Google made a big splash in the AI world by launching Gemini 3 and promoting its Tensor chips. This move got the attention of both OpenAI and Nvidia.
On November 25, Nvidia posted on X: “We are excited about Google’s success — they have made great progress in AI and we continue to supply Google.” However, they added that “NVIDIA offers greater performance, flexibility, and interchangeability than ASICs” like Google’s.
At the same time, Sam Altman, CEO of OpenAI, congratulated Google publicly: “Congratulations to Google for Gemini 3! It looks like an excellent model.”
Meanwhile, ChatGPT has reached 800 million active users each week, according to OpenAI. The Gemini app has 650 million active users every month.
The “leap” of Gemini 3 and market reactions
Interest in Gemini 3 soared quickly as it was heavily promoted with the new Tensor chips. Salesforce CEO Marc Benioff shared on X that after testing Google’s model, he won’t go back to ChatGPT. He called the performance leap “crazy,” noting improvements in reasoning, speed, images, and videos.
Google is now a popular choice for collaboration. According to The Information, Meta is talking with Google about Tensor chip supply, while Anthropic plans to use more of Google’s technology.
On the stock market, Google’s shares rose nearly 8%, while Nvidia’s dropped just over 2% during the same time.
Who will dominate the AI ecosystem?
CNN highlights that the stakes go beyond a few deals or company images. AI is set to impact the economy, markets, and investment portfolios for everyone. The company that shapes the leading model could affect us all.
Nvidia’s post seems to downplay Google’s threat. Google’s chips are different and don’t directly replace Nvidia’s GPUs. Still, both Nvidia and OpenAI recognizing Google’s success shows they take these developments seriously.
“For now, they have the lead until someone else brings the next model,” says Angelo Zino from CFRA.
Google from “caught off guard” to protagonist
Even though Google has used AI technologies in its services (like Search and Translate) for years, the rapid rise of ChatGPT in 2022 caught them off guard. According to The New York Times, Google’s management issued a “red alert” in December that year.
Today, however, Gemini has become one of the top players alongside ChatGPT. ChatGPT has 800 million active users weekly while Gemini has 650 million monthly users, showing that the battle for dominance is now quite close.
Gemini 3: Benchmark leader – with caveats
Released on November 18, Gemini 3 tops many benchmarks for text creation, image processing, and text-to-image tasks. It outperforms Western competitors like ChatGPT, xAI’s Grok, and Anthropic’s Claude in specific areas.
Google reported that over 1 million users tested the model within the first day through its AI coding tools and APIs linked with other apps.
However, as Ben Barringer from Quilter Cheviot points out, users often choose different models based on their needs. In some tests focused on search capabilities, xAI and Perplexity models seem to outperform Gemini 3. “It doesn’t mean Alphabet will monopolize AI,” he says. “It’s a strong piece in an ever-expanding ecosystem.”
The Tensor chips versus Nvidia’s GPUs
Google has invested in Tensor chips for years before generative AI took off. However, Nvidia still leads the AI chip market. The company saw a sales increase of 62% year-on-year in October’s quarter and a profit increase of 65%.
The difference partly lies in product philosophy: Nvidia and AMD focus on GPUs—flexible units for complex computations across various applications. Google’s Tensors are ASICs—chips designed for specific uses with a narrower application scope.
As Jacob Feldgoise from Georgetown explains, ASICs are very powerful at their specific tasks but lack the general flexibility of GPUs.
Why Nvidia is still not threatened
Apart from chip performance, Nvidia has built an entire ecosystem: integrated solutions for data centers and software that helps developers optimize applications for its platforms. They also have strong partnerships with major market players—including Google itself.
“No one can really match Nvidia’s offering breadth right now,” notes Ted Mortonson from Baird.
While Google’s chips won’t replace Nvidia’s solutions anytime soon, more companies are turning to ASIC solutions. Increased competition from AMD shows a trend where major players are trying to lessen their dependence on one supplier.
Ben Barringer believes Google represents another strong “pillar” in a more multipolar AI market: “We likely won’t see a single winner but rather a new balance,” he concludes.
