Move less. Keep inference on the device.
Fanshi organises its technology around where model weights are stored and computed: MaskROM compute-in-memory, SRAM compute-in-memory and near-memory compute.

Cast the whole model
into a single chip.
GPUs and NPUs spend most of their time and power shuttling weights between memory and compute. Fanshi turns that around: the weights live in the memory array, and the math happens where they are.
Make the weights part of the silicon.
External weight traffic per token
0MB
What remains at runtime
In-place compute · no external weight traffic
0.8B · INT8 estimate, not system powerSeparate memory
Every token moves every weight
In a conventional NPU or GPU, memory and compute are separate pieces of silicon. For every token generated, all of the weights cross the bus: bandwidth sets the ceiling, and most of the energy is spent in transit.
Compute-in-memory
Put the weights in the memory array
Fanshi builds the multiply-accumulate circuits into the memory array itself. Weights stay in the array, activations flow in, results flow out. No weights on the bus.
MaskROM
Cast at tape-out, never moved at runtime
Mask ROM compute-in-memory writes the weights physically into the via array during tape-out. The chip leaves the fab with the model already inside, so power and marginal cost per inference drop to the minimum.
One goal,
three routes.
All three move the weights less, or not at all. They differ in where the weights live and whether the model can still change. The first question to answer: how often does your model change?
- Where weights live
- Mask layers, cast at tape-out
- Runtime movement
- None
- Model updates
- Fixed at tape-out
- Best for
- Stable models, high volume, lowest power and marginal cost
- Where weights live
- Inside the SRAM array
- Runtime movement
- MAC inside the array
- Model updates
- Loaded at boot, updatable
- Best for
- Mid-size models, devices that need to swap models in the field
- Where weights live
- High-bandwidth external memory
- Runtime movement
- On-chip streaming
- Model updates
- Continuous iteration
- Best for
- Large, fast-moving models in heavy-load nodes
On-device decode speed
Forge design target · Qwen 3.5 0.8B
System power
Forge, lightweight edge class
Data stays on the device
Local inference, no network needed
Marginal cost per inference
After the one-time chip purchase
Further reading
What Is a Compute-in-Memory AI Chip? Three Routes ExplainedA compute-in-memory AI chip runs multiply-accumulate inside the memory array, so weights stop moving. How MaskROM, SRAM CIM and near-memory compute differ.
MaskROM Compute-in-Memory: What Casting Weights in Silicon MeansMaskROM compute-in-memory writes model weights into a via array at the mask stage, so nothing moves at runtime. How it works, why it costs least, its limits.
The Memory Wall: Why Edge LLMs Stall and How to Get Around ItLLM decoding is bandwidth-bound: every token reads all the weights, and edge devices cannot sustain that traffic. How high the wall is and how to get past it.
Choose the route for your project
Send the target model, performance and power goals, and how often the model changes. We reply with a route and an evaluation plan.

