Google is giving a new acceleration to the global race for artificial intelligence putting under pressure OpenAI on the front of generative models e Nvidia on the chip side. The launch of Gemini 3, presented on November 18 as the company’s “smartest” model, it has garnered very positive ratings and reopened competition at the highest levels of the industry.
According to Axios, Gemini 3 obtained benchmarks superior to GPT-5.1 in various areas of reasoning, coding, multimodality thanks also to the immediate integration into the Google ecosystem: from Search with AI mode to Android. Added to this is a gigantic amount of data coming from different entities in its sphere, such as YouTube. In short, everything that Alphabet has built over the years could now guarantee a decisive role in the most talked about and discussed sector of the tech and financial world.
Unlike the more famous and used OpenAI model (ChatGPT has exceeded 700 million users per week), Gemini 3 can be rolled out to billions of users without going through dedicated apps or additional onboarding. And the first market reactions confirm a small shift in inertia and interest. Salesforce CEO Marc Benioff, after a two-hour test, wrote on X: “I’m not going back.” He won’t go back. The reason is simple and multiple: “Jumping is crazy: reasoning, speed, images, videos.”
Not all experts, however, are totally convinced of the solidity of the model which, like its competitors, is not at all stingy with “hallucinations”. However, Google’s advance has generated a certain nervousness in OpenAI. A Sam Altman’s internal note to employeesreported by newspapers such as The Information, talks about “rough vibes”, bad weather, and “temporary economic headwinds”, economic headwinds. All signs that indicate a more complex phase for the company. The GPT-5.1 model, launched in August with mixed performances, it did not fully convince several sector analysts, contributing to the perception of a slowdown phase.
The difference could be entirely in the pre-training, a decisive step on which Google has returned to focus with dedication. In addition to the competition on models, there is also competition on AI chips. According to several newspapers, Google is proposing to large groups the adoption of its TPUs, the proprietary chips already used in Google Cloud data centers. Among the potential customers there would also be Meta, while other large financial institutions would have been involved in the discussions. A move perceived as direct challenge to Nvidiathe undisputed leader in AI hardware, which today recorded a decline in trading on Wall Street (up to -5%), a sign of growing competition.
Google, in a statement reported by the Wall Street Journal, says it has one demand “strongly accelerating” for both TPUs and Nvidia GPUs, confirming a “dual track” strategy that aims both to strengthen its hardware platform and to intercept new customers. Overall, the Gemini 3 advance and chip offensive suggests a rebalancing phase in the generative AI industry. Google enjoys structural advantages – cloud infrastructure, computing power, immediate deployment – while OpenAI retains strong audience loyalty and is working on next-generation models designed to correct current limitations. But the competitive picture, for the first time in three years, seems less unbalanced.