake-up call. Sergey Brin literally came back to the office to write code (there were rumors he was committing fixes to Gemini), they declared Code Red internally — everyone working hard and fast because the competitor was way ahead.
Looking at December 2025 traffic charts where Gemini is eating into OpenAI's share, it seems like the strategy worked. They let startups skim the first hype cream, watched the mistakes others made, and now they're simply rolling out their machine to billions of Android and Workspace users. The exact same thing happened with BMW/Mercedes vs Tesla.
In the Valley they say: "Google is not first, but Google is usually last standing." Hard to disagree. In my opinion, Gemini 3 Pro and Nano Banana Pro are the best models on the market. Better yet — affordable and even free in Google AI Studio.