Morfics CGRA: programmable AI compute for cloud and embedded systems.
Morfics begins with software, open FPGA implementations, and FPGAI. Over time, the same ecosystem will target Morfics-owned coarse-grained reconfigurable architecture hardware for efficient inference, training, and real-time adaptation.
Programmable compute
A reusable compute fabric positioned between fixed accelerators and fully programmable FPGA logic.
One software ecosystem
Models, operators, community packages, and FPGAI contracts can evolve toward the same runtime.
Cloud and embedded
The architecture is planned for both managed infrastructure and Morfics-owned embedded hardware.
Open contribution path
The public FPGA ecosystem remains useful now and becomes a source of implementations, workloads, and benchmarks later.
See how today’s ecosystem leads to Morfics-owned compute.
The roadmap keeps current software, FPGA, and embedded capabilities useful while showing how they can evolve toward cloud and embedded CGRA infrastructure.
Roadmap
Useful today, built toward CGRA
Morfics is a working platform now; the CGRA is a future infrastructure layer.
AI Cloud
Managed training & inference on FPGA compute
Web AI Apps
Full-stack AI apps as one managed deployment
Embedded AI
On-device models for edge and robotics
Community Ecosystem
Models, benchmarks, and publications
FPGAI
One import path from source to hardware
FPGA support
AMD/Xilinx, Intel, Lattice, custom boards