Our Partners
Direct partnerships with leading GPU manufacturers to power your AI workloads with advanced computing solutions.
Products
Alibaba Cloud
Alibaba Cloud, also known as Aliyun, is a cloud computing resource company that is a subsidiary of Alibaba Group. International registration in Singapore.
Products
Baidu Kunlun
Baidu Kunlun is a line of AI accelerators from the Chinese company Baidu, designed for cloud computing, edge devices, and AI tasks including neural network training and inference. Developments are based on the proprietary XPU architecture with a focus on high performance in NLP, computer vision, and multimodal model tasks. Architecture and Specifications Kunlun is built on Samsung Foundry's 14-nm process with 2.5D packaging and HBM2 memory (up to 16 GB, bandwidth 512 GB/s). The base K200 model delivers 256 TOPS in INT8, 64 TOPS in INT16/FP16 at TDP 150 W and PCIe 4.0 x8 interface; it outperforms NVIDIA T4 by 2–3 times in tests like BERT and YOLOv3 in QPS. The architecture includes XPU-SDNN for tensor operations and XPU clusters with SIMD cores for scalar tasks. New Developments In 2025, Baidu announced Kunlun M100 (for MoE model inference, launch in early 2026) and M300 (for training multimodal models, 2027). They support PaddlePaddle, TensorFlow, PyTorch frameworks via graph compiler and XPU C/C++. Tianchi 256 supernodes enable connectivity for up to 256 chips with 4x higher bandwidth than previous versions. Applications and Performance Kunlun chips are optimized for Ernie Bot and real-world tasks (NLP, vision, speech), with software masking differences from NVIDIA GPUs. They are used in Baidu's cloud and show 3x efficiency over FPGA/GPU in inference.
Products
Enflame Technology
Enflame Technology is a Chinese startup specializing in AI chips and cloud computing solutions. Founded in 2018 in Shanghai by former AMD employees, it positions itself as an alternative to NVIDIA in the GPGPU segment for neural network training and inference. Key Products Enflame produces high-performance chips like Suisi (DTU) for cloud data centers, focusing on deep learning. These GPGPUs are used in computer vision, NLP, and recommendation systems, with deliveries to Chinese state enterprises and local governments. In 2024, the company adapted designs to US restrictions, partnering with TSMC. Funding and Plans Enflame has raised significant investments: Series C (1.8 billion yuan in 2021 from CITIC, Primavera, Tencent) and Series D (2 billion yuan in 2023 from Shanghai International Group, Tencent, Meitu). It plans an IPO on Shanghai's STAR Market; Tencent co-developed the Zixiao chip for finance and transportation. In 2024, construction began on an AI hub in Wuxi. Market and Competitors The company competes with Cambrian, Horizon, and Deephi Tech in AI inference; its chips are applied in education, smart cities, and research. Enflame is recognized as a Challenger in the AI inference processors market alongside AMD and Samsung.
Products
Huawei Technologies
Huawei is a leading Chinese technology giant specializing in telecommunications, smartphones, networking equipment, and AI solutions. The company is actively developing its own GPUs and AI accelerators in response to US sanctions limiting access to NVIDIA technologies. Main Directions Huawei focuses on the Ascend chip series, such as Ascend 910C and 910D, positioned as alternatives to NVIDIA H100 and Blackwell. These accelerators are manufactured using SMIC's 7-nm process and are used in supercomputers like CloudMatrix 384 with 384 chips, providing up to 300 petaflops in BF16—almost twice as much as NVIDIA GB200 NVL72. The systems stand out with large memory capacity (up to 49.2 TB HBM, 3.6 times more than NVIDIA analogs) and high bandwidth (1229 TB/s). GPU Developments In 2025, Huawei shifted from specialized ASICs to universal GPUs with CUDA compatibility via an intermediate layer that converts NVIDIA code into commands for its chips. This reduces dependence on embargoes and simplifies migration for developers; conversion efficiency exceeds 95%. Plans include producing 600,000 accelerators in 2026, including Ascend 910D (4 chips in one) and 950DT. Technologies and Optimization Huawei's software masks differences between its chips and NVIDIA GPUs, doubling the utilization rate from 35% to 70% in AI clusters. Ascend 910C achieves 320–800 TFLOPS in FP16 at 310 W, though it lags behind NVIDIA in energy efficiency due to the process technology. In mobile devices, GPU Turbo is used to accelerate graphics by 60%.
Products
Iluvatar Corex
Iluvatar Corex: Leading Chinese GPGPU and AI Computing Solutions Provider Iluvatar Corex (Tianshu Zhixin) is a premier Chinese developer of high-performance general-purpose GPU (GPGPU) chips and integrated computing systems. The company is dedicated to creating independent, globally competitive, and sustainable computing engines to power the AI era. Core Product Lines The company’s hardware portfolio is built around three main series designed for diverse computational needs: Tiangai (天垓) Series: Specialized for high-end cloud training. The Tiangai 100, released in March 2021, was China’s first fully self-developed cloud-based training chip using a GPU architecture, marking a significant technological breakthrough. It supports traditional machine learning, encryption, and digital signal processing. Zhikai (智铠) Series: Focused on AI inference. With the launch of Zhikai 100 in December 2022, Iluvatar Corex became the only domestic provider capable of offering a complete "training + inference" solution under a unified GPU architecture. This chip supports multi-precision computing (FP32, FP16, INT8) and advanced video decoding. Tongyang (彤央) Series: Includes models such as the TY1000, TY1100, and TY1200. Software Stack: All hardware is supported by the proprietary Tianshu Zhisuan (天数智算) software stack. Community and Ecosystem Iluvatar Corex leads the DeepSpark Open Source Community. This platform is designed to build a collaborative ecosystem for AI and general computing, featuring hundreds of open-source algorithms and models tailored for various industry scenarios to help customers accelerate application deployment. Market Applications Their solutions are widely applied across multiple sectors, including: Finance and Retail: Data processing and analytics. Healthcare and Education: Intelligent diagnostics and educational platforms. Smart Manufacturing and Energy: Industrial automation and resource exploration. Autonomous Driving: Supporting vehicle-road coordination systems. Multimedia: Content generation, modeling, and rendering.
Products
MetaX Integrated Circuits
Muxi Integrated Circuit (Shanghai) Co., Ltd. was established in Shanghai in September 2020, and has established wholly-owned subsidiaries and R&D centers in Beijing, Nanjing, Chengdu, Hangzhou, Shenzhen, Wuhan and Changsha. Muxi shares have a team with complete technology and rich experience in design and industrialization. The core members have an average of nearly 20 years of end-to-end research and development experience in high-performance GPU products. They have led the research and development and mass production of more than a dozen mainstream high-performance GPU products in the world, including GPU architecture definition, GPU IP design, GPU SoC design and GPU systemThe whole process of mass production and delivery of the solution. Muxi shares are committed to providing full-stack GPU chips and solutions for heterogeneous computing, which can be widely used in cutting-edge fields such as smart computing, smart cities, cloud computing, autonomous driving, digital twins, meta-universe, etc., to provide strong computing power support for digital economic development.
Products