Jensen Huang, CEO of Nvidia, holds a press briefing at Gyeongju Arts Center in North Gyeongsang Province on October 31 in the afternoon./Courtesy of News1

Nvidia moved to check Google's progress in developing artificial intelligence (AI) chips.

Nvidia wrote on the social media (SNS) platform X (formerly Twitter) on the 25th (local time) that it was "glad for Google's success," but added, "We are a generation ahead of the industry and provide the only platform that runs every AI model and works everywhere computing happens." As projections emerged that Google could supply its AI inference chip "Tensor Processing Unit" (TPU) to big tech companies such as Meta, the company moved to draw a line by stressing a "technological gap." Nvidia also added, "Google has made significant progress in AI, and we continue to supply products to Google."

Google's TPU is an application-specific integrated circuit (ASIC) designed for a single purpose. Nvidia said its own products "deliver better performance, versatility, and fungibility than ASICs designed for specific AI frameworks or functions."

Nvidia, via its official X account, checks Google and stresses that its product is a generation ahead of the industry./Courtesy of X Capture

Nvidia controls more than 90% of the AI chip market through its graphics processing units (GPUs). Google has emphasized that the TPU, which it has manufactured for a decade, can be an alternative to GPUs. That view has been spreading since Google unveiled the AI model "Gemini 3," which runs on TPUs, earlier this month. Gemini 3 is being assessed as outperforming OpenAI's ChatGPT.

Google had until now allowed TPUs to be used only through its own cloud. Anthropic, which operates the AI chatbot "Claude," also signed a cloud usage agreement late last month that equips 1 million of Google's TPUs.

However, after Google recently released the 7th-generation TPU "Ironwood," signs have emerged that it is shifting policy toward supplying the product directly. As word spread that potential TPU customers such as Meta are considering deploying TPUs in their own data centers in place of GPUs, some analysts said Nvidia's stronghold could be shaken.

Huang Jen-hsun, Nvidia's chief executive officer (CEO), recently said after an earnings release regarding Google's TPU, "Google is a customer, and (Google's AI model) Gemini also runs on Nvidia technology."

※ This article has been translated by AI. Share your feedback here.