Skip to main content

12 posts tagged with "Domestic Substitution"

AI chip domestic substitution process

View all tags

IDC 数据实锤:国产 AI 芯片国内份额合计 50%,竞争转向系统级能力

· 4 min read
Industry Research Team

「国产替代」讲了三年,这次有了一组完整的第三方数据。IDC 2026 上半年口径下,中国数据中心 AI 芯片市场自主品牌合计已占 50%——一半的市场,第一次有了数据实锤。

一、IDC 份额数据(2026 上半年,中国数据中心 AI 芯片)​

玩家出货量份额同比
英伟达67.8 万颗34%-44%
华为昇腾57.2 万颗28%领跑国产
其他国产(平头哥/寒武纪/昆仑芯/海光/摩尔线程等)合计 40 万+颗合计 22%高增长
自主品牌合计—50%首次过半

补充口径:IDC 预计 2026 年中国 AI 加速器整体出货量同比增长 19%——市场在涨,英伟达在跌,国产在吃增量。

二、集微咨询:产业进入「业绩兑现」阶段​

集微咨询《2026 中国 CPU/GPU 行业上市公司研究报告》(以寒武纪、海光、摩尔线程、沐曦、壁仞、燧原等 11 家 A 股/港股公司为样本):

  • 2026 上半年 11 家合计营收约 254.11 亿元,6 家盈利、5 家亏损——从「概念验证」迈入「业绩兑现」
  • 全球背景:上半年全球半导体市场 7020 亿美元(+102%),全年有望冲击 1.655 万亿美元;高盛测算 2026–2031 全球 AI 基础设施累计资本开支 7.6 万亿美元
  • 全球 GPU 市场:2024 年 773.9 亿美元 → 2025 年约 1046 亿 → 2026 年预计超 1400 亿美元,两年接近翻倍
  • 中国 AI 芯片市场:2025 年 3039 亿元(2021 年以来 CAGR 77.9%),2030 年有望接近 1.6 万亿元
  • 本土份额第一:华为海思以 23% 居中国 AI 加速芯片市场本土首位
  • 全球 AI 服务器:2026 年出货 980 万台(+44%);亚太以 48.22% 居全球第一大 GPU 消费与制造区域

三、竞争焦点:从单芯片到系统级​

IDC 指出,市场竞争正从单颗芯片性能转向系统整体能力——芯片在大型集群中的协同表现。原因很直接:

  • 大规模场景下,芯片间数据传输、内存调度、并行效率、软件适配同样决定实际性能
  • 推理场景企业真正关心的是:每秒处理多少 Token、时延能否达标、单位 Token 综合成本
  • 峰值性能高不等于业务算力利用效率高

这与我们在超节点元年综述中的判断互相印证:华为围绕「超节点 + 集群」做系统架构创新(汪涛,HC 2026),国产各家的超节点产品在工博会集体亮相。

四、规模化应用的微观证据​

  • 平头哥:真武系列已通过阿里云服务自动驾驶、互联网、金融等 20 多个行业的超过 650 家外部客户,覆盖训练、微调、推理全类型负载
  • 昆仑芯:P800 完成大规模集群验证,依托百度智能云百舸平台实现 3.2 万卡集群点亮与规模化应用
  • 寒武纪:产品在金融、互联网等核心行业规模化商用部署
  • 智谱:GLM-5.3-Flash 全部线上流量由超过 10 万张国产芯片集群承载

结论​

50% 这个数字的意义不在大小,而在结构:国产芯片不再是政策驱动的备胎,而是市场份额、财务表现、客户规模三条曲线同时向上的主流选项。下一步的分水岭是 IDC 点名的系统级能力——超节点互联、集群调度、单位 Token 成本,这些维度上的差距会比单卡规格更难追赶,也更有价值。

数据来源与口径说明​

  • IDC 份额/出货/增速数据:IDC 报告(2026 上半年),腾讯新闻/搜狐转述(2026-10-24/25)
  • 集微咨询报告数据:集微咨询《2026 中国 CPU/GPU 行业上市公司研究报告》(2026-10-25 发布),中商产业研究院/弗若斯特沙利文/高盛数据转引
  • 平头哥/昆仑芯案例:腾讯新闻(2026-10-24),阿里/百度口径
  • 「50%」为自主芯片份额加总口径,统计边界为中国数据中心 AI 芯片市场

国产算力芯片告别「躺赢」:Day 0 适配普及后,竞争转向软件迁移与供货保障

· 4 min read
Industry Research Team

业绩全线飘红的另一面,是「靠外部环境躺赢」的阶段正在过去。当每家头部厂商都能在开源模型发布当天完成适配,差异化优势从「能不能跑」转向「迁移多快、效率多高、供货多稳」。

一、三季报成绩单:增长普遍亮眼​

2026 年三季报季,国产算力芯片公司陆续交卷:

公司业绩口径数字
沐曦股份前三季度营收同比预增 105.60%–151.29%,归母净利润扭亏为盈(10-15 预告)
海光信息前三季度营收同比+57.58%;Q3 单季 58.56 亿元创单季新高(10-19 三季报)
寒武纪2026 上半年营收同比+108.13%(三季报未出)
摩尔线程2026 上半年营收同比+147.42%(三季报未出)

增长背后是三重红利叠加:AI 算力需求快速增长、厂商自身技术积累、以及出口管制 limiting 英伟达高端芯片对华销售带来的「竞争真空」——国产芯片过去几年订单有保障,海外竞争相对有限。

二、Day 0 适配:从稀缺能力到行业标配​

模型适配曾是国产芯片最硬的差异化壁垒,如今这条护城河正在快速填平:

  • 2025 年 1 月,DeepSeek-R1 开源后,国产芯片大多需要十多天才陆续完成适配
  • 2025 年 9 月,DeepSeek-V3.2-Exp 开源,华为昇腾、寒武纪、海光信息等已能当天完成适配
  • 2026 年 10 月 15 日,阶跃星辰开源 Step 5 Preview,华为昇腾、摩尔线程、沐曦股份、壁仞科技、海光信息 DCU 全部当天完成适配

「Day 0 适配」在头部国产芯片厂商中已逐渐普及。技术路径也在收敛:各家芯片厂商的基础软件接入开源推理框架后,新模型发布时的主要工作只剩补齐新算子、再做精度与性能调优——适配从「攻坚工程」变成「流水线作业」。

三、竞争的下一站:三个新战场​

当「能不能跑」不再是问题,竞争延伸到算力落地的各个环节:

1. 存量应用迁移。新模型好适配,存量软件难迁移。行业用户的软件大多在国外软硬件环境上开发,商业软件无源码、迁移困难;自研算法团队的移植往往需要几十人、甚至接近两百人的深度适配团队。中科曙光高端计算首席工程师布静德对媒体表示,随着基础软件完善,科学计算主流软件环境已基本支持,但传统 CPU 应用迁移仍有工作量。

2. 系统效率。同样一张卡,调度、通信库、算子融合的差距可以拉开数倍的实际吞吐——这正是超节点方案(机柜级互联)成为各家新战场的原因,参考我们的超节点元年综述与国产超节点工博会盘点。

3. 供货保障。先进制程产能是全行业共同的紧约束:寒武纪依赖中芯国际 N+2 制程,而该产线最大产能客户是华为。据彭博社 10 月初报道,寒武纪 2026 年出货目标约 50 万颗(含思元 590 与思元 690),代工配额直接决定这盘生意的天花板。「能产多少供多少」正在成为比性能参数更硬的竞争力。

四、对采购方的含义​

  • 适配差距抹平意味着可以放开手脚比硬指标:在 Day 0 适配成为标配后,选型应回到单卡/单柜实际吞吐、单位 Token 成本与生态迁移成本
  • 把迁移人力写进 TCO:存量业务迁移的成本(人员、周期、精度回退风险)在不同厂商间差异巨大,这部分隐性成本常被报价单忽略
  • 供货承诺要写进合同:在产能紧约束下,交期违约条款与备货缓冲比纸面参数更值得谈判

数据来源与口径说明​

  • 三季报数字:沐曦 10-15 业绩预告、海光 10-19 三季报、寒武纪/摩尔线程 2026 半年报——经济观察报/腾讯新闻(2026-10-22)汇总
  • Day 0 适配时间线与曙光表态:经济观察报报道(2026-10-22)
  • 寒武纪出货目标与中芯产能约束:彭博社报道(2026-10 初),为知情人士口径,非公司披露

长鑫第五代 DRAM 量产:从「跟跑存储」到「补强 AI 内存自主链」

· 3 min read
Industry Research Team

人民日报 10 月 21 日在头版文章中把「长鑫存储第五代 DRAM 技术平台实现量产」列为「练好内功」的标志性成果之一——与华为昇腾 960 超节点、宁德时代电池专利并列。在全球存储涨价的背景下,这条消息的分量比标题重。

一、长鑫的位次​

指标数据(2026-10 行业口径)
全球 DRAM 排名第 4 位,份额 9.5%
技术平台G5(第五代)DRAM 量产
前沿方向4F² DRAM 技术推进
与头部差距评估约 1-3 年

二、为什么 DRAM 量产与 AI 算力有关​

直接关系藏在存储超级周期的结构里:

  1. HBM 挤出效应:HBM 在 DRAM 产能中占比每升 1 个百分点,消费级 DRAM 供给收缩约 3 个百分点——全球 DRAM 合约价 4Q26 环比涨 10-15%,交期 40 周以上
  2. 国产 AI 服务器的内存成本:国产芯片卡的显存是 HBM(受制),但服务器主存、KV Cache 扩展内存、向量库存储大量使用标准 DRAM——长鑫放量直接压低国产算力方案的 BOM
  3. HBM 的前置能力:HBM 本质是 3D 堆叠的 DRAM。五代 DRAM 平台的量产意味着存储微缩工艺的持续积累——这是未来国产 HBM 的必要非充分条件

三、全球背景下的时间窗口​

当前存储市场处于「15 年来最严重短缺」(高盛口径):三星、SK 海力士把产能大幅倾斜给 HBM 长约,传统 DRAM 与 NAND 供给被动收缩,价格上行周期预计延续到 2027-2028 年新产能落地。长鑫 G5 的量产正好落在需求最旺的窗口——份额提升与价格上行双重红利。

对照全球格局:SK 海力士 HBM 收入份额 ~50%、三星 33%、美光 18%,HBM4 已进入三倍报价时代;国产存储在 HBM 上尚有代差,但在标准 DRAM 上以 9.5% 份额站稳第四,构成「标准存储自主 + HBM 跟进」的现实路径。

四、对 AI 算力采购的提示​

  1. 国产服务器主存方案开始具备「长鑫平替」选项,采购 RFP 里可以列入
  2. 存储涨价周期内,提前锁定内存采购与芯片采购同等重要
  3. 关注长鑫 HBM 进展的官方披露——一旦进入国产 AI 芯片的 HBM 供应链验证,国产算力全链自主的最后一环将补齐

延伸阅读:HBM4 三倍报价 · 存储超级周期 · 完整对比表

寒武纪 MLU690 规模部署与订单交付最新进展

· 5 min read
Industry Research Team

国产 AI 芯片市场对 MLU690 的期待,从 2025 年一路延续到 2026 年。但据公开报道,MLU690 在 2026 年并没有形成大规模出货计划——寒武纪全年晶圆资源优先保障已成熟量产的 MLU580 与 MLU590。摩根士丹利 2026 年 10 月的研报也确认了这一点:MLU690 系列(特别是 MLU692)收入确认延迟,构成三季度业绩不及预期的主因。本文梳理 MLU690 的真实交付进展。

MLU690 交付现状:产能优先给成熟型号​

据公开报道,寒武纪 2026 年的产能分配策略已经明确:全年晶圆资源优先供给成熟量产的 MLU580、MLU590 两款芯片,MLU690 中芯版本暂不规划批量出货,也没有主动向客户推广。核心制约仍是产能瓶颈,即便拿到客户订单也无法按时交付,因此公司索性收缩 MLU690 的市场投放节奏。

摩根士丹利的研报从财务角度印证了同一件事:预计寒武纪 2026 年第三季度收入为 36.5 亿元人民币,低于此前预期的 47.7 亿元。不及预期的主要原因在于 MLU690 系列(特别是 MLU692)的收入确认延迟以及 MLU580 的交付量有限。寒武纪在第三季度已将 MLU692 交付给主要的 CSP(云服务提供商)进行测试,但预计该产品的大部分收入将在第四季度才能确认。

这就是 MLU690 当前的真实状态——不是"卖不动",而是"没有足够的晶圆去卖"。

型号2026 年产能地位交付状态
MLU580主力,产能倾斜交付量有限,三季度提价后毛利率仍偏低
MLU590主力,产能倾斜上半年内可交付
MLU690 / MLU692产能优先级靠后已交付主要 CSP 测试,收入多在四季度确认
MLU790(下一代)尚未推进受 690 验证进度影响,节奏更谨慎

据公开报道,MLU690 的合同虽在上半年签订,但交付预计在第三季度。这与摩根士丹利所述"大部分收入在第四季度确认"基本一致。

央国企市场:可见的订单空间​

在消费级与互联网市场之外,央国企与训练场项目是 MLU690 的主要落地方向。据公开报道,2026 年央国企市场的可见商机空间预计在 10 万至 20 万卡,其中上半年预计可落地订单约为 5 万卡。这 5 万卡中,有 3 万卡将用于与京能合作的国家训练场项目,另外 2 万卡来自其他央国企计算项目扩容(如浙江台州、昆山的项目)以及上海、安徽合肥等地的训练场和电信运营商项目。

值得注意的是,央国企订单的竞争格局并不宽松。据公开报道,在央国企市场寒武纪并非绝对的第二供应商,市场地位大约能排进前五;在信创领域华为是绝对的第一供应商,占据超过 60% 的市场份额。对寒武纪构成最大竞争压力的是平头哥和昆仑芯——这两家公司背靠阿里云和百度云,具备算力消纳能力,能通过云服务输出算力,而寒武纪无法做出此类承诺。

超节点布局:停滞状态​

超节点是 2026 年国产算力的主战场,但寒武纪在这条线上基本缺席。据公开报道,寒武纪自己没有做超节点,也没有明确的自建超节点计划。当前主要是给一些做超节点的厂商提供产品做验证和适配,2026 年还没有客户明确要求寒武纪去部署超节点,而且产能也不支持这么做。

现有产能仅能勉强满足字节的八卡基础服务器订单,字节自身也没有超节点部署计划,相关计划更偏 2027 年。腾讯和阿里虽有超节点方向的规划,但寒武纪没有办法保证供货,所以没有真正推进到客户部署层面。目前更多是与 OEM 厂商做平台适配,比如阿里的盘古、浪潮、新华三等平台已经做过适配,但没有量产安排。

还有一个被忽视的连锁反应:MLU690 的延迟量产会进一步拖累下一代 MLU790 的流片进度。MLU690 本身承担 790 前置大规模商用验证任务,缺少足量客户真实场景数据迭代,下一代芯片的研发与流片节奏会持续延后。

研判​

寒武纪当前的处境可以用一句话概括:需求不缺,晶圆不够。据公开报道,摩根士丹利预计寒武纪 2027 年向 CSP 客户的出货量将接近 35 万颗,主要得益于 MLU692 的产能爬坡和下一代 MLU580 产品的推出;代工方面主要依赖中芯国际的 N+2/N+1 产能。HBM3e 仍是关键供应瓶颈,但供应链调研显示寒武纪拥有一定库存,并有望随国内产能爬坡获得长鑫存储的一部分 HBM3e 产量。

财务侧的压力同样真实:HBM 成本上升对单位经济效益造成压力,MLU580 即使三季度提价,毛利率仍将保持在相对较低水平。摩根士丹利维持增持评级,目标价从 1408 元上调至 1700 元,但明确将三季度称为"业绩低谷"。

对客户而言,MLU690 在 2026 年不是可规模化规划的选项;2027 年随着产能爬坡与 790 产品线就位,它才会真正进入放量期。这一年多的窗口期,寒武纪要证明的是晶圆产能能否转化为可持续的交付能力。

国产化替代迁移清单:从 CUDA 迁到昇腾/寒武纪/海光要做什么

· 4 min read
Industry Research Team

国产化替代最常见的失败模式不是"卡买不到",而是"代码跑不动、精度对不齐、性能差五倍"。迁移本质是一个软件工程问题,用清单化的方式管理,成功率会高得多。

一、三条生态路线先分清​

昇腾(CANN 加 torch_npu):生态最完整的国产路线。PyTorch 官方支持度较好,主流模型(LLaMA、Qwen、DeepSeek 系列)的适配案例最多,社区踩坑记录丰富。硬件侧 Atlas 800T A2(8 卡 910B)训练机约 75 万至 200 万元区间,910C 8 卡服务器渠道价约 110 万至 120 万元,推理侧 Atlas 300I Duo 单卡约 2 万至 4 万元。

寒武纪(Neuware):训练卡 MLU 系列、推理卡加边缘盒子的组合,推理场景渗透率提升明显,国产推理卡渠道价已进入 3 万至 5 万元区间。

海光(DTK,ROCm 系):DTK 从 ROCm 分支演化,对 CUDA 代码的兼容性是三条路线中相对友好的,GPGPU 架构使通用算子迁移成本较低,适合已有大量自定义 CUDA 算子的团队。

二、六阶段迁移清单​

阶段工作内容典型工作量常见坑
1. 资产盘点扫描代码中的自定义 CUDA 算子、专有库依赖(NCCL、TensorRT、FlashAttention)1 至 2 周低估第三方闭源库依赖
2. 环境搭建驱动、固件、CANN 或对应 SDK、框架版本对齐1 至 2 周版本矩阵组合错误,驱动与 SDK 不匹配
3. 框架适配安装 torch_npu 或对应插件,跑通开源模型 baseline2 至 4 周依赖库里隐藏的 CUDA 专属调用
4. 自定义算子迁移用等价 API 重写或请求原厂支持2 至 8 周,取决于算子数量算子性能回退数倍,需原厂深度调优
5. 精度对齐FP16 或 BF16 下逐层对比 loss 与输出1 至 3 周舍入差异累积导致 loss 曲线发散
6. 性能调优与压测通信库替换、算子融合、混合精度策略2 至 6 周集群线性加速比不达标,多因网络配置

整体而言,一个"标准开源模型加常规训练脚本"的项目迁移到昇腾,顺利情况下约 6 至 10 周;有大量自定义算子和闭源依赖的项目,周期可能拉长到半年,且强依赖原厂驻场支持——签约时务必把驻场支持写进合同。

三、最容易翻车的五个环节​

1. NCCL 依赖。 分布式训练脚本里的 NCCL 调用要替换为昇腾 HCCL 或对应通信库,超参与拓扑配置需要重新调试,多机通信效率是国产集群性能差距的最大来源。

2. 量化推理工具链。 从 TensorRT 到昇腾 MindIE 或各家的量化工具,量化校准数据集和精度评估都要重做,生产环境的精度回归测试不可省略。

3. 生态外围工具。 DeepSpeed、vLLM、Transformer Engine 等常用库在国产平台上的版本支持滞后,很多团队卡在"模型能跑,加速库不兼容"这一步。

4. 运维体系。 现有监控(DCGM 类工具)、故障诊断、驱动升级流程全部要重建,这部分工作量经常被漏算。

5. 人才储备。 国产平台的调优经验集中在原厂和头部客户,招聘有昇腾或寒武纪实战经验的工程师成本高,提前安排团队参加原厂认证培训更实际。

四、结尾建议​

给准备迁移的团队三条建议:第一,先迁推理、后迁训练——推理负载对生态的依赖更浅,Atlas 300I Duo 这类推理卡性价比高,能用最小代价验证国产路线的可行性;第二,选硬件前先选生态,优先挑你的核心模型已有官方适配案例的平台,能省一半工作量;第三,保留双栈能力,迁移期让 CUDA 与国产平台并行运行至少一个季度,用同一份测试集做持续对比,确认精度与性能达标后再下线旧集群。国产化替代是一场马拉松,清单化管理、分阶段验收,才能避免"买回来跑不起来"的困境。

国产 AI 芯片替代采购决策框架:昇腾、寒武纪、摩尔线程、昆仑芯怎么选

· 7 min read
Industry Research Team

国产替代不是"把英伟达换成国产卡"一句话能概括的决策。合规压力、负载类型、迁移成本、供货确定性、预算五个维度都要过一遍,本文用站内芯片卡数据把每一步落到具体型号上。

一、替代决策五问​

动手选型前,先回答五个问题:

  1. 合规要求:是政策强制的信创场景(政府、国企、金融),还是成本驱动的自愿替代?前者看《安全可靠测评》等资质,后者看性价比。
  2. 负载类型:训练还是推理?Prefill 还是 Decode?不同负载的答案完全不同——昇腾 950PR 是推理 Prefill 卡,拿它做 FP16 密集训练就是选型错误(站内卡明确提示:FP16 训练应选 910C,其 800 TFLOPS 对 950 系列的 486–547 TFLOPS 保持领先)。
  3. 生态迁移成本:现有 CUDA 代码库有多少算子依赖?团队有没有非 NVIDIA 平台经验?迁移成本有时比卡价更高。
  4. 供货确定性:产能爬坡与 HBM 供应决定交付周期,头部型号已经在涨价排队。
  5. 预算口径:国产卡价格已从"替代品折价"转向"供给稀缺溢价",预算要按动态报价做。

二、国产主要选项对比​

以下数据全部取自本站芯片卡:

型号架构FP16/BF16 算力低精度显存 / 带宽互联软件栈参考价
昇腾 910CDa Vinci 双芯BF16 800 TFLOPS不支持 FP8/FP4128GB HBM2e / 3.2 TB/sHCCS 784 GB/s + UB 392 GB/sCANN渠道价超 ¥11 万
昇腾 950DTDa Vinci v5BF16 547 TFLOPSFP8 1,034 TFLOPS、FP4 2,007 TFLOPS144GB 自研 HiZQ 2.0 / 4 TB/s灵衢 2 TB/sCANN Next渠道指示价超 ¥25 万
寒武纪 思元690MLUarch 双 dieFP16 700+ TFLOPSINT4 2,800+ TOPS196GB HBM3 / 3.35 TB/sMLU-Link 890 GbpsNeuWare约 ¥17.5 万
摩尔线程 MTT S5000MUSA 平湖(第四代)BF16/FP16 400 TFLOPSFP8 稠密 1,000 TFLOPS80GB / 1.6 TB/sMTLink 784 GB/sMUSA / MUSIFY官方指导价 ¥5.5 万
天数智芯 天垓150ivcore11FP16 192 TFLOPSINT8 384 TOPS64GB HBM2e / 1.2 TB/sPCIe 4.0IXUCA未公开
燧原 云燧T21邃思2.0FP16 未公开TF32 160 TFLOPSHBM2E(容量未公开)自研多卡互联燧原智算软件栈约 ¥7 万

怎么读这张表:训练优先看昇腾 910C/950DT 与思元690(FP16 算力与显存容量领先,且都有千卡/超节点级方案——思元690 已落地 256 卡超节点,950DT 的 Atlas 950 SuperPoD 1024 卡版本已在 WAIC 2026 真机展示);FP8 推理看 MTT S5000(FP8 稠密 1,000 TFLOPS 是国产卡中的亮点,且已通过《安全可靠测评》2026 年第 2 号);视频与低功耗推理看昆仑芯 R200(150W、INT8 256 TOPS、108 路视频解码);信创训练集群可评估云燧T21(OAM 标准模组,8000+ 卡集群已验证)。

三、软件栈迁移成本评估​

迁移成本是国产替代里最容易低估的一项,各家生态进展差距明显:

  • CANN(华为):进展最快。CANN Next 提供 CUDA 兼容层,约 80% 的标准 PyTorch 推理代码仅需配置修改;2026 年 10 月 DeepSeek 与华为联合发布面向昇腾 950 的开源编程工具与库。生态拐点已现——站内昇腾 960 卡口径显示,CANN 外部开发者占比升至 61%,首次超越内部团队,月活跃开发者突破 5,200 人;PyTorch 官网也已将昇腾收录为中国算力平台。
  • NeuWare(寒武纪):云边端统一一套代码是国内独一份,PyTorch/TensorFlow 适配版齐全,MagicMind 是全球首批商用 MLIR 图编译推理框架;短板在千卡以上分布式训练的工程成熟度,单机 8 卡与百卡小集群表现优异。
  • MUSIFY(摩尔线程):算子覆盖率约 70%–80%(CUDA 为 99%+),复杂 LLM 需手工优化,但 Day-0 适配能力快速进步——2026 年 6–7 月连续完成 MiniMax M3、智谱 GLM-5.2、Qwen3.5 和美团万亿参数 LongCat-2.0 的适配。
  • IXUCA(天数智芯):高度兼容 CUDA 生态,官方口径迁移时间下降 50% 以上,已适配 200+ 模型,覆盖 LLM、CV、NLP、语音等领域。

四、供货现实与价格走势​

供需关系是当前国产算力市场的第一现实:

  • 需求侧:华为副董事长徐直军在 HC2026 访谈中的表态是"能产多少供多少"——需求远未见顶,第一约束已从市场转移到产能。
  • 产能侧:站内 950DT 卡引用的行业口径显示,中芯国际产能可支撑年产超 100 万颗昇腾芯片,但国产 HBM 产能仅够约 25 万–30 万颗成品加速器,缺口约六成——瓶颈在 HBM 而非逻辑制程。
  • 价格侧:HBM 成本攀升推高全线上调——950DT 渠道指示价超 ¥25 万(较两个月前 +20%–50%,Bloomberg 称三个月涨约 60%,已对标 B200)、950PR 突破 ¥8 万(+30%)、910C 超 ¥11 万(+22%)、寒武纪 690 等同期上调 20%–30%。预算必须按动态报价做,且注意以上均为渠道指示价,实际成交随采购规模与捆绑条件浮动。

五、混合部署策略与风险清单​

务实的企业路径是混合部署:NVIDIA 存量卡继续承担成熟训练负载与 CUDA 专有工作负载,国产卡承接增量需求——通常从推理切入(生态迁移成本最低),训练负载在 CANN/MUSA 成熟度验证后逐步迁移。昇腾 910C 集群已验证万亿参数大模型全参数后训练能力(站内卡记载的 DeepSeek V4 Pro 案例),训练迁移的技术可行性正在快速提高。

同时把风险清单贴在采购文档里:

  1. 生态风险:算子覆盖不全导致的性能回退与排期延误,先做 POC 再批量采购。
  2. 价格风险:涨价周期中锁价窗口短暂,长单尽量锁量锁价。
  3. 供货风险:HBM 约束短期无解,关键项目准备备选型号。
  4. 人才风险:非 CUDA 平台的工程师储备需要提前 6–12 个月建设。
  5. 残值风险:国产卡二手市场流动性弱于 NVIDIA,折旧年限要单独评估。

小结​

国产替代选型的正确姿势是:用五问框架定位自己的真实约束,按负载类型从对比表里筛型号(训练看昇腾与思元690、FP8 推理看 MTT S5000、低功耗视频看昆仑芯),把软件迁移成本算进总预算,在涨价与供货的现实下混合部署、增量替换。更多型号的全字段对比见站内完整对比表,各卡详细规格请进入对应芯片卡页面。

出口管制下的供应链重构:2026 年的版图变化

· 4 min read
Industry Research Team

2026 年,美国对华 AI 芯片出口管制完成了从"推定拒绝"到"逐案审查"的历史性转向,同时附加关税、产能与安全三重枷锁。管制形态从"按芯片型号划线"演变为"按买家总部划线",全球算力供应链随之重新绘制版图。

一年之内:政策的密集摇摆​

时间事件影响
2025 年 4 月H20 出口需申请许可英伟达计提 45 亿美元相关损失
2025 年 5 月AI 扩散规则被暂停全球分级许可框架搁置
2026 年 1 月H200 转为逐案审查TPP 低于 21000 且 DRAM 带宽低于 6500GB/s 的芯片可申请许可
2026 年 1 月同步加征 25% 关税相关芯片出口附带"收入分成"
2026 年 5 月D:5 总部指南生效管制对象从产品延伸到买家总部所在地
2026 年 8 月英伟达财报指引不假设任何来自中国的数据中心计算收入

新规则并非简单放松:对华出口总量受"不超过同期对美供应总 TPP 一半"的产能配额约束,每批产品须经美国第三方实验室检测验证,云端用途还需提交远程终端用户清单。英伟达甚至修改付款条款,要求中国客户全款预付,把政策风险完全转嫁给买方。

中国市场的受控交织​

尽管约束重重,需求依然旺盛。据公开报道,中国企业累计订购的 H200 级别芯片超过 200 万颗,字节跳动单年度采购金额据报高达 140 亿美元。作为对价,中国工信部引入"配对"要求:企业每进口一定数量海外芯片,须按比例配套采购以昇腾为主的国产处理器,形成"国产芯片承担推理与通用负载、海外芯片保留高价值训练"的混合算力部署模式。中美监管的双重合规压力,成为跨国采购团队的新常态。

与此同时,管制反向催化了国产替代的加速。据公开报道,华为 AI 芯片年收入正逼近 120 亿美元;BIS 甚至发布指南,提示昇腾 910B、910C 等芯片在全球范围内的交易也可能触发美方许可要求——管制的长臂正从硬件延伸到远程算力访问。2025 年 11 月出台、2026 年 11 月到期的"关联规则"暂停令,则让实体清单的覆盖范围悬而未决,进一步放大了政策的不确定性。

供应链的连锁反应​

合规物流让每颗输华芯片先赴美国检测再出口亚洲,交期延长、成本上升。更深层的挤压在制造端:H200 与 Blackwell 共享台积电 CoWoS 先进封装产能,对华订单直接排挤更高毛利产品的排产。英伟达已锁定台积电 2026 年约六成先进封装产能,其余厂商只能在剩余份额中竞争。2026 年 5 月,黄仁勋向媒体表示英伟达"基本已让出中国 AI 芯片市场"——对芯片巨头而言,中国市场的财务贡献正在从增长项退化为风险项。

结语:版图的三条断层线​

2026 年的供应链版图上,三条断层线已然清晰:美国以"逐案审查加关税"维持受控开放;中国以"配对采购加国产放量"构建双轨算力体系;其余国家则在两种体系之间摇摆观望。管制已经无法阻止算力需求的增长,只能改变算力的地理分布与成本结构。对行业观察者而言,比门槛数字更值得跟踪的,是每一次政策摇摆背后,供应链各方重新谈判的定价权。

NVIDIA's China Share Crashes to 8% as Ascend Rises to 50%: The Dramatic Reshuffle of China's AI Compute Market in a September Report

· 5 min read
Industry Research Team

This article is based on market analysis published by Bloomberg Intelligence and Bernstein in September 2026 and public reporting; market share and revenue figures are analyst estimates, not official disclosures.

On September 28, two Wall Street reports simultaneously painted a dramatic picture: the balance of market share in China's AI compute market completed an almost total flip within 12 months.

1. The Share Flip: 40% to 8%, Ascend to 50%​

Bernstein's latest forecast:

  • NVIDIA's share of the China AI chip market will fall from about 40% to about 8% by the end of this year
  • Huawei Ascend rises to nearly 50%
  • The immediate trigger: China's September ban on new H20 orders from NVIDIA, which shut the last channel for NVIDIA's return to the Chinese market
  • ByteDance, Alibaba, and Tencent have all placed large Ascend orders

On the timeline, this was not a sudden shock but the cumulative result of three years of tightening export controls: NVIDIA's share of the China AI chip market once peaked at 95%; as high-end chips were cut off, Chinese hyperscale cloud providers pivoted wholesale to Huawei and domestic alternatives. Even when Washington allowed H200 sales to China in January 2026, the gears of that shift had already meshed — there was no going back.

2. Ascend 950PR: The "Chinese H200" in Analysts' Eyes​

What underpins this share data is product strength itself:

  • The Ascend 950PR entered mass production in March this year, with FP4 compute of up to 2 PFLOPS and 128GB of domestic HBM
  • Analysts rate it as roughly on par with the NVIDIA H200
  • Huawei's AI chip revenue is expected to reach $12 billion this year, up 60% from $7.5 billion in 2025

For the full specifications of the 950 series, see our earlier in-depth articles: Ascend 950PR/950DT Dual-Configuration Analysis and Ascend 960 Official Launch — the 960 was ready three quarters ahead of schedule, confirming a one-generation-per-year cadence.

3. The Model Side: Chinese Open Models Take the OpenRouter Token Pie​

While chip shares flipped, the model usage data is equally striking (OpenRouter platform statistics):

MetricData
DeepSeek token share (June)16.3%, surpassing Google, Anthropic, and OpenAI individually to rank first
Combined token share of Chinese open-weight models (May)About 61% (DeepSeek, Qwen, MiniMax, Tencent Hunyuan, etc.)
Weekly token consumption of Chinese modelsAbout 18 trillion, versus about 5.5 trillion for US models — a gap of more than 3x, with the overtake completed within a year
China-US frontier model performance gap (June)Narrowed to 6%, a historic low (9% in May)

Compute and models form a positive feedback loop: cards that cannot be bought force domestic compute to scale up; scaled-up domestic compute produces cheap tokens; and cheap tokens let Chinese open-source models swallow the bulk of global inference traffic.

4. But the Stock Market Isn't Buying It​

Bloomberg Intelligence pointed out a contradiction in the same period: the valuation discount of the China Tech 8 relative to the US Magnificent Seven has widened to more than 50%, the widest of the year, and BI believes only a "true AI breakthrough" can close it. Year-to-date stock performance: Alibaba about -7%, Tencent about -23%, while US AI infrastructure leaders broadly gained more than 15%.

There are two readings of this divergence: either the market is underestimating China's AI fundamentals, or earnings execution (Alibaba's EPS missed expectations by 17.3% last quarter, Baidu's by 36%) has dragged down the delivery of the AI narrative. For industry observers, the notable point is: the compute-side data (share, revenue, shipments) has moved ahead, while application-side monetization is still on the way.

5. Three Takeaways for Compute Buyers​

  1. For deployments in China, the Ascend ecosystem is already the default option. The cost of migrating from CUDA to CANN is one-time, while supply uncertainty is persistent; after the September ban, there is no compliant path to procuring new NVIDIA cards in the Chinese market.
  2. Domestic compute "matching the H200" is a watershed. The implicit assumption of domestic substitution used to be "discounted performance, discounted price"; when single-card performance catches up to the H200 generation, procurement decisions return to pure TCO and supply-security calculations — we recommend using the TCO Calculator to put electricity, depreciation, and utilization into the same table.
  3. The token cost curve determines the application landscape. Chinese models' 61% share of OpenRouter tokens shows that the chain of compute self-sufficiency leading to token price drops leading to application prosperity has completed its first leg; the next question is whether inference costs can keep falling.

Summary​

From 40% to 8% is not a slogan but a market reshuffle driven jointly by the ban, product strength, and the model ecosystem. For the Chinese market, the mass-production cadence of Ascend 950/960 and the maturity of the CANN ecosystem will determine whether this share can be held; for the global market, the combination of China's compute "internal circulation + open-source models going overseas" is rewriting the geographic distribution of inference traffic.

(Market share and revenue forecasts come from Bernstein / Bloomberg Intelligence analyst reports; actual figures are subject to each company's financial reports.)

Three Signals for the Domestic Compute Ecosystem in One Week: China Telecom Open-Sources Ascend-Trained Model, Inspur 128-Chip Inference Appliance, EVAS Raises RMB 2 Billion

· 5 min read
Industry Research Team

Beyond the official Ascend 960 unveiling and the Zhenwu V900 launch, late September also delivered three ecosystem signals that headline coverage easily drowned out, yet whose structural significance rivals flagship chips: a carrier open-sourcing a model trained on Ascend, an OEM delivering a hundred-chip-class domestic inference system, and a cloud-compute chip startup closing a new funding round. This article breaks down each one.


1. China Telecom Open-Sources Xing4.0: Carrier-Grade Endorsement for End-to-End Ascend Training​

On September 23 (as reported), China Telecom open-sourced Xing4.0-29B-A4B — an agentic mixture-of-experts (MoE) model that the company says was trained end-to-end on Huawei Ascend accelerators.

Why it matters:

  • Third-party proof that "it can train": until now, evidence that "Ascend can train large models" came mainly from Huawei itself (40+ natively trained models); a carrier running training with its own data, engineering teams, and clusters — and open-sourcing it — is the first scaled training case outside the Huawei ecosystem;
  • Agentic positioning: the 29B-A4B (29B total / 4B active) MoE form factor targets agent workloads directly — aligned with the open-source cadence of DeepSeek and Qwen, rather than a small experimental model;
  • Open-source spillover: publishing model weights plus the training recipe means Ascend training engineering know-how can be reused by other institutions — the standard playbook for ecosystem diffusion.

Combined with Huawei's disclosure that "CANN has entered routine open-source operations, with external developers at 61%", the Ascend ecosystem is shifting from "vendor-led" to "community-operated".

2. Inspur MetaBrain SD200 Ultra: Measured Throughput Claims for a 128-Chip Domestic Inference System​

Inspur launched the MetaBrain SD200 Ultra inference system around the same time:

MetricValue (company figures)
Domestic AI chips128 chips (vendor did not disclose specific models)
Model servedKimi K3
Token throughput2.8T tokens (system throughput figure under that methodology)
Latency5.85 ms

Three readings:

  • Hundred-chip-class system integration: 128 chips cooperating across a heterogeneous setup to run very large MoE inference is a test of interconnect topology, scheduling, and fault tolerance — OEMs are now demonstrably capable of assembling domestic chips into large systems;
  • Aimed at top open-source models: targeting Kimi K3 (already live on AWS Bedrock) as the benchmark workload shows the acceptance standard for domestic inference systems is "runs today's most popular models", not a self-referential demo;
  • Mind the methodology: the 2.8T token throughput and 5.85ms latency are both vendor claims; the tested model version, concurrency, and batch size were not disclosed, so procurement evaluations should re-test under real workloads.

3. EVAS Closes RMB 2 Billion B+ Round: The Cloud Compute Chip Race Still Attracts Capital​

EVAS (Yixing Intelligence) completed a B+ round of RMB 2 billion, at a post-money valuation near RMB 15 billion, with Oriza Capital following on; earlier in the first half of the year it had closed a RMB 1.5 billion Series B and brought in China Mobile as a strategic investor. The company focuses on next-generation cloud compute chips.

In 2026, when the flagship chip lineup (Ascend / Cambricon / Moore Threads / Hygon / Zhenwu) looks settled, a cloud compute chip newcomer still raising RMB 2 billion suggests:

  • Primary-market conviction in the "second tier of domestic compute": with top vendors' capacity booked into 2027, overflow demand gives newcomers a window;
  • The role of carrier strategic investment: China Mobile is both a buyer and an investor — the demand side of domestic compute is using capital to lock in the supply side;
  • Differentiated room remains for cloud inference DSA routes (see the ecosystem positioning of routes like Tsingmicro and Houmo).

4. Assembling the Week's Signals: The Loop Is Taking Shape​

Put the three signals together with this month's flagship launches, and every segment of the domestic compute loop now has players filling the gaps:

SegmentThis Month's Evidence
Flagship chipsAscend 960 early official unveiling (9-17), Zhenwu V900 launch (9-22)
Training validationChina Telecom's Xing4.0 end-to-end trained on Ascend and open-sourced, DeepSeek betting on Ascend training
Inference systemsInspur SD200 Ultra running Kimi K3 on 128 chips
Software ecosystemCANN routine open-sourcing, external developers at 61%
Capital supplyEVAS B+ round of RMB 2 billion (post-money near RMB 15 billion)
Demand sideDeepSeek's alignment, Kimi K3 live on overseas cloud platforms

Independent players are investing across all four segments — "chips, systems, models, capital" — the biggest difference from the "isolated breakthroughs" of 2024-2025.

5. Takeaways​

  • China Telecom Xing4.0-29B: the first open-source model end-to-end trained on Ascend outside the Huawei ecosystem;
  • Inspur SD200 Ultra: a 128-chip domestic inference system at 2.8T token throughput / 5.85ms (company figures, pending re-testing);
  • EVAS RMB 2 billion B+ round: the cloud compute chip second tier is still getting real money;
  • What to watch: Xing4.0's training cluster scale and MFU, third-party re-tests of the SD200 Ultra, and EVAS's tape-out cadence.

Further Reading​

References​

  • The GPU Daily (2026-09-24): China Telecom open-sources Xing4.0-29B; Inspur MetaBrain SD200 Ultra
  • Toutiao (2026-09-21): EVAS completes RMB 2 billion B+ round, post-money valuation near RMB 15 billion
  • CSDN AI Daily (2026-09-21): Huawei CANN enters routine open-source operations, OceanStor M900 launched

This article is compiled from public reports. Xing4.0 training details and SD200 Ultra performance figures are company statements; testing conditions are subject to subsequent disclosures.

Domestic Big Three 2026 H2: Localization Rate Crosses 40% Toward 60%, Ascend 960 Roadmap, MLU690 and S5000 Ecosystems Ramp Up

· 6 min read
Industry Research Team

In 2026, China's AI chip market landscape has shifted from "NVIDIA unipolar dominance" to "overseas vendors leading, domestic multi-route catch-up." According to industry research, China's overall AI accelerator market was ~4M units in 2025, of which 1.65M were domestic, with share first breaking 40%; as products iterate and fabs follow up, the localization rate is expected to rise to 60%-70% by 2027. This article focuses on the latest H2 2026 progress of Huawei Ascend, Cambricon, and Moore Threads — the domestic "Big Three."


1. Huawei Ascend: 950 Capacity Fully Booked, 960 Roadmap Unveiled​

Ascend's core advantage is "architecture + full-stack ecosystem synergy," with ~800K units shipped in 2025, capturing 50% of the total domestic vendor share. The product iteration cadence is clear:

TimeProductNote
2025 Q1Ascend 910CMain transitional model
2026 Q1Ascend 950PRInference flagship
2026 Q4 (planned)Ascend 950DTTraining flagship, drives domestic HBM iteration
2027-2028Ascend 960 / 970Roadmap products

950 series capacity has entered a "fully booked" state: 950PR entered mass production in April 2026; June monthly capacity jumped to 500K-600K units (nearly 10x MoM), with a full-year target of 1.2M units at 100% certainty; ByteDance locked in 350K units for $5.6B, while Tencent / Alibaba / Baidu combined locked in 400K units.

Ascend 960 roadmap specs (per roadmap disclosure):

MetricAscend 960
ArchitectureAscend 6th gen (Da Vinci v6)
FP8 compute~4 PFLOPS
Memory288GB
Memory bandwidth9.6 TB/s
Super-nodeAtlas 960 SuperPoD, 15,488 cards, Lingqu optical-electrical converged bus
Debut2027 Q4 (roadmap)

The previous-gen Ascend 384 super-node has cumulatively shipped over 750 sets, deployed across 20+ industries including internet, operators, finance, education, and healthcare — Huawei calls it "the only domestic super-node that has trained a SOTA model."


2. Cambricon MLU690: H2 Mass Production, Entering ByteDance Bidding Window​

Cambricon is the core domestic compute leader in the absence of an Ascend IPO, with the technology gap continuously narrowing:

  • Siyuan 590 (7nm): Performance equivalent to 80% of A100, already supports DeepSeek, continuously adapting to mainstream large models like Qwen 3 and GLM
  • Siyuan 690 series: Will enter mass production in H2 2026, expected to achieve order scale-up during ByteDance's H2 bidding window
  • Revenue certainty: Equity incentive targets show >100% revenue growth for the next 3 years: 2026 revenue target 13.5B RMB, 2027 27B RMB, 2028 60B RMB

Cambricon fully benefits from the industry dividend of "domestic CSP capex + full adaptation of domestic large models and domestic chips," making it the most direct elasticity play on rising localization rate.


3. Moore Threads MTT S5000: Full-Function GPU + Ecosystem Breakthrough​

Moore Threads takes a differentiated "full-function GPU" route, with the flagship MTT S5000 based on the 4th-gen "Pinghu" MUSA architecture:

MetricMTT S5000
Dense AI compute1000 TFLOPS
Memory80GB
Memory bandwidth1.6 TB/s
Inter-card interconnect784 GB/s
PrecisionFP8 to FP64 full precision (training + inference)
SecurityFirst batch to pass national "Safe and Reliable Evaluation" (Level I)

Its engineering capability is verified: the Kuae (KUAE) intelligent computing cluster based on S5000 achieves 95% training linear scaling efficiency, with compute efficiency loss within 5% at ten-thousand-card scale; supports checkpoint-resume training with effective training time ratio >90%; and has trained a MoE-236B base model with >25 trillion tokens of corpus from scratch.

The ecosystem is Moore Threads' deepest moat: MUSA has achieved 100% core math library compatibility, 3000+ PyTorch operator compatibility, covers 55 categories of core AI operators, has official vLLM and SGLang support, Day-0 adaptation of mainstream models, and 800K+ developers. Its PD heterogeneous-disaggregation solution achieves equivalent replacement of international high-end GPUs at a 2:1 ratio with S5000, significantly reducing inference cost.

The 5th-gen "Huagang" architecture (released 2025-12) supports FP4 to FP64 full precision, with 50% higher compute density and 10x better energy efficiency than the previous gen, supporting 100K+ card clusters; cumulative R&D investment in the "Huashan" (train-infer integrated) and "Lushan" (graphics rendering) new chips based on this architecture exceeds 900M RMB.


4. Software Ecosystem Decides: Day-0 Adaptation Becomes Routine​

Beyond hardware, software ecosystem realization is the watershed for domestic compute in 2026:

  • Huawei's CANN heterogeneous computing architecture and MindSeries suite are fully open-sourced, with the community incubating 67 projects, 12.44M+ lines of code, and 3,500+ monthly active developers
  • The "release-and-adapt" closed loop between domestic large models and domestic chips has basically formed: Tencent Hunyuan T3 (295B), DeepSeek-V4, and GLM-5.2 all completed Day-0 adaptation
  • 2026 is regarded as the "first year of domestic super-nodes"; Huatai Securities estimates China's super-node architecture market will reach 341.4B RMB by 2028, with a 2026-2028 CAGR of 194%

5. Industry Judgment: From "Can It Be Built" to "Can It Be Used Well"​

The domestic Big Three are converging along three paths:

  1. Huawei: Locks government/enterprise and internet big customers with super-node system-level capability + full-stack software
  2. Cambricon: Impacts the revenue inflection point by narrowing the training-side gap + scaling up via big-customer bidding
  3. Moore Threads: Covers cloud-edge-end full scenarios with full-function GPU generality + mature CUDA-compatible ecosystem

The common shortcoming of all three remains advanced process and HBM supply — precisely the core link of overseas controls. But as domestic HBM iterates and fabs follow up, a realistic path to 60%-70% localization by 2027 exists.

References​


This article is compiled from public industry research, broker views, and corporate announcements as of August 2026. Some shipment and market-share figures are third-party estimates, not officially confirmed data.