Singapore chip startup Symatics unveiled a new AI chip architecture, MPU, aimed at reducing inference latency

深潮 TechFlow reported: On November 6, according to Tech in Asia, Singapore-based semiconductor startup Symatics unveiled the Morphic Processing Unit (MPU) architecture at the 2026 Business and Philanthropy Forum, aiming to improve AI response speed by optimizing data organization and processing.

Symatics said the architecture uses row-column parallel processing and compiler-scheduled execution mechanisms to reduce data transfer between memory and compute units, primarily targeting interactive AI and inference tasks. The company plans to collaborate with Singaporean research institutions and industry partners through an open architecture and licensing model.

Currently, the MPU remains in the R&D stage, with technical validation underway; no progress on funding, prototype chips, or tape-out has been disclosed.

Source: 深潮 TechFlow