微软将于下个月推出其首款人工智能芯片 以减少对英伟达的依赖
中文原文
10月7日消息，据外媒援引知情人士消息透露，微软计划在下个月举行的年度开发者大会上推出该公司首款专为支持人工智能（AI）而设计的芯片。此举是多年来努力的结果，可能有助于微软减少对英伟达AI芯片的依赖。随着需求的激增，英伟达的AI芯片一直供不应求。
微软的芯片类似于英伟达图形处理器（GPU），是为训练和运行大语言模型的数据中心服务器设计的，大语言模型是OpenAI的Chat GPT等对话式人工智能功能背后的软件。微软的数据中心服务器目前使用英伟达的GPU，为包括OpenAI和Intuit在内的云客户以及微软生产力应用中的人工智能功能提供支持。
这款芯片代号为“雅典娜”（Athena），可能会在11月14日于西雅图举行的微软Ignite大会上揭晓。雅典娜预计将与英伟达的旗舰级微处理器H100 GPU竞争，为数据中心的人工智能加速。这种定制芯片已经由微软及其合作伙伴OpenAI的团队秘密测试。
微软在2019年左右开始开发雅典娜芯片，寻求削减成本的同时，也希望增加与英伟达谈判的筹码。Azure目前依赖于英伟达的GPU来实现微软、OpenAI和云客户使用的AI功能。但有了雅典娜，微软可以跟随竞争对手AWS和谷歌的脚步，向云用户提供自主研发的人工智能芯片。
雅典娜的性能细节尚不清楚，但微软希望这款芯片能够与英伟达的H100相媲美。尽管许多公司都在兜售卓越的硬件和成本效益，但得益于该公司的CUDA平台，英伟达GPU仍然是人工智能开发者的首选。吸引用户使用新的硬件和软件将是微软的关键。
微软自主研发AI芯片还可能在GPU供应紧张的情况下减少对英伟达的依赖。据报道，在开始与OpenAI密切合作后，微软订购了至少数十万颗英伟达芯片，以支持OpenAI的产品和研究需求。通过使用自己的芯片，可以节省大量的成本。
OpenAI可能也在考虑减少对微软和英伟达芯片的依赖。近日曾有报道称，这家人工智能研究实验室正在考虑制造自己的人工智能芯片。OpenAI网站上最近的招聘信息也表明，该公司打算招聘人才来评估和共同设计人工智能硬件。
虽然微软和其他云服务提供商没有立即停止从英伟达购买GPU的计划，但从长远来看，说服他们的云客户更多采用内部芯片而不是英伟达的GPU服务器可能在经济上有利。微软还与AMD在即将推出的人工智能芯片MI300X上密切合作。随着人工智能工作量的激增，这种多样化的方法提供了多种选择。云计算的竞争对手也在采用类似的策略来避免供应商锁定。
亚马逊和谷歌已将其人工智能芯片战略性地整合到其云业务的促销活动中。亚马逊向OpenAI的竞争对手Anthropic提供资金支持的条件是，Anthropic将使用亚马逊的人工智能芯片，分别名为Trainium和interentia。与此同时，谷歌云宣布，人工智能图像开发商Midjourney和Character AI等客户使用了该公司的张量处理单元。
随着人工智能芯片成为数据中心的重要组成部分，押注该领域获得的回报可能很高。随着这一发展，微软也将加入竞争对手的行列，在加速发展的人工智能芯片领域争夺市场份额。有了雅典娜，微软可以为云客户提供更多的选择，同时在下一代人工智能基础设施方面制定更独立的路线。
英文译文
On October 7th, according to sources quoted by foreign media, Microsoft plans to launch its first chip specifically designed to support artificial intelligence (AI) at its annual developer conference next month. This move is the result of years of hard work and may help Microsoft reduce its dependence on NVIDIA AI chips. With the surge in demand, Nvidia's AI chips have been in short supply.
Microsoft's chips are similar to NVIDIA graphics processors (GPUs) and are designed for data center servers that train and run large language models, which are software behind conversational artificial intelligence features such as Open AI's Chat GPT. Microsoft's data center servers currently use NVIDIA's GPU to support cloud clients including OpenAI and Intuit, as well as artificial intelligence features in Microsoft productivity applications.
This chip is codenamed "Athena" and may be unveiled at the Microsoft Ignite conference in Seattle on November 14th. Athena is expected to compete with NVIDIA's flagship microprocessor H100 GPU to accelerate artificial intelligence in data centers. This custom chip has been secretly tested by a team from Microsoft and its partner OpenAI.
Microsoft began developing Athena chips around 2019, seeking to reduce costs while also increasing bargaining chips with NVIDIA. Azure currently relies on NVIDIA's GPU to implement the AI functionality used by Microsoft, OpenAI, and cloud clients. But with Athena, Microsoft can follow in the footsteps of its competitors AWS and Google, providing cloud users with independently developed artificial intelligence chips.
The performance details of Athena are not yet clear, but Microsoft hopes that this chip can rival NVIDIA's H100. Although many companies are touting excellent hardware and cost-effectiveness, thanks to the company's CUDA platform, Nvidia GPUs remain the preferred choice for artificial intelligence developers. Attracting users to use new hardware and software will be the key for Microsoft.
Microsoft's independent development of AI chips may also reduce its dependence on NVIDIA in the face of tight GPU supply. According to reports, after starting close cooperation with OpenAI, Microsoft ordered up to a few hundred thousand Nvidia chips to support OpenAI's product and research needs. By using your own chips, you can save a lot of costs.
OpenAI may also be considering reducing its reliance on Microsoft and NVIDIA chips. Recently, it was reported that this artificial intelligence research laboratory is considering manufacturing its own artificial intelligence chips. The recent recruitment information on the OpenAI website also indicates that the company intends to recruit talents to evaluate and jointly design artificial intelligence hardware.
Although Microsoft and other cloud service providers have not immediately stopped plans to purchase GPUs from Nvidia, in the long run, convincing their cloud customers to use internal chips more than Nvidia's GPU servers may be economically advantageous. Microsoft is also closely collaborating with AMD on the upcoming artificial intelligence chip MI300X. With the surge in artificial intelligence workload, this diverse approach provides multiple options. Cloud computing competitors are also adopting similar strategies to avoid supplier lock-in.
Amazon and Google have strategically integrated their artificial intelligence chips into their cloud business promotions. The condition for Amazon to provide financial support to OpenAI's competitor, Anthropic, is that Anthropic will use Amazon's artificial intelligence chips, named Training and Interentia, respectively. Meanwhile, Google Cloud announced that customers such as artificial intelligence image developer Midjournal and character AI have used the company's tensor processing units.
As artificial intelligence chips become an important component of data centers, the returns on betting in this field may be high. With this development, Microsoft will also join the ranks of competitors and compete for market share in the accelerating field of artificial intelligence chips. With Athena, Microsoft can provide cloud customers with more choices and develop more independent paths in the next generation of artificial intelligence infrastructure.
重点词汇
下个月 next month ; in the coming months ; coming] month ; last month ; ensuing
人工智能 artificial intelligence
多年来 for years ; over the years ; through the years
有助于 contribute to ; be conducive to ; conduce to ; go ; advantage ; favour ; tend to
英伟 Yingwei ; tall and handsome ; strapping
供不应求 Demand exceeds supply ; Supply falls short of demand ; Supply fails to meet the demand ; The supply is not adequate to the demand
类似于 be similar to ; resemble ; correspond ; be analogue to
图形处理器 GPU ; Graphic Processing Unit
语言模型 language model
数据中心 Data center ; datacenter ; IDC ; MDC
以上是秒词邦为您整理编写的文章《微软将于下个月推出其首款人工智能芯片 以减少对英伟达的依赖》的全部内容。秒词邦是国内权威分题型分考点背诵中高考/四六级考研/专升本/出国单词的专业单词软件。扫描如下小程序码，进入秒词邦官方小程序获取更多英语相关资料！ 【关键词：高考单词；高考英语；高中单词；高中英语；单词app；单词软件；记单词app；记单词软件；背单词软件；背单词app；英语单词；四六级单词；四六级英语；四六级单词app；四六级单词软件；考研单词app；考研单词软件；核心单词；高考冲刺复习；高考英语教材；高考英语真题；四六级真题；四六级试题；考研真题；考研英语单词；考研英语真题】