Nvidia在2026年3月GTC公布新的AI推论芯片,提高速度和效率,OpenAI作为主要客户。 Nvidia unveils new AI inference chip at March 2026 GTC, boosting speed and efficiency, with OpenAI as a major customer.

Nvidia正在开发一个新的AI推论芯片,准备在2026年3月的GTC会议上首次启用,将Groq启动后的技术结合起来。 Nvidia is developing a new AI inference chip, set to debut at its March 2026 GTC conference, integrating technology from startup Groq. 芯片旨在提高AI模型的反应速度和效率,满足对实时处理日益增长的需求。 The chip aims to improve response speed and efficiency for AI models, addressing growing demand for real-time processing. 预计OpenAI将成为主要客户,获得3千兆瓦的专用推断能力,而Nvidia与Groq的200亿美元许可协议结束了OpenAI与初创公司的先前谈判. OpenAI is expected to become a major customer, securing 3 gigawatts of dedicated inference capacity, while Nvidia’s $20 billion licensing deal with Groq ended OpenAI’s prior talks with the startup. 尽管谷歌,亚马逊和其他提供推断优化芯片的竞争日益激烈, 但此举强化了Nvidia在AI硬件方面的统治地位. The move strengthens Nvidia’s dominance in AI hardware, despite rising competition from Google, Amazon, and others offering inference-optimized chips.