技术档案

AI定制芯片AI application-specific integrated circuit

为AI任务定制的芯片

每条都摘自下方编号的来源原文；只写事实，不写预测或建议。

技术简介

AI基础设施定制芯片是为企业自身AI工作负载专门设计的芯片与系统，而非标准商用产品。5
Custom silicon for AI infrastructure refers to chips and systems designed specifically for a company’s AI workloads, rather than purchased as standard commercial products.
ASIC是为特定计算任务设计的专用半导体芯片，而非通用计算芯片。6
For example, application-specific integrated circuits (ASICs), are specialized semiconductor chips designed for specific compute tasks, as opposed to general-purpose computing.

已有应用

Google为其数据中心开发的TPU是用于AI训练与推理大矩阵计算的专用加速器。3
Tensor Processing Units (TPUs), developed by Google for its data centers, are specialized AI accelerators designed to efficiently execute large-matrix operations used in both AI training and inference.
Google的第七代TPU Ironwood已向云客户提供。4
Ironwood our seventh-generation Tensor Processing Unit (TPU) is now available for Cloud customers.

主要参与者

联发科提供面向数据中心的定制ASIC设计服务。2
To build world-class custom data center ASIC solutions, MediaTek starts with a team of veteran chip design experts.
联发科为客户设计定制ASIC，用于构成XPU。3
At MediaTek, we design custom ASICs for our customers to create best-in-class XPUs.

未定事项

联发科表示2028年供应链安排仍在敲定。1
We're still in the process of finalizing the supply chain situation.
联发科称正与IDM伙伴及基板供应商合作，优先改善良率。1
And we are working very closely with IDM partner and substrate suppliers, first and foremost on the yield improvements.

来源