在AI算力赛道，NVIDIA早已凭借Hopper、Blackwell等架构GPU，在AI训练领域建立起难以撼动的优势。但随着语音对话、实时翻译等即时AI场景需求爆发，传统GPU在推理延迟上的短板逐渐凸显。近日，据行业消息披露，NVIDIA正瞄准这一痛点，计划于2028年推出新一代Feynman（费曼）架构GPU——这款产品将首次引入X3D堆叠设计，整合G ...

作为继Hopper、Blackwell之后的最新一代AI计算平台，NVIDIA Rubin不仅是一次硬件升级，更是一场由六款全新芯片深度协同构建的系统级革命。其核心目标直指当前AI发展的最大瓶颈：高昂的推理成本与训练效率。

TL;DR: NVIDIA is reducing production of its B40 AI GPU for China from 1.5-2 million to 900,000 units in 2025, as Chinese AI firms favor RTX 5090 gaming GPUs, Hopper AI chips, and local alternatives.

The pressure point is evident in Nvidia's stock price, which has mainly flatlined since the fall despite a 37% return in 2025 ...

Having an annual cadence for the improvement of AI systems is a great thing if you happen to be buying the newest iron at ...

在今日的消费电子展（CES）上，Nvidia 首席执行官黄仁勋正式发布了公司全新的 Rubin 计算架构，他称其为当前人工智能硬件的尖端之作。Rubin 架构目前已经进入生产阶段，预计将在下半年加速推广。

Nvidia has been able to increase Blackwell GPU performance by up to 2.8x per GPU in a period of just three short months.

Nvidia’s long-rumored RTX 50 Super refresh is not arriving on cue, and the pause is not a minor scheduling slip. The company ...

NVIDIA announced that GPU cloud platform CoreWeave is among the first cloud providers to bring NVIDIA GB200 NVL72 systems online at scale, with Cohere, IBM and Mistral AI using them for model training ...

SINGAPORE, Jan. 14, 2026 (GLOBE NEWSWIRE) -- Bitdeer AI, part of Bitdeer Technologies Group (NASDAQ: BTDR), an emerging AI cloud service provider and NVIDIA Cloud Partner, today announced the launch ...