跳到正文
The Elec· JEONG IL JOO·· 25 天前AI 評分66

Nvidia 將向 Meta 投資 $3 billion,並擴大 AI 合作

Nvidia to Invest $3 Billion in Meta, Expand AI Partnership

AI 導讀

Nvidia 表示將向 Meta 的增資投資 $3 billion,並與 Meta 合作提升 Muse 系列 AI 模型的訓練與推理效能。Meta 同日宣佈將發售 250 million 股 A 輪存託股份,每股 $50,並發售價值 $25 billion 的 A 類普通股,每股 $625。

正文

Nvidia and Meta are strengthening their collaboration on AI models. (Image: Nvidia)

Nvidia and Meta are strengthening their collaboration on AI models. (Image: Nvidia)

Nvidia and Meta are strengthening their collaboration on AI models. (Image: Nvidia)
Nvidia and Meta are strengthening their collaboration on AI models. (Image: Nvidia)

Nvidia said Oct. 26 local time that it will invest $3 billion (approximately 3.98 trillion won) in Meta's capital increase. Meta announced the same day that it would offer 250 million Series A depositary shares at $50 per share and $25 billion (approximately 33.2 trillion won) worth of Class A common stock at $625 per share.

The companies will collaborate to improve the training and inference performance of Meta's Muse family of artificial intelligence (AI) models. The partnership will focus on optimizing Nvidia's AI computing infrastructure for Muse, maximizing performance across AI computing workflows while reducing the total cost of ownership (TCO) of data centers.

Meta will expand purchases and deployment of Nvidia's Vera Rubin AI computing infrastructure and Vera central processing unit (CPU) server racks in its data centers. It will also integrate Nvidia's Spectrum-X high-performance Ethernet networking technology and BlueField-4 data processing units (DPUs). Meta plans to use Nvidia's hardware and technologies to efficiently handle large-scale AI computing workloads.

Meta is also considering adopting Nvidia's OpenShell reference architecture, which provides infrastructure covering the development and operation of AI agents. It combines Nvidia hardware with OpenShell, a secure runtime software platform. The companies are also discussing the use of Nvidia's Sentry integrated security software framework. Both technologies are designed to isolate and monitor AI agents.

"AI agents represent a new computing paradigm that understands our goals, reasons and independently uses tools to perform tasks on our behalf," said Jensen Huang, founder and chief executive officer of Nvidia. "We will optimize the entire computing infrastructure to run AI agents efficiently and securely, helping billions of people achieve greater accomplishments and ambitions."

來源:The Elec · thelec.net