China's GPU pioneer, Moore Threads Technology Co.,Ltd., is charting a path toward limitless possibilities.
### 01 #### Why Cash Management? Recent discussions have swirled around Moore Threads' plan to utilize up to ¥7.5 billion in idle raised funds for cash management. Questions arose: "What’s the actual use of the raised capital?" "Are they prioritizing wealth management over R&D?" "Is this prudent planning or a distraction?"
However, a closer look at Moore Threads' December 12 announcement clarifies the situation. The company intends to deploy idle funds—without disrupting ongoing projects or compromising capital security—into low-risk, high-liquidity instruments such as structured deposits, certificates of deposit, and other principal-guaranteed products.
The raised capital is earmarked for multi-year projects, with expenditures phased over three years. A company spokesperson emphasized that the ¥7.5 billion represents an upper limit for cash management, not the actual amount deployed.
This approach reflects responsibility to shareholders: idle funds are optimized conservatively, avoiding even bond market risks.
### The GPU Challenge: A National Priority Moore Threads is tackling full-function GPU development—a critical battleground in China’s AI rivalry with global powers. Unlike GPGPUs (focused on AI computation) or specialized ASICs (e.g., Google’s TPU), full-function GPUs, like NVIDIA’s, integrate AI, graphics rendering, and scientific computing.
Intel’s decades-long struggle in this space underscores the difficulty. Yet Moore Threads has broken through, deploying its MUSA architecture to deliver multi-capability GPUs for both consumer (gaming GPUs like MTT S80) and industrial applications (data centers, design tools).
### 02 #### To B & To C: Dual-Market Traction The 2024 launch of *Black Myth: Wukong*, a AAA game, inadvertently tested Moore Threads’ mettle. Its MTT S80—powered by the self-developed "Chunxiao" chip—became the only domestic GPU to run the game smoothly after a driver update boosted frame rates to 35–55 FPS.
Beyond gaming, Moore Threads’ "cloud-edge-device" strategy has solidified its B2B presence. From its 2021 debut chip "Sudi" to the 2023 "Kuae" AI training cluster (scaling to 10,000 cards), the company has built a full-scenario GPU portfolio.
By 2024, AI compute products drove 77.63% of revenue, marking a pivotal shift from earlier reliance on graphics accelerators and SoCs.
### 03 #### Ecosystem Breakthrough: The MUSA Advantage NVIDIA’s CUDA ecosystem—a de facto "OS" for AI—poses a formidable barrier. In 2024, tightened CUDA compatibility restrictions targeted rivals, including Chinese firms.
Moore Threads’ response? Its fully independent MUSA architecture, coupled with the MUSIFY migration tool, enables seamless code transitions from CUDA with minimal developer effort. This "bridge via compatibility, root in autonomy" strategy is gaining traction as developers seek alternatives to CUDA lock-in.
### Market Outlook & Financials Analysts project NVIDIA’s market cap could hit $9 trillion by 2030, reflecting GPU sector optimism. In China, Frost & Sullivan forecasts AI chip sales to surge from ¥142.5 billion (2024) to ¥1.34 trillion (2029), with GPUs dominating 77.3% share.
Moore Threads’ revenue skyrocketed from ¥46 million (2022) to ¥702 million (H1 2025), underscoring its growth trajectory. While challenges remain, its dual-market progress and ecosystem strategy position it as a contender in China’s GPU revolution.
Investors watch closely—not for another Cambricon, but a future rival to NVIDIA.
Comments