高通正在为 AWS 设计多代定制芯片，重点关注 AI 推理。这两家公司还在研究光互连，带宽可达 1.6 太比特，以应对 AI 基础设施中日益增长的数据流量。高通带来了其在节能芯片设计方面的经验，而亚马逊则提供其云基础设施。作为回报，高通使用 AWS 服务，如 Amazon Bedrock，加快芯片设计流程。
自 6 月以来，亚马逊成为高通的第三个大型数据中心客户。Meta 正在采用 Dragonfly C1000 服务器处理器：https://the-decoder.com/qualcomm-enters-the-data-center-market-with-its-own-processor/，微软则在 Azure 中部署高通的 HBC 内存架构。高通的目标是在 2029 年达到 150 亿美元的数据中心收入。
AWS 正在将高通的设计与其自有的 Trainium：https://the-decoder.com/amazons-nova-2-undercuts-openai-and-google-on-price-but-still-trails-top-tier-models/、Graviton：https://the-decoder.com/meta-buys-tens-of-millions-of-aws-graviton-5-processor-cores-from-amazon/ 以及 Nitro 芯片系列一起使用。这一举措瞄准推理工作负载，其中每个令牌的能源成本决定盈利。效率也是高通在另一端——边缘设备上的关注重点。今年 3 月，高通的 AI 研究部门发布了一个可以在智能手机上运行推理模型的框架：https://the-decoder.com/qualcomm-shrinks-ai-reasoning-chains-by-2-4x-to-fit-thinking-models-on-smartphones/。
保持对 AI 的关注。内容清晰、有用，无废话。
关注 The Decoder，获取 AI 新闻、背景故事和专家分析。
解码器：https://the-decoder.com/
Qualcomm is designing custom chips for AWS across multiple product generations, with a focus on AI inference. The two companies are also working on optical interconnects with bandwidth up to 1.6 terabits to handle the growing data traffic in AI infrastructure. Qualcomm brings its track record in power-efficient chip design, while Amazon contributes its cloud infrastructure. In return, Qualcomm uses AWS services like Amazon Bedrock to speed up its chip design process.
Amazon is Qualcomm's third big data center win since June. Meta is taking the Dragonfly C1000 server processor：https://the-decoder.com/qualcomm-enters-the-data-center-market-with-its-own-processor/, and Microsoft is deploying Qualcomm's HBC memory architecture in Azure. Qualcomm aims to hit $15 billion in data center revenue by 2029.
AWS is adding Qualcomm's designs alongside its own Trainium：https://the-decoder.com/amazons-nova-2-undercuts-openai-and-google-on-price-but-still-trails-top-tier-models/, Graviton：https://the-decoder.com/meta-buys-tens-of-millions-of-aws-graviton-5-processor-cores-from-amazon/, and Nitro chip families. The move targets inference workloads, where energy cost per token drives the bottom line. Efficiency is also Qualcomm's focus at the other end of the stack, on edge devices. In March, Qualcomm's AI research division released a framework that runs reasoning models on smartphones：https://the-decoder.com/qualcomm-shrinks-ai-reasoning-chains-by-2-4x-to-fit-thinking-models-on-smartphones/. Ad Ad
Stay in the loop on AI. Clear, useful, no fluff.
Follow The Decoder for AI news, background stories and expert analyses.
The Decoder：https://the-decoder.com/
扫码直达文章详情