Yizhu Technology raises over RMB 1 billion to commercialize compute-in-memory AI chips
The AMW Read
The financing advances an alternative AI chip architecture, but the absence of deployment results or measured economics limits its demonstrated impact to the compute-in-memory subsegment.
Yizhu Technology raises over RMB 1 billion to commercialize compute-in-memory AI chips
Yizhu Technology has completed financing of more than RMB 1 billion to support research, product iteration and commercialization, according to Leiphone. The Chinese AI chip company began operations in 2022 and focuses on general-purpose compute-in-memory architecture, branded GP-IMC, combined with a heterogeneous computing approach. Its development work includes 3D DRAM compute-in-memory technology. The report does not disclose the financing stage, investors or valuation.
The investment backs an alternative hardware architecture for data centers and AI cloud computing, placing Yizhu among infrastructure suppliers seeking to compete on inference economics and software usability. The company claims high throughput, low power consumption and lower token costs, alongside compatibility with CUDA and PyTorch. That combination makes software integration part of the commercial proposition: an alternative chip architecture must fit the tools customers already use. However, the report provides no performance measurements or customer deployment results to substantiate those claims.
For builders and investors, the concrete diligence question is whether the claimed efficiency survives deployment with existing AI workloads. Useful evidence would include measured token costs, power consumption and the engineering effort needed to run CUDA- or PyTorch-based applications. The financing supports further development, but the announcement alone does not establish production readiness or competitive economics.
