The global AI wave is sweeping across the world, and China’s large-scale AI model companies are flourishing. Although Chinese AI enterprises started later compared to international giants like NVIDIA and Intel, they are making continuous breakthroughs as they grow. Companies like Infinigence AI and lisuantech have developed rapidly, while others such as VastaiTech, Birentech, Zhipuai, Moore Threads, and Enflame have secured investments from several state-backed funds.

China Internet Investment Fund Invests in Domestic GPU Maker VastaiTech

Recently, VastaiTech (Shanghai) Co., Ltd. underwent a corporate restructuring, with new shareholders including the China Internet Investment Fund (Limited Partnership), Guoshou (Shenzhen) Technology Innovation Private Equity Fund (Limited Partnership) under China Life, and Qingdao Yidun New Industry Investment Fund (Limited Partnership). At the same time, VastaiTech’s registered capital increased from approximately RMB 470 million to around RMB 520 million.

VastaiTech is a high-end GPU chip provider, offering full-stack chip solutions for core AI computing power, graphics rendering, content generation, and AI-generated content (AIGC).

Birentech Prepares for IPO, Backed by China Ping An and China Merchants Capital

On September 12, Shanghai-based Birentech released its “Initial Public Offering and Listing Counseling Report,” signaling its intention to go public. Birentech has completed its Series B funding round, raising over RMB 5 billion. Investors include Zhuhai Da Heng Qin Group, Gree Ventures, China Ping An, New World Group, Country Garden Ventures, Qiming Venture Partners, IDG Capital, among others.

Birentech’s focus is on general-purpose GPU (GPGPU) computing systems. The company launched the BR100 GPGPU chip, built on a 7nm process, and introduced a proprietary heterogeneous GPU co-training solution, which allows Chinese GPUs to coexist with NVIDIA GPUs.

State Capital Invests Again as ZhipuAI Completes New Round of Multi-Billion-Yuan Financing

In early September, ZhipuAI, a large AI model company founded only five years ago, announced the completion of a new financing round, raising several billion yuan. The lead investor is Zhongguancun Science City, a platform established by the Haidian District Government of Beijing.

ZhipuAI has completed 11 rounds of financing and is currently the highest-valued AI large-model unicorn in China. Its investors include well-known institutions such as Hillhouse Capital, Qiming Venture Partners, and Junlian Capital, alongside internet giants like Meituan, Alibaba, Tencent, and Xiaomi.

Enflame Initiates IPO Counseling, Backed by National IC Fund Phase II

On August 23, 2024, Shanghai-based Enflame signed an IPO counseling agreement with China International Capital Corporation, formally launching its A-share IPO process.

Enflame specializes in AI training and inference products. The company has developed the Suisi series of chips, Yunsui training and inference acceleration cards, and Yunsui intelligent computing systems, providing computing power support to internet companies, cloud service providers, network operators, research institutions, and local intelligent computing centers.

Infinigence AI Raises Nearly RMB 500 Million in Series A Funding

On September 2, Infinigence AI announced it had raised nearly RMB 500 million in its Series A funding round, marking the largest single financing record for a startup in China’s AI infrastructure sector. The round was co-led by the Social Security Fund’s Zhongguancun Innovation Fund, Qiming Venture Partners, and Hongtai Fund, with additional participation from Lenovo Ventures, Xiaomi, and other strategic investors.

According to public information, Infinigence AI, founded just 16 months ago, has raised nearly RMB 1 billion in total. Past investors include Sequoia China, Baidu, ZhipuAI, and Tongge Ventures.

Lisuantech Secures RMB 200 Million Investment from Dongxin Semiconductor

On August 19, Dongxin Semiconductor announced an external investment, stating it would use RMB 200 million of its own funds to increase its stake in Shanghai-based GPU company Lisuantech by subscribing to RMB 5 million in additional registered capital. After this capital injection, Dongxin Semiconductor will hold about 37.88% of Lisuantech’s equity.

Lisuantech is dedicated to developing scalable chip designs for multi-layer graphics rendering, meeting the chip and computing power needs in various scenarios such as digital content creation, gaming, animation, film production, AR/VR, digital twin cities and factories, cloud desktops, intelligent cockpits, AI, and large models.

Moore Threads Expands to Wuxi, Attracting Market Attention

Over the past two months, Moore Threads has drawn significant attention for two reasons. First, in September, the company announced plans to “focus on Wuxi strategically” and to establish a manufacturing headquarters for AI SoC chips and intelligent terminal manufacturing, as well as a headquarters for intelligent computing cluster business in Wuxi’s Huishan District.

In terms of financing, Moore Threads has already secured investments from government funds, leading venture capital firms, and capital across the industry chain, including Guosheng Capital, China Mobile, Shenzhen Capital Group, Sequoia China, Five Source Capital, GGV Capital, ByteDance, and Tencent.

Read more

• [News] China’s Core AI Industry Achieved a Market Size of Nearly CNY 600 Billion

(Photo credit: Birentech)

As the U.S. and its allies continue to impose technology restrictions on China’s semiconductor sector, Beijing has accelerated its efforts to develop homegrown alternatives. Chinese firms are aggressively pursuing advanced AI chip development, aiming to rival Nvidia, the global leader in AI semiconductors. A recent CNBC report highlighted seven Chinese companies to watch, including Huawei and Alibaba.

Huawei, the first of Nvidia’s Chinese challengers, is gaining attention with its new Ascend 910C AI chip, which is expected to compete with Nvidia’s H100.

Alibaba follows closely behind. After acquiring C-Sky Microsystems in 2018, the company integrated it with its in-house chip division to form T-Head. In 2019, T-Head launched its first AI inference chip, the Hanguang 800, which has since been deployed at scale in Alibaba’s hyperscale data centers.

Baidu ranks third with its self-developed AI chip, Kunlun. The chip has matured significantly, and in June, Baidu received a strategic investment from Beijing’s AI Industry Investment Fund, marking the first time a state-owned entity has invested in an AI chip firm, boosting Baidu’s growth prospects.

Biren Technology, in fourth, focuses on GPUs like Nvidia, with a software platform to build applications on top of its hardware. Biren’s Bili series of chips are designed for AI training in data centers. Last week, Biren registered for IPO guidance with the Shanghai Securities Regulatory Bureau, marking the start of its public listing journey.

Cambricon Technologies, ranked fifth, designs a wide range of semiconductors, from chips that train AI models to those running AI applications on devices. Known as China’s first AI chip stock, Cambricon has faced setbacks since being blacklisted by the U.S. in late 2022, with reports of large-scale layoffs last year.

Moore Threads, founded in 2020, is developing GPUs for training large AI models. Its data center product, MTT KUAE, integrates GPUs and is aimed at competing with Nvidia.

Enflame Technology, the seventh company on the list, positions itself as a domestic alternative to Nvidia, focusing on AI training chips for data centers. Enflame began IPO guidance on August 26, and is expected to list on the STAR Market alongside Biren either by the end of this year or early next year.

(Photo credit: Huawei)

According to a report from Tom’s Hardware, it’s reported that one of GPU giant NVIDIA’s key advantages in data centers is not only its leading GPUs for AI and HPC computing but also its effective scaling of data center processors using its own hardware and software. To compete with NVIDIA’s CUDA, Chinese GPU manufacturer Moore Threads has developed networking technology aimed at achieving the same horizontal scaling of GPU compute power with its related clusters, addressing market demands.

Moore Threads was founded in 2020 by former senior executives from NVIDIA China. After being blacklisted due to U.S. export restrictions, they were unable to access advanced manufacturing processes but continued to develop gaming GPUs.

Per another report from South China Morning Post, Moore Threads has upgraded the AI KUAE data center servers, with a single cluster connecting up to 10,000 GPUs. The KUAE data center server integrates eight MTT S4000 GPUs, designed for training and running large language models and interconnected using MTLink network technology, similar to NVIDIA’s NVLink.

These GPUs use the MUSA architecture, featuring 128 tensor cores and 48GB of GDDR6 memory with a bandwidth of 768GB/s. A cluster with 10,000 GPUs can have 1,280,000 tensor cores, though the actual performance depends on various factors.

However, Moore Threads’ products still lag behind NVIDIA’s GPUs in performance. Even NVIDIA’s 2020 A100 80GB GPU significantly outperforms the MTT S4000 in computing power.

Moore Threads has established strategic partnerships with major telecommunications companies like China Mobile and China Unicom, as well as with China Energy Engineering Group and Big Data Technology Co., Ltd., to develop three new computing clusters aimed at boosting China’s AI development.

Recently, Moore Threads completed a new round of financing, raising CNY 2.5 billion (roughly USD 343.7 million) to support its expansion plans and technological development. However, the inability to access advanced processes from TSMC, Intel, and Samsung presents significant challenges for developing next-generation GPUs.

Read more

• [News] ASML Former CEO Predicts Chip War Between China and the US May Persist for Decades
• [News] US to Lobby Netherlands and Japan to Contain China’s Chip Development, Focusing on HBM

(Photo credit: Moore Threads)