NVIDIA

A leading supplier of GPUs, accelerated computing platforms, and software used to train and run advanced AI models.

Key technologies: Blackwell architecture, data-center GPUs, CUDA, networking.

AMD

Develops data-center GPUs and CPUs for AI computing, high-performance computing, and enterprise infrastructure.

Key technologies: Instinct accelerators, EPYC processors, ROCm software.

Google

Designs Tensor Processing Units (TPUs) for AI workloads within Google’s infrastructure and Google Cloud.

Key technologies: TPU accelerators, AI infrastructure, custom silicon.

Amazon Web Services

Develops custom silicon to support AI model training and inference on AWS cloud infrastructure.

Key technologies: Trainium and Inferentia families.

Apple

Designs its own chips for smartphones, computers, and other devices, including neural processing capabilities.

Key technologies: Apple silicon, Neural Engine, M-series and A-series chips.

Intel

Develops processors and accelerator technologies for data centers, enterprise computing, and AI workloads.

Key technologies: Xeon processors and Gaudi AI accelerators.

Cerebras Systems

Builds wafer-scale computing systems designed for large AI models and demanding machine-learning workloads.

Key technology: Wafer-Scale Engine processors and AI systems.

Graphcore

Developed Intelligence Processing Units (IPUs) for machine learning and parallel computing; now part of SoftBank Group.

Focus: AI processor architecture and computing systems.

Qualcomm

Develops mobile, edge, automotive, and computing processors with on-device AI capabilities.

Key technologies: Snapdragon platforms and Hexagon neural processing.

Broadcom

Supplies networking and semiconductor technologies and works on custom accelerators for large-scale computing.

Focus: Custom AI silicon and high-speed data-center networking.

Marvell

Develops data infrastructure semiconductors, including custom compute and connectivity technologies used in AI systems.

Focus: Custom silicon, interconnects, and data-center infrastructure.

Tenstorrent

Develops AI processors and computing systems based on its processor and RISC-V technology approach.

Focus: AI compute, processor IP, and accelerator systems.

TSMC

The world’s leading dedicated semiconductor foundry, manufacturing advanced chips designed by many AI and technology companies.

Focus: Advanced process technologies and chip packaging.

Samsung Semiconductor

Produces logic chips and memory technologies used across computing, mobile devices, and AI infrastructure.

Focus: Semiconductor manufacturing, DRAM, and advanced memory.

GlobalFoundries

Manufactures semiconductors for communications, automotive, industrial, and other specialized computing applications.

Focus: Specialty semiconductor manufacturing.

ASML

Supplies advanced lithography equipment used by semiconductor manufacturers to produce leading-edge chips.

Focus: EUV lithography and semiconductor manufacturing technology.

SK hynix

A major memory manufacturer supplying DRAM and high-bandwidth memory technologies used in advanced computing systems.

Focus: HBM and high-performance memory.

Micron Technology

Develops memory and storage products for data centers, computing, networking, and AI systems.

Focus: High-bandwidth memory, DRAM, and data-center memory.

Samsung Memory

Provides memory technologies used in AI accelerators, servers, and high-performance computing systems.

Focus: DRAM, HBM, and advanced memory products.

Huawei

Develops Ascend AI processors and associated computing systems for AI training and inference.

Focus: Ascend processors and AI computing infrastructure.

Moore Threads

A Chinese semiconductor company developing GPUs and computing products for graphics and AI workloads.

Focus: GPU and accelerated computing technologies.

MetaX

A Chinese semiconductor company focused on GPU products and computing accelerators.

Focus: GPU and AI computing solutions.

Cambricon

Develops AI processors and accelerator technologies for intelligent computing applications.

Focus: AI chips and computing systems.

Biren Technology

A Chinese chip-design company focused on general-purpose GPU and AI computing technologies.

Focus: GPU and high-performance computing products.

Alibaba

Develops cloud services and custom semiconductor technologies for its broader computing ecosystem.

Focus: Cloud infrastructure and in-house chip development.

AI Training

Training large models requires substantial parallel computing power, fast memory, and high-speed connections between processors.

AI Inference

Inference chips run trained models for chatbots, assistants, recommendations, image generation, and other real-world applications.

Memory and Networking

High-bandwidth memory, networking, and data movement help keep AI accelerators supplied with the information they need.

Manufacturing and Packaging

Advanced chip fabrication and packaging help turn chip designs into practical, high-performance computing products.

Software Ecosystems

Compilers, libraries, drivers, and developer tools influence how easily AI models can run efficiently on different hardware.

Power and Cooling

AI servers require significant power and thermal management, making energy efficiency an important part of system design.