作者丨骆轶琪 编辑丨倪雨晴 2025年上半年，国产AI芯片持续获得热捧，多家厂商业绩持续高增。在旺盛的AI推理需求支撑下，国 内头部AI芯片厂商左手积累产成品，右手储备关键材料，以应对未来市场发展。 21世纪经济报道记者梳理公司财报发现，总收入规模与海光信息还有一定差距的寒武纪，在上半年却与 前者的净利润规模已经十分接近；海光信息在上半年相比同业，有较大比例花费用在了销售上；龙芯中 科虽然盈利能力依然面临压力，但在市场周期回暖进程中，旗下核心产品毛利率表现已经逐渐回归，近 期其携"三剑客"进入市场，接下来有望看到进一步商用的好消息。 当然，国内AI芯片厂商的份额仍有较大提升空间。从沐曦股份披露的回复问询公告可见，政策支持对 国内市场采用国产AI芯片有一定驱动，但更大程度还是来自于内生竞争力提升和国内开源生态的密切 联动。 订单与存货"齐飞" 从收入规模看，分别有CPU和DCU（GPGPU的一种）业务的海光信息一直都在稳健增长态势中。上半 年海光信息实现营业收入54.64亿元，同比增长45.21%；归母净利润12.01亿元，同比增长40.78%。 同时，在旺盛的AI推理需求支撑下，正式摆脱"多年亏损"帽子的寒武 ...

作者丨骆轶琪 从收入规模看，分别有CPU和DCU（GPGPU的一种）业务的海光信息一直都在稳健增长态势中。上半年海光信息实现营业收入 54.64亿元，同比增长45.21%；归母净利润12.01亿元，同比增长40.78%。 同时，在旺盛的AI推理需求支撑下，正式摆脱"多年亏损"帽子的寒武纪追赶势头猛烈。上半年实现营业收入28.81亿元，同比增长 4347.82%；归属于上市公司股东的净利润10.38亿元，上年同期为亏损5.3亿元，同比增长295.85%。 寒武纪无论从总资产还是营收规模都与海光信息有一定差距，但在净利润层面实现了快速接近。 寒武纪在公告中提到，净利润大幅 增长源于营业收入增长，期内公司在持续拓展市场，助力人工智能应用落地。 编辑丨倪雨晴 2025年上半年，国产AI芯片持续获得热捧，多家厂商业绩持续高增。在旺盛的AI推理需求支撑下，国内头部AI芯片厂商左手积累产 成品，右手储备关键材料，以应对未来市场发展。 21世纪经济报道记者梳理公司财报发现， 总收入规模与海光信息还有一定差距的寒武纪，在上半年却与前者的净利润规模已经十分 接近； 海光信息在上半年相比同业，有较大比例花费用在了销售上；龙芯中科虽 ...

21世纪经济报道记者骆轶琪 2025年上半年，国产AI芯片持续获得热捧，多家厂商业绩持续高增。在旺盛的AI推理需求支撑下，国 内头部AI芯片厂商左手积累产成品，右手储备关键材料，以应对未来市场发展。 21世纪经济报道记者梳理公司财报发现，总收入规模与海光信息还有一定差距的寒武纪，在上半年却与 前者的净利润规模已经十分接近；海光信息在上半年相比同业，有较大比例花费用在了销售上；龙芯中 科虽然盈利能力依然面临压力，但在市场周期回暖进程中，旗下核心产品毛利率表现已经逐渐回归，近 期其携"三剑客"进入市场，接下来有望看到进一步商用的好消息。 当然，国内AI芯片厂商的份额仍有较大提升空间。从沐曦股份披露的回复问询公告可见，政策支持对 国内市场采用国产AI芯片有一定驱动，但更大程度还是来自于内生竞争力提升和国内开源生态的密切 联动。 订单与存货"齐飞" 从收入规模看，分别有CPU和DCU（GPGPU的一种）业务的海光信息一直都在稳健增长态势中。上半 年海光信息实现营业收入54.64亿元，同比增长45.21%；归母净利润12.01亿元，同比增长40.78%。 同时，在旺盛的AI推理需求支撑下，正式摆脱"多年亏损"帽子的寒武 ...

Core Viewpoint - The article highlights the significant progress of domestic GPU companies, particularly Moore Threads and Muxi, as they both received acceptance for their IPO applications, marking a historic moment for the domestic GPU industry [1][2]. Group 1: Company Profiles - Moore Threads was founded by Zhang Jianzhong, a former NVIDIA executive, who established a complete GPU ecosystem in China during his 15 years at NVIDIA [2][5]. - Muxi's founding team has a strong background in AMD, with key members having extensive experience in GPU design and architecture [5][6]. - Both companies aim to break the monopoly of international giants in the high-performance GPU market, despite their different strategic focuses [9]. Group 2: Product Development and Market Strategy - Moore Threads targets both data center (B-end) and consumer gaming graphics card (C-end) markets, aiming to create a comprehensive ecosystem [3][10]. - Muxi focuses primarily on the data center market, starting with general-purpose computing (GPGPU) and gradually expanding into graphics rendering [7][17]. - Moore Threads has developed the MUSA architecture, which supports AI computing, graphics rendering, and physical simulation on a single chip, making it the only domestic company to fully compete with NVIDIA [10][13]. Group 3: Financial Performance and R&D Investment - Both companies are investing heavily in R&D, with Moore Threads' R&D investment accounting for 626.03% of its revenue over the past three years, totaling 3.8 billion yuan, while Muxi's R&D investment is 282.11% of its revenue, totaling 2.2 billion yuan [25][21]. - Muxi's revenue has shown remarkable growth, with a compound annual growth rate of 4074.52% from 2022 to 2024, while Moore Threads also experienced a significant growth rate of 208.44% during the same period [28][30]. - Both companies are improving their gross margins, with Moore Threads increasing from 18.08% in 2022 to 36.60% in 2024, and Muxi achieving a gross margin of 62.88% in 2023 [32][32]. Group 4: Market Challenges and Future Outlook - The domestic GPU industry faces challenges, particularly in software ecosystems, where NVIDIA has a significant advantage with its CUDA platform [37]. - Both companies are working to build their developer ecosystems, with Moore Threads offering a software stack that allows for easier migration from CUDA, and Muxi supporting major frameworks like PyTorch and PaddlePaddle [37]. - Despite high R&D costs leading to losses, both companies are showing strong revenue growth and potential for future success in the GPU market [21][33].