Skymizer has launched HyperThought, its next-generation AI accelerator IP, following the success of its EdgeThought LPU in 2024. The new technology is designed to meet the growing demands of real-time, multimodal, and agent-based AI applications.
HyperThought is a fully integrable subsystem that allows SoC teams to design custom AI silicon tailored to their specific applications. It is powered by Skymizer’s Language Instruction Set Architecture (LISA) v3, enabling multimodal processing, concurrent multi-model execution, and native support for agentic AI workflows.
The platform delivers unmatched efficiency in compute, memory, and bandwidth, making it suitable for edge AI deployments in smart devices and autonomous systems. According to William Wei, CMO of Skymizer, “EdgeThought brought LLM inference to the edge. HyperThought makes the edge intelligent.”
Key features of HyperThought include its support for dynamic graph control, parallel agent execution, and advanced memory operations, all fully hardware-supported. It also enables seamless execution of text, vision, and instruction-based models across shared compute resources.
Key highlights of HyperThoughtLuba Tang, CTO of Skymizer, described HyperThought as “the most advanced LPU we’ve engineered,” stating that LISA v3 enables support for real-time intelligent systems that can react, infer, and adapt.
Skymizer will showcase HyperThought at COMPUTEX Taipei from May 20–23, 2025, with demonstrations of FPGA-verified IP, technical sessions on LISA v3, and previews of the software tools and developer stack. The event will take place at the Taipei Nangang Exhibition Center Hall 1.
All Rights Reserved. Copyright , Central Coast Communications, Inc.