公共雲端中的客製化晶片——人工智慧和機器學習正在推動晶片設計的創新

Custom Silicon in Public Cloud- Artificial Intelligence and Machine Learning are Driving Innovation in Chip Design

出版日期: 2024年05月20日 | 出版商:

Frost & Sullivan | 英文 48 Pages | 商品交期: 最快1-2個工作天內

價格

簡介目錄

創新技術和新需求正在推動雲端領導者設計自己的客製化晶片

近幾十年來，對晶片的需求經歷了指數級成長。目前，由於產生人工智慧（AI）和機器學習（ML）等創新的工作量需求不斷增加，這些關鍵硬體組件的重要性甚至更高，這些創新也推高了能源消耗和成本。然而，疫情引發的供應鏈中斷和短缺暴露了該行業在應對這些課題方面的脆弱性，可能導致全球經濟癱瘓。認知到僅依賴基於 CPU 的運算和現成硬體的不永續性，主要科技公司現在正在設計自己的晶片，以控制關鍵組件、降低開支並擴展服務，從而贏得競爭優勢。客製化晶片提供差異化和獨特的功能，儘管初始成本較高，但它們有望透過消除對外部供應商的依賴和最佳化工作負載來實現長期節省。

在這項研究中，Frost & Sullivan 探討了晶片產業的整體演進、晶片生產和需求的全球背景，以及技術超大規模企業（尤其是雲端運算服務供應商）如何解決客製化晶片的設計和製造以滿足客戶的特定需求，特別是與人工智慧和機器學習相關的需求。

The demand for chips has experienced exponential growth in recent decades. Presently, the importance of these key hardware components is even higher due to increasing workload demands from innovations like generative artificial intelligence (AI) and machine learning (ML), which are also driving up energy consumption and costs. However, pandemic-induced supply chain disruptions and shortages have laid bare the industry's vulnerability in addressing these challenges, risking global economic paralysis. Recognizing the unsustainability of relying solely on CPU-based computing and off-the-shelf hardware, major technology companies are now designing their own chips to gain control over critical components, reduce expenses, and expand their services to earn competitive advantages. Custom chips offer differentiation and unique features-although at high initial costs, they promise long-term savings by eliminating reliance on external suppliers and optimizing workloads.

In this study, Frost & Sullivan explores the evolution of the chip industry as a whole, the global context of chip production and demand, and how technology hyperscalers, especially cloud computing service providers, are addressing the design and manufacturing of custom chips to meet the specific needs of their clients, particularly those related to AI and ML.