en.Wedoany.com Reported - Chinese GPU company Moore Threads Intelligent Technology (Beijing) Co., Ltd. disclosed its first-quarter report for 2026 and its annual report for 2025 on April 26th. In the first quarter of 2026, the company achieved an operating income of 738 million yuan, a year-on-year increase of 155.35%; net profit attributable to parent company reached 29 million yuan, turning a profit from a loss of 113 million yuan in the same period last year, representing an increase of 142 million yuan; net profit attributable to parent company excluding non-recurring gains and losses was a loss of 54 million yuan, narrowing the loss by 60.10% year-on-year. For the full year of 2025, the company achieved operating income of 1.505 billion yuan, a year-on-year increase of 243.37%; total gross profit reached 987 million yuan, a 218.43% increase year-on-year; the losses for net profit attributable to parent company and net profit attributable to parent company excluding non-recurring gains and losses narrowed by 38.16% and 33.38% respectively compared to the same period last year. After excluding the impact of share-based payments, the company's net profit loss for 2025 was 648 million yuan, narrowing by 847 million yuan compared to a loss of 1.495 billion yuan in the same period last year, a reduction of 56.65%.

R&D investment is the most prominent feature of Moore Threads' financial structure. The full-year R&D expenditure for 2025 reached 1.305 billion yuan, accounting for 86.68% of operating income, a ratio among the highest for Chinese GPU companies. The company continues to invest heavily in areas such as its full-function GPU chip architecture, MUSA software ecosystem, thousand-card-level smart computing clusters, and super-node technologies, with R&D personnel accounting for over 85% of its workforce. This high-intensity R&D investment supports the company's technological iteration across three levels: chip design, computing frameworks, and cluster scheduling, and has also laid an engineering foundation for subsequent large-scale commercial delivery.

The commercialization process accelerated in the first quarter. In Q1, Moore Threads secured a major order worth 660 million yuan for its "Kua'e" smart computing cluster from a central state-owned enterprise. This order, covering thousand-card-level training clusters and inference service nodes, represents the highest single contract value since the company's inception. The company also disclosed that its smart computing cards have been deployed in multiple smart computing centers and cloud service platforms, making it one of the few GPU suppliers in the Chinese market to achieve truly large-scale commercial application of thousand-card and ten-thousand-card level clusters. As of the end of Q1, the company's cumulative delivery for smart computing clusters exceeded 15,000 full-function GPU cards, covering key computing hub regions including Beijing-Tianjin-Hebei, the Yangtze River Delta, and the Greater Bay Area.

Based on its next-generation "Huagang" GPU architecture, Moore Threads is advancing the construction of 100,000-card-scale super-large smart computing clusters and core technology research and development, such as super nodes. The "Huagang" architecture features targeted optimizations in instruction sets and interconnection topology for large model training and inference scenarios, improving single-node computing density approximately threefold compared to the previous "Pinghu" architecture. The MUSA computing platform has adapted mainstream AI frameworks such as PyTorch, vLLM, and DeepSpeed, completing structural verification for over 300 models covering large language models, multimodal models, and scientific computing. On April 24th, the company announced that its MUSA operator library, based on TileLang, has achieved Day-0 support for DeepSeek-V4, with an operator unit test coverage rate exceeding 95%.

Relying on deep cooperation with wafer foundries in China, Moore Threads' GPU chip shipments more than doubled year-on-year for the full year of 2025, with shipments in the first quarter of 2026 reaching 45% of the 2025 total. The company simultaneously expanded its headquarters in Beijing and the R&D center in Shanghai, with total employees exceeding 2,500 people. As AI computing demands continue their large-scale shift from training to inference, the penetration rate of Moore Threads' full-function GPU roadmap continues to increase in applications within the Xinchuang industry (domestic IT infrastructure), finance, energy, and telecommunications sectors. The company anticipates that the total annual revenue for 2026 could exceed 3 billion yuan, with R&D investment projected to remain above 1 billion yuan.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com