In October, the domestic AI compute chip sector witnessed a surge in funding activity. Several startups announced billion-yuan-level fundraising rounds, with distinct technological focuses. Beyond general-purpose computing, differentiated paths such as processing-in-memory (PIM) and embodied intelligence have also secured sustained capital investment.
Huixi Technology Secures Over CNY 2.5 Billion in Cumulative Funding This Year, Pivoting to High-Compute Chips for Embodied Intelligence
On October 22, Huixi Technology, a provider of high-compute chips for embodied intelligence, announced that following the completion of its Pre-B+ and Series B rounds, the company's cumulative equity financing in 2026 has exceeded CNY 2.5 billion. The latest round was led by the Beijing Artificial Intelligence Industry Investment Fund, with participation from the Beijing Robotics Industry Development Investment Fund, Yinshan Capital, Zhengxin Gu Capital, Jinpu Investment, and Jiantou Investment. Existing shareholders including Shunwei Capital, YuanSheng Capital, and KaHui Capital participated in the follow-on investment.
Huixi Technology will primarily allocate the new funds to the R&D of next-generation high-compute chips for embodied intelligence, while accelerating large-scale delivery and industrial ecosystem construction.
Founded in 2022, Huixi Technology initially focused on advanced autonomous driving chips. In October 2024, it launched the Guangzhi R1, a chip manufactured using a 7nm automotive-grade process. According to Economic Observer, Huixi Technology has now repositioned itself as a supplier of high-compute chips for embodied intelligence.
During the 2026 World Robot Conference in August, the company unveiled its Huixi Embodied compute product series for embodied intelligence, based on the Guangzhi R1 and accompanying computing platforms. The Guangzhi R1 chip series has cumulatively received orders for thousands of domain controllers in the embodied intelligence sector. Deployments cover humanoid robots for industrial use (such as AgiBot's Spirit G2), factory production line robots (Leju's Kuavo 4 Pro), and L4 unmanned logistics vehicles by Zelos. The company is expanding into scenarios such as mining and industrial inspection.
Reexen Technology: Raises CNY 1 Billion in Six Months, Attracting Strategic Investors Including Meituan, XPeng, and miHoYo
According to reports from the SSE STAR Market Daily, Reexen Technology, an AI chip company specializing in PIM fusion, announced the completion of cumulative financing totaling CNY 1 billion. Strategic investors such as Meituan Strategic Investment, XPeng Motors (via Xinghang Capital), YMTC Industrial Fund, Yuanhe Puhua, and OmniVision Group have successively entered. Recently, Zhiwei Capital and miHoYo Strategic Investment joined, while existing shareholders Yinghao Capital and Weihao Chuangxin made excess additional investments.
Founded in 2018 and headquartered in Shenzhen, Reexen Technology focuses on the R&D of neuromorphic sensing-storage-computing integrated chips. Founder and CEO Liu Hongjie stated that the company entered both the SRAM-based PIM and HBF (High Bandwidth Flash) fields before consensus formed, aiming to enable trillion-parameter large models to transition from clusters to single machines/cards, and eventually to every terminal device.
The funds from this round will mainly support the R&D and tape-out of HBF and SRAM PIM fusion chips, the MPM (Model-Processor-Memory) software stack, and product/ecosystem development for cloud, edge, and end devices. The company's developing product is a 3D fused PIM chip vertically stacking SRAM-based PIM with HBF.
Reexen Technology claims that HBF will be equally important as HBM (High Bandwidth Memory). A single card can support the deployment of large models exceeding xTB in size, reducing costs by more than tenfold at equivalent performance levels. The company has initiated Alpha cooperation with a leading vendor's HBF AI chip system, is coordinating with multiple internet companies for cloud solutions, and collaborating with vivo, Honor, and Lenovo for end-side applications. Its PIM products are already in mass production and sales.
Yizhu Technology: Completes Over CNY 1 Billion Financing, 3D DRAM PIM Chip Has Been Taped Out
On October 23, Suzhou-based AI high-compute chip company Yizhu Technology announced the completion of a new financing round exceeding CNY 1 billion. The funds will be used for R&D investment, product iteration, and industrialization promotion. Founded in June 2020 and officially operational since 2022, Yizhu Technology is headquartered in Suzhou, with subsidiaries in Shanghai, Hangzhou, Chengdu, and Changsha.
Public records indicate that Yizhu Technology has chosen a technical path combining general-purpose PIM with super-heterogeneous architectures, centering on 3D DRAM. It pioneered a fully digital general-purpose PIM architecture, covering a full-stack matrix from compute cards and workstations to all-in-one machines and inference clusters. The company also launched its self-developed YICA software platform, compatible with mainstream ecosystems like CUDA and PyTorch.
Regarding product progress, Yizhu Technology disclosed that its fully digital general-purpose PIM AI chip based on 3D DRAM has been taped out. Having passed full-chain verification from theoretical architecture to chip hardware, commercial chip test results have basically met preset design specifications.
In terms of financing history, Yizhu Technology has completed multiple rounds since its inception. In October 2024, the company announced the completion of a several-hundred-million-yuan financing round led by overseas fund Prosperity7 Ventures, with follow-on investment from Shengshi Technology and Xingzhi Capital. In October 2025, it completed another round involving Xingxiang Capital, ABC International, Yingke Zhidi, and Heding Gong Capital.
AI Compute Chips: Fierce Competition Among Players
Viewed over the full year, the AI compute chip sector has seen multiple ultra-large financing rounds of similar magnitude in 2026.
In April, inference GPU company Sunrise Technology announced the completion of a new financing round exceeding CNY 1 billion. In June, PIM chip company Micro-Nano Core completed a Series B financing package totaling over CNY 1 billion. In September, RISC-V cloud compute chip company YiXing Intelligent completed a financing round of nearly CNY 2 billion. Combined with the three rounds announced in October, the year's large financings cover various sub-sectors including general inference GPUs, PIM, cloud RISC-V, and embodied intelligence computing, continuously broadening the boundaries of capital allocation.
Currently, as applications such as large models, intelligent agents, and humanoid robots accelerate their rollout, and intelligent computing infrastructure continues to advance, market demand for differentiated compute solutions is rising rapidly, boosting the popularity of the domestic AI compute chip sector. Financing amounts in the AI compute chip field are relatively large, with billion-yuan levels being common due to the naturally high capital barriers inherent in this industry.
From an industry perspective, after years of iteration, the domestic AI compute chip landscape has formed a clearly stratified vendor structure. In the mature commercialization tier, vendors such as Huawei Ascend, Hygon Information, Cambricon, Moore Threads, MetaX, and Biren Technology have products entering stages of scaled delivery and cluster deployment, covering training, inference, and general graphics computing scenarios. On the path of differentiated innovation, companies like Reexen Technology, Yizhu Technology, and Houmo Intelligent focus on PIM and PIM fusion solutions, aiming to bypass traditional GPU bandwidth bottlenecks through architectural innovation. Companies like Huixi Technology and YiXing Intelligent anchor themselves in niche markets such as embodied computing for humanoid robots and RISC-V cloud computing, respectively, to open up new territories.
AI compute chips represent an unavoidable hard-tech segment in the AI supply chain, characterized by high R&D thresholds and significant challenges. Consequently, both corporate teams and industrial capital continue to invest heavily and push forward relentlessly. Multiple technical paths are running in parallel; some focus on deepening general-purpose computing, while others bet on new directions like PIM and embodied intelligence. Who will truly achieve commercial viability by delivering stable, user-friendly, and cost-effective products remains to be seen, warranting continued observation of future developments.