When OpenAI’s ChatGPT came out three years ago, Google was caught unawares at the lightning-fast popularity of the AI chatbot. Seemingly overnight, Google seemed to be facing an existential crisis.
But Google’s turning point came last August when a Google DeepMind AI project manager uploaded a new, ultra-fast image generator to an AI ranking platform under a placeholder name, ‘Nano Banana,’ according to The Wall Street Journal. Google sought to exploit a known weakness in ChatGPT – image generation.
Nano Banana, a combination of the project manager’s nicknames, quickly topped performance charts, went viral and exceeded internal usage expectations. By September, Google’s Gemini app became the most downloaded app in Apple’s App Store.
Within months, Google released its most powerful Gemini model yet, overtaking competitors on several benchmarks and briefly leapfrogging OpenAI in perceived technical leadership. Gemini’s momentum was supported by Google’s long-term bets on multimodal AI, custom tensor-processing units and organizational changes that unified its Brain and DeepMind teams.
The company also began successfully integrating generative AI into its core businesses, including search ads, paid Gemini subscriptions and in-house AI chips, while protecting its dominant search franchise. AI-driven features such as AI Overviews and AI Mode marked the biggest overhaul of Google search in years.