A former Nvidia China executive set out to build a universal Chinese GPU: Moore Threads’ rapid rise
Moore Threads was founded in Beijing in 2020 by Zhang Jianzhong, a long-time graphics and semiconductor executive who had previously led Nvidia’s China business.
The company set out to build what it calls Universal GPUs capable of graphics, AI, video and scientific-computing workloads on one architecture.
The early business looks very different from the company visible today. Following the milestones in order shows which decisions changed Moore Threads’s direction and which capabilities remained constant.
The first MUSA architecture arrived within two years
Moore Threads launched its MUSA architecture and first Universal GPU products in 2022, including discrete graphics cards aimed at both visual and compute workloads.
The AI era increased the value of capabilities Moore Threads had already been developing. Compute, data, networking, software or infrastructure that once served narrower workloads became relevant to much larger AI systems.
A domestic software stack became as important as the chip
The company developed drivers, developer tools and the MUSA Toolkit because a GPU platform requires software compatibility as much as semiconductor performance.
The AI era increased the value of capabilities Moore Threads had already been developing. Compute, data, networking, software or infrastructure that once served narrower workloads became relevant to much larger AI systems.
Thousand-GPU clusters moved the company into AI infrastructure
Moore Threads demonstrated MTT KUAE clusters and expanded from individual cards towards systems for training large AI models.
The AI era increased the value of capabilities Moore Threads had already been developing. Compute, data, networking, software or infrastructure that once served narrower workloads became relevant to much larger AI systems.
The 2025 STAR Market listing financed a larger AI-factory ambition
Moore Threads listed on Shanghai’s STAR Market and in 2026 outlined plans for large AI-factory infrastructure using its own accelerator ecosystem.
For Moore Threads, access to public markets was more than a branding milestone. It increased financial transparency and created another route to fund expansion, acquisitions or long-term product development.
The GPU start-up is trying to build an ecosystem at start-up speed
As of 18 September 2026, moore Threads is now one of China’s highest-profile domestic GPU developers. Its short history has unfolded under strong demand for local alternatives to overseas accelerators and under export restrictions that make software and supply-chain independence strategic priorities.
For readers following the brand today, the history explains why newer announcements often connect back to technology or business decisions made much earlier.
For TechnologyBlog.co.za, this page is intended to serve as the company-history cornerstone for future Moore Threads coverage, so product and company-news articles can link back to one factual brand background instead of repeating the full origin story.
Primary sources checked include Moore Threads’ official company timeline. Information reflects public material available on 18 September 2026.
