Models, Resources, Distribution, and Data—‘Gemini’ Equipped on All Fronts, Momentum Builds for an Overtake Beyond Chasing ChatGPT

Google’s ‘Gemini’ Puts the Brakes on OpenAI’s ‘ChatGPT’ Dominance
Rated as “the company with the most complete portfolio”
Control of the global AI ecosystem seen as only a matter of time

Google’s artificial intelligence (AI) model ‘Gemini’ is emerging as a formidable threat to competitors including OpenAI, armed with a balanced arsenal spanning technological capability, resources, distribution channels, and access to data, according to a new analysis. Once sidelined by the debut of ChatGPT, Google is now widely seen as having returned to the center of the AI race in just three years. As the competitive balance tilts toward Google, the company is also consolidating its grip on Apple’s Siri, further entrenching its dominance.

Gemini’s Counteroffensive, Closing in on the AI Throne With In-House Chips

On the 15th (local time), U.S. technology outlet The Verge reported in an analysis article titled “Gemini Is Gaining the Upper Hand” that Google appears to be the only company to have effectively secured all the critical elements required to win the AI race. The Verge noted that decisive advantage in AI competition requires, simultaneously, top-tier model performance, massive computing resources, services with large real-world user bases, and access to vast troves of user data. This context explains why OpenAI CEO Sam Altman has repeatedly stated publicly that “trillions of dollars” will be needed for computing resources alone.

By these criteria, Google is widely regarded as possessing the most complete portfolio. With the launch of its next-generation AI model ‘Gemini 3’ last year, Google is assessed as owning one of the most powerful large language models (LLMs) currently on the market. The model has posted strong results across multiple benchmarks, and there is broad consensus among experts that it delivers top-tier performance across a wide range of tasks. In practice, Gemini 3 has already begun to outperform OpenAI’s GPT-5.1 in multimodal processing and complex reasoning. Its image generation model ‘Nano Banana’ also boasts capabilities so advanced that photo editing software such as Photoshop is effectively unnecessary.

A key pillar of Gemini 3’s competitiveness is Google’s TPU, which the company has developed in-house for years. TPU, short for Tensor Processing Unit, is a custom semiconductor designed by Google specifically for AI computation. Google has been manufacturing TPUs for a decade. This has enabled the company to build a structure that optimizes AI training and inference using its own chips, without relying entirely on Nvidia’s graphics processing unit (GPU) supply chain. The Verge assessed that “Google is the only company with a full-stack structure that controls the entire AI pipeline end to end.”

Explosive Growth in Gemini Users

Gemini’s advance is also evident in the data. According to figures from global web analytics firm Similarweb, Gemini’s share of web traffic, which stood below 6% a year ago, surpassed 21% as of early this month. Over the same period, ChatGPT’s share fell from 85% to 65%. In the six weeks following the release of Gemini 3 last November, ChatGPT traffic declined by 22%, while weekly active users fell by approximately 158 million.

Experts expect this trend to accelerate further. Deepak Mathivanan, an analyst at global financial firm Cantor Fitzgerald, said, “In the long run, there may be room for two AI assistants to coexist, but the core competitive moat is likely to shift from ‘model intelligence’ to ‘practical usefulness.’” He added, “In this regard, Google, which has invested for more than a decade in improving and commercializing search, holds a significant advantage.” Influential figures in Silicon Valley are also increasingly backing Google. Last month, Salesforce CEO Marc Benioff said, “I used ChatGPT every day for three years, but after trying Gemini 3, I decided never to go back,” adding, “Google’s leap forward is insane.”

OpenAI, meanwhile, appears increasingly uneasy about Google’s Gemini-led offensive. Altman previously disclosed at a developer conference last October that ChatGPT’s weekly active users had surpassed 800 million, but sentiment has since cooled amid intensifying competition and rising cost burdens. Massive computing expenses required to maintain high-performance models have failed to translate into improved profitability. The Economist projected that OpenAI’s cash burn would surge from $9 billion last year to $17 billion this year, prompting Altman to declare a “code red” emergency early last month.

By contrast, Google already enjoys a data advantage over OpenAI, improving its odds in a prolonged contest. Sustained AI performance depends on continuous access to diverse training data. While OpenAI must source most data externally beyond user interactions within ChatGPT, Google operates the world’s leading search engine and video platform, YouTube, generating mountains of data daily. Moreover, as AI expands into real-world understanding and robotics applications, access to sensor-based and real-environment data becomes critical. Google’s ownership of smartphone hardware businesses and its autonomous driving subsidiary Waymo places it in a far stronger position in this data diversification race.

‘Hardware-Handicapped’ OpenAI Faces Mounting Risks

In terms of user reach, Google also holds a clear edge over OpenAI. Apple, which had long resisted external AI partnerships, joined forces with Google on the 12th. Apple urgently needed a partner to compensate for the limitations of its in-house AI, Apple Intelligence, while Google sought a decisive ally to pull further ahead of OpenAI. With interests aligned, the two companies have moved beyond a search partnership to establish an ‘AI alliance front’ centered on Gemini.

As a result, Google has effectively expanded Gemini’s potential footprint from roughly 3 billion Android devices worldwide to include Apple’s approximately 2 billion active devices, including iPhones and iPads. While the pace of rollout may vary by device, the structure now effectively allows Google AI to permeate the vast majority of smartphones on the planet.

This development is potentially devastating for OpenAI. Even with superior model performance, competing against OS-integrated, default AI assistants on mobile devices is an uphill battle. Even if ChatGPT ranks highly in app stores, the accessibility gap with built-in AI operating at the OS level is likely to widen over time.

Google has yet another strategic card to play: the recently unveiled ‘Personal Intelligence’ feature. When enabled by users, this function allows Gemini to generate responses using personal data held by Google, including search history, YouTube viewing records, Gmail, photos, and files. The implications are significant, as Google’s vast reserves of personal data effectively become part of Gemini’s intelligence. Users can receive deeply personalized responses without repeatedly explaining context, representing a fundamentally different approach from conventional chatbots. The feature is currently being piloted among a subset of paid users, but Google plans to expand it across its search services. One IT expert observed, “Google’s vision of turning Gemini into a gateway that connects personal data, the internet, and global information is becoming increasingly clear.”