The company that casts models into chips.
Hefei Fanshi Semiconductor Technology Co., Ltd. builds edge LLM inference chips, inference IP and engineering validation platforms around compute-in-memory and near-memory compute, on a mature SMIC 28 nm supply chain.

What we do
We cast large language models into compute-in-memory silicon, so inference runs without a network, data never leaves the device, and the marginal cost of each inference approaches zero. Customers either use a production ASIC directly, or license the inference cores into their own ASIC or FPGA.
- Company
- Hefei Fanshi Semiconductor Technology Co., Ltd.
- Location
- Hefei, Anhui, China
- Process and supply chain
- Mature SMIC 28 nm
- Routes
- MaskROM CIM · SRAM CIM · near-memory compute
- Products
- Forge · Trident · Argus · custom ASIC
- Inference IP
- FS-Fusion · FS-Attention · FS-MoE · FS-Linear
- Languages
- English · Chinese
- Contact
- info@fanshi-silicon.com
How we work with customers
01
Pick the route
Target model, performance and power goals, update cadence, interfaces. The first question is always: how often does the model change?
02
Evaluate
Run the model on the evaluation board, measure accuracy and throughput, and get a working engineering sample before any silicon is committed.
03
Design in
Interfaces, system software and field validation built around the evaluation results, turning the sample into a product prototype.
04
Deliver the ASIC
Once the model is frozen: a production chip, or the IP integrated into your own silicon.


Products
Edge ASICs for lightweight language models, heavy-load inference and multimodal vision.
Technology
MaskROM, SRAM CIM and near-memory compute: three routes for three update cadences.
Contact Fanshi Semiconductor
Business, IP licensing, samples and careers: info@fanshi-silicon.com.
