Singapore-headquartered semiconductor startup Symatics unveiled its Morphic Processing Unit, or MPU, architecture. The company develops AI chips and said the design aims to speed up AI responses. The announcement was made at the Business & Philanthropy Forum 2026.
How the architecture works
The MPU uses row-column parallelism. It also uses compiler-planned execution. These methods reduce data movement between memory and processing components.
Target workloads
That data movement can slow AI workloads. The MPU is aimed at interactive AI and reasoning tasks. Symatics said the design changes how data is organized and processed.
Partnership plans
Symatics plans to work with research and industry partners in Singapore. It will use an open architecture and licensing model. Reviewed materials showed no funding, prototype, or tape-out milestones.
Symatics said technical validation is ongoing. The MPU remains in development. Singapore seeks to strengthen its role in the global semiconductor ecosystem.