- Memory chip leader joins Xiaomi and other backers in supporting Hangzhou-based computing-in-memory developer
- Company targets AI inference market with architecture designed to reduce power consumption and data-transfer bottlenecks
Nano-Core Chip (微纳核芯), a Hangzhou-based AI semiconductor startup, has completed a Series B+ funding round, with the addition of memory chip maker GigaDevice Semiconductor Inc.(兆易创新) as a strategic investor.
The cash infusion further strengthens Nano-Core Chip’s ties with major players across China’s semiconductor and consumer electronics industries.
Corporate registration records show GigaDevice joined the latest financing round after Xiaomi had already invested through its venture arm in a pre-Series A round in January 2022.
From training to deployment
The investment comes as the AI industry shifts from training large models to deploying them at scale.
That transition is driving demand for inference chips that can operate efficiently in servers, AI PCs, smartphones and robots, where power consumption and cost constraints are often more stringent.
Founded in 2021 and incubated by the Peking University Information Technology Institute in Zhejiang, Nano-Core Chip develops computing-in-memory processors designed for large-model inference workloads.
The company’s proprietary 3D-CIM architecture combines computing-in-memory technology, three-dimensional near-memory computing and a RISC-V-based computing-storage framework.
The approach seeks to address one of the biggest constraints in AI computing: the movement of data between processors and memory.
Off the beaten path
In conventional von Neumann architectures, processors and memory operate separately, with data transfer accounting for more than 70% of total energy consumption.
Nano-Core Chip compares its approach to processing goods directly inside a warehouse rather than transporting them back and forth.
The company says its technology can deliver more than four times the computing density and reduce power consumption by more than tenfold in large-model inference applications.
Its processors target AI smartphones, AI PCs, internet-of-things devices, servers and embodied AI robots.
The company expects to move from laboratory testing to customer deployments within this year and enter volume production.
Unlike many AI chip developers, Nano-Core Chip does not rely on the most advanced manufacturing nodes, a strategy intended to improve supply-chain resilience and reduce production costs.
Shaping standards
Beyond chip development, Nano-Core Chip plays a role in shaping industry standards.
China’s RISC-V industry committee appointed the startup to lead the RISC-V Computing-in-Memory Application Group, where it is working with GigaDevice, Xiaomi Mobile, vivo, Inspur and dozens of other companies to develop what it describes as the world’s first RISC-V computing-in-memory standard.
The company has completed more than six financing rounds since its founding. Existing backers include HongShan, Xiaomi, Luxshare’s corporate venture funds, Lanchi Ventures, SMIC-backed China Fortune-Tech Capital, Addor Capital and Lenovo Capital.
