📰 DigiTimesDigiTimesDoubao · 实盘不做把握 4/10不做原因:无增量信息
北京时间·🕒 发布 09-07 12:02:52·⟳ 抓取 09-07 18:03:49 +6h00m·⚡ 决策 09-15 12:46:45 +7d18h·42.0 天前
DigiTimes 复盘中国智驾芯片商 H1 2026：定点不等于量产、高算力撞上整车降本、车企自研挤压外购，地平线与黑芝麻硬件毛利率承压
这是 DigiTimes 一篇付费产业评论，全部论据（出货量、毛利率下滑、研发超收入、车企自研）来自地平线与黑芝麻早已披露的 H1 2026 财报，无任何新增订单、定点、指引或官方表态；方向虽对纯硬件芯片商偏负面，但属对旧财报的二次解读，不构成新催化，且距事件已 8 天、可能已被消化。
📄 原文DigiTimes↗ 原文
China smart driving chips hit bottlenecks in mass production, cost, in-house auto R&D
中文網 繁體 简体 SUBSCRIBE My account Sign out SUBSCRIBE Trending Apple Huawei Nvidia AI Search Scaling AI Glasses Inference E…
不做 地平线机器人（09660.HK）
无增量事实：引用数据为已披露 H1 2026 财报，本文为发布已 8 天的产业评论，偏负面但属旧闻二次解读，不构成事件驱动 edge
不做 黑芝麻智能（02533.HK）
无新增催化：研发投入与 A2000 进展均为既有信息，评论性内容无法提供可交易的预期差，且时点偏旧
展开详情(成交明细 / 研究档案 / 原文全文 / 决策记录链)
研究档案
结论摘要
文章是对已公开 H1 财报的结构性复盘，指出智驾芯片商面临量产转化、降本压毛利、车企自研替代和全栈研发烧钱四重压力；信息无增量、时点偏旧，无可交易 edge，故回避。
🔎 新鲜度:最早出现 2026-09-07 · 扩散 low · price-in 风险 high ↗
关键理由
- 核心财务数据（地平线出货 221.8 万套、毛利率 44.2%→36.2%、研发 27.55 亿元；黑芝麻研发 8.60 亿元）均为已披露中报内容，本文无新增事实
- 产业评论偏长期结构判断，缺乏订单/指引级别催化剂，且发布已 8 天，不具备短线事件驱动价值
主要风险
- 若市场此前未充分定价毛利率下滑与车企自研替代的长期叙事，相关标的情绪仍可能偏弱，但这属基本面趋势而非本条新闻的增量
- 车企自研（比亚迪/蔚来/小鹏）进展若出现新的具体芯片量产/上车事件，才会构成对地平线、黑芝麻的新增利空，需另行研判
先例如何影响了这次决策
对本次决策的影响:未改变结论：缺乏新增事实与新鲜度，直接回避，不进入仓位/买点判断
方向:无对齐先例；本条为旧财报的二次产业评论，本身不构成可锚定的事件催化，故不据此设定涨跌幅度
涨幅量级:无高相似、锚干净的同市场先例可供校准，且无交易意图，不定 tp/sl
形态:不适用
工具执行结果 · 成功 1 · 降级 0 · 失败 0 · 取消 0
- search_processed_events 成功 stock_tools HTTP · 18 ms
{"result": "search_processed_events:agent query=`horizon robotics black sesame smart driving chip bottleneck mass production margin | 地平线 黑芝麻 智能驾驶芯片 量产 成本 毛利率 h1 2026 | 征程系列 出货 221.8万套 毛利率 36.2% 研发费用`；结构化索引覆盖完整 72 小时；旧记录兼容扫描近 0.0/72 小时、0 条本臂决策(完整)，返回 0 条候选。未命中只表示这些 agent query 表达式没有召回候选，不能据此证明消息首发或不
原文全文
展开原文 · 9,384 字
中文網
繁體
简体
SUBSCRIBE
My account
Sign out
SUBSCRIBE
Trending Apple
Huawei
Nvidia
AI Search
Scaling AI Glasses
Inference Era AI
Advanced Packaging
Trending
SUBSCRIBE
CONNECT WITH US
Sign out
REALTIME NEWS
Corning sees CPO and glass core as new semiconductor battlegrounds
Tomorrow's Headlines
Sep 14, 18:47
China optical transceivers avoid US FCC curbs as NPO and CPO compete in high-speed links
Tomorrow's Headlines
Sep 14, 18:46
China's AIDC token costs hinge on access to advanced GPUs
Tomorrow's Headlines
Sep 14, 18:46
Home
Tech
Electric Vehicles
China smart driving chips hit bottlenecks in mass production, cost, in-house auto R&D
Staff reporter, Taipei
Sep 7, 2026, 12:02
0
Credit: Horizon Robotics
China's intelligent driving market is advancing rapidly, but the true trial has only just begun for domestic smart driving chipmakers.
Historically, competition among Chinese chip vendors was easily oversimplified into a "TOPS war," where securing higher compute specifications and locking in more vehicle design wins seemed to guarantee market dominance. However, as urban Navigate on Autopilot (NOA) transitions into commercial deployment, industry rivalry is shifting away from raw specification races toward far more pragmatic challenges: scaling chips into mass-production vehicles, driving down system costs, and sustaining continuous software updates.
The financial results for the first half of 2026 from Horizon Robotics and Black Sesame Technologies illustrate this structural transition.
1. Design wins fail to guarantee mass production volume
The first major bottleneck lies in the gap between securing design wins and achieving true mass-production. By the end of the first half of 2026, Horizon Robotics had secured nearly 500 vehicle model projects, including close to 130 mid-to-high-end intelligent driving programs, yet actual shipments for its Journey series processors totaled 2.218 million sets, representing a 12.1% year-over-year increase. Similarly, Black Sesame Technologies secured high-end program wins, with its Huashan A2000 processor nearing commercial production.
A substantial time lag exists between design designation and volume delivery. Once an automaker selects a chip platform, the product must undergo extensive chip validation, software stack integration, domain controller consolidation, functional safety testing, road trials, and production certification before translating into actual vehicle installations. Moving forward, evaluating smart driving chipmakers requires looking beyond design win totals to examine project-to-production conversion rates and the shipment volume per model post-launch.
2. High compute specifications conflict with automotive cost structures
Advanced autonomous driving demands increased compute capacity, but passenger cars are not data centers. Automakers will not continually inflate bill of materials (BOM) costs simply to acquire additional TOPS. Chip power consumption, thermal dissipation, advanced packaging, memory bandwidth, and domain controller integration directly dictate final vehicle retail pricing.
This represents one of the most pressing dilemmas for Chinese chip suppliers. As car companies seek to offer urban NOA capabilities at lower price points, chip vendors face severe pricing pressure even if they supply high-compute platforms.
During the first half of 2026, Horizon Robotics saw the gross margin of its core products and solutions business contract from 44.2% to 36.2%. Although its licensing and service division achieved a high gross margin of 90.4%, margin compression on the hardware side underscores a structural flaw in the traditional chip supply model. Higher chip specifications do not guarantee superior profitability; if automakers retain strong bargaining power, compute upgrades can quickly become a margin-eroding burden for silicon vendors.
3. In-house auto R&D reshapes market supply dynamics
A third bottleneck stems from Chinese automakers, including BYD, NIO, and Xpeng Motors, aggressively bolstering their internal autonomous driving capabilities, with several pursuing proprietary chip development. For merchant chip suppliers, market demand expansion no longer correlates directly with a proportional rise in silicon orders.
Automakers are investing in internal R&D not merely to reduce procurement costs, but to secure end-to-end control over the entire "chip-algorithm-data-vehicle" technology stack. As urban NOA transitions from an optional feature into a primary selling point for smart electric vehicles (EVs), the underlying autonomous driving system is no longer viewed as a component, but as a critical axis for brand differentiation.
Consequently, merchant chipmakers relying solely on hardware sales face eroding bargaining power. Their position in the supply chain will increasingly depend on their capacity to deliver comprehensive software toolchains, foundational AI models, development platforms, and turnkey autonomous driving solutions. Horizon Robotics' recent expansion into licensing and service models reflects this strategic shift as silicon vendors pivot from pure hardware suppliers into platform-level partners.
4. Advanced autonomous driving demands full-system capabilities
The fourth bottleneck highlights that high-end autonomous driving requires complete system integration. As autonomous driving models evolve from legacy perception algorithms toward end-to-end architectures, world models, and vision-language-action (VLA) frameworks, the chip represents one component of the broader stack. CPUs, NPUs, ISPs, memory, toolchains, compilers, model optimization engines, and vehicle-side software must operate in tight synchronization.
This architectural requirement explains why Black Sesame emphasizes pairing its A2000 silicon with proprietary NPUs, software SDKs, and AI toolchains, while Horizon Robotics continues to advance its HSD platform and foundation models. Both vendors are actively mitigating their commercial exposure to pure hardware sales.
However, building end-to-end system capabilities requires massive capital allocation. In the first half of 2026, Horizon Robotics allocated CNY2.755 billion (US$410.5 million) to R&D, exceeding its revenue of CNY2.055 billion (US$306.2 million) for the same period. Black Sesame's R&D expenditure reached CNY860 million (US$128.1 million), representing nearly 1.9 times its total revenue.
This financial strain highlights the core dilemma facing Chinese smart driving chipmakers: scaling back R&D risks ceding market share to in-house auto development and international computing platforms, yet maintaining heavy R&D spending prolongs financial losses.
Ultimately, the next phase of competition among Chinese smart driving chipmakers will no longer hinge on raw compute metrics. Success will be determined by effective compute utilization, software ecosystem maturity, mass-production deployment efficiency, and the ability to enable automakers to commercialize advanced driving features at manageable cost structures.
China's smart driving market possesses an abundance of high-compute chips and design wins; the scarcity lies in scalable, cost-controlled, software-upgradable turnkey solutions that automakers are willing to standardize on for the long term.
Article
translated by Yunnie Cheng and edited by Jerry Chen
AI Search
Related stories
BYD's 4nm smart-driving chip signals deeper EV supply-chain integration
How physical AI and ADAS-cockpit fusion are rewiring China's smart driving supply chain
Nvidia leads China assisted-driving chip market; Horizon Robotics rises to second
China's Horizon Robotics aims at Tesla with new self-driving platform
China automotive AI chipmaker Horizon Robotics targets EV cost cuts with integrated chip
Show more
Categories
Automotive IC EV
Tags
autonomous driving China chipmakers chips cost design Horizon Robotics market production software vehicle
Share this article
Other links
Print
Related topics
Global supply chain: Emerging tech, businesses
Global supply chain: China market
Asia
Regions
BIZ FOCUS
Sep 14, 08:00
Hauman Unveils Micro LED Testing Breakthrough for AI Optical Interconnects
Monday 14 September 2026
Scaling Photonics: Assembly, Test and Fiber Automation at SEMICON Taiwan
Friday 11 September 2026
DAS Expands in Taiwan and Unveils World First at SEMICON
Friday 11 September 2026
Toward Technologies Redefines Relay Solutions at SEMICON Taiwan 2026
MOST-READ
7 DAYS NEWS
Google's v10 TPU opens door to AMD without sidelining Broadcom, MediaTek
Update: ABF substrate maker Unimicron searched in origin-labeling probe
Qualcomm's HBC play pulls Samsung and SK hynix into a new race beyond HBM
SK keyfoundry expands 8-inch capacity as China demand strains supply
Samsung, SK Hynix step up China NAND upgrades as US tool risks grow
Column: When digital trust breaks — quantum risk rewrites supply chain resilience
Wuliangye skips carmaking, turns liquor waste into silicon-carbon anodes
The 1,000-watt problem: a TSMC veteran on the brute force forging Nvidia's AI chips
eCloudvalley expands enterprise AI Agent portfolio
CXMT rides AI boom to 80% operating margin
Full list
Members only
×
Sorry, the page you are trying to open is available only for our paid subscribers.
Please login to read more
New users, please register first
Notification
×
繁體
简体
SUBSCRIBE
My account
Sign out
SUBSCRIBE
Trending Apple
Huawei
Nvidia
AI Search
Scaling AI Glasses
Inference Era AI
Advanced Packaging
Trending
SUBSCRIBE
CONNECT WITH US
Sign out
REALTIME NEWS
Corning sees CPO and glass core as new semiconductor battlegrounds
Tomorrow's Headlines
Sep 14, 18:47
China optical transceivers avoid US FCC curbs as NPO and CPO compete in high-speed links
Tomorrow's Headlines
Sep 14, 18:46
China's AIDC token costs hinge on access to advanced GPUs
Tomorrow's Headlines
Sep 14, 18:46
Home
Tech
Electric Vehicles
China smart driving chips hit bottlenecks in mass production, cost, in-house auto R&D
Staff reporter, Taipei
Sep 7, 2026, 12:02
0
Credit: Horizon Robotics
China's intelligent driving market is advancing rapidly, but the true trial has only just begun for domestic smart driving chipmakers.
Historically, competition among Chinese chip vendors was easily oversimplified into a "TOPS war," where securing higher compute specifications and locking in more vehicle design wins seemed to guarantee market dominance. However, as urban Navigate on Autopilot (NOA) transitions into commercial deployment, industry rivalry is shifting away from raw specification races toward far more pragmatic challenges: scaling chips into mass-production vehicles, driving down system costs, and sustaining continuous software updates.
The financial results for the first half of 2026 from Horizon Robotics and Black Sesame Technologies illustrate this structural transition.
1. Design wins fail to guarantee mass production volume
The first major bottleneck lies in the gap between securing design wins and achieving true mass-production. By the end of the first half of 2026, Horizon Robotics had secured nearly 500 vehicle model projects, including close to 130 mid-to-high-end intelligent driving programs, yet actual shipments for its Journey series processors totaled 2.218 million sets, representing a 12.1% year-over-year increase. Similarly, Black Sesame Technologies secured high-end program wins, with its Huashan A2000 processor nearing commercial production.
A substantial time lag exists between design designation and volume delivery. Once an automaker selects a chip platform, the product must undergo extensive chip validation, software stack integration, domain controller consolidation, functional safety testing, road trials, and production certification before translating into actual vehicle installations. Moving forward, evaluating smart driving chipmakers requires looking beyond design win totals to examine project-to-production conversion rates and the shipment volume per model post-launch.
2. High compute specifications conflict with automotive cost structures
Advanced autonomous driving demands increased compute capacity, but passenger cars are not data centers. Automakers will not continually inflate bill of materials (BOM) costs simply to acquire additional TOPS. Chip power consumption, thermal dissipation, advanced packaging, memory bandwidth, and domain controller integration directly dictate final vehicle retail pricing.
This represents one of the most pressing dilemmas for Chinese chip suppliers. As car companies seek to offer urban NOA capabilities at lower price points, chip vendors face severe pricing pressure even if they supply high-compute platforms.
During the first half of 2026, Horizon Robotics saw the gross margin of its core products and solutions business contract from 44.2% to 36.2%. Although its licensing and service division achieved a high gross margin of 90.4%, margin compression on the hardware side underscores a structural flaw in the traditional chip supply model. Higher chip specifications do not guarantee superior profitability; if automakers retain strong bargaining power, compute upgrades can quickly become a margin-eroding burden for silicon vendors.
3. In-house auto R&D reshapes market supply dynamics
A third bottleneck stems from Chinese automakers, including BYD, NIO, and Xpeng Motors, aggressively bolstering their internal autonomous driving capabilities, with several pursuing proprietary chip development. For merchant chip suppliers, market demand expansion no longer correlates directly with a proportional rise in silicon orders.
Automakers are investing in internal R&D not merely to reduce procurement costs, but to secure end-to-end control over the entire "chip-algorithm-data-vehicle" technology stack. As urban NOA transitions from an optional feature into a primary selling point for smart electric vehicles (EVs), the underlying autonomous driving system is no longer viewed as a component, but as a critical axis for brand differentiation.
Consequently, merchant chipmakers relying solely on hardware sales face eroding bargaining power. Their position in the supply chain will increasingly depend on their capacity to deliver comprehensive software toolchains, foundational AI models, development platforms, and turnkey autonomous driving solutions. Horizon Robotics' recent expansion into licensing and service models reflects this strategic shift as silicon vendors pivot from pure hardware suppliers into platform-level partners.
4. Advanced autonomous driving demands full-system capabilities
The fourth bottleneck highlights that high-end autonomous driving requires complete system integration. As autonomous driving models evolve from legacy perception algorithms toward end-to-end architectures, world models, and vision-language-action (VLA) frameworks, the chip represents one component of the broader stack. CPUs, NPUs, ISPs, memory, toolchains, compilers, model optimization engines, and vehicle-side software must operate in tight synchronization.
This architectural requirement explains why Black Sesame emphasizes pairing its A2000 silicon with proprietary NPUs, software SDKs, and AI toolchains, while Horizon Robotics continues to advance its HSD platform and foundation models. Both vendors are actively mitigating their commercial exposure to pure hardware sales.
However, building end-to-end system capabilities requires massive capital allocation. In the first half of 2026, Horizon Robotics allocated CNY2.755 billion (US$410.5 million) to R&D, exceeding its revenue of CNY2.055 billion (US$306.2 million) for the same period. Black Sesame's R&D expenditure reached CNY860 million (US$128.1 million), representing nearly 1.9 times its total revenue.
This financial strain highlights the core dilemma facing Chinese smart driving chipmakers: scaling back R&D risks ceding market share to in-house auto development and international computing platforms, yet maintaining heavy R&D spending prolongs financial losses.
Ultimately, the next phase of competition among Chinese smart driving chipmakers will no longer hinge on raw compute metrics. Success will be determined by effective compute utilization, software ecosystem maturity, mass-production deployment efficiency, and the ability to enable automakers to commercialize advanced driving features at manageable cost structures.
China's smart driving market possesses an abundance of high-compute chips and design wins; the scarcity lies in scalable, cost-controlled, software-upgradable turnkey solutions that automakers are willing to standardize on for the long term.
Article
translated by Yunnie Cheng and edited by Jerry Chen
AI Search
Related stories
BYD's 4nm smart-driving chip signals deeper EV supply-chain integration
How physical AI and ADAS-cockpit fusion are rewiring China's smart driving supply chain
Nvidia leads China assisted-driving chip market; Horizon Robotics rises to second
China's Horizon Robotics aims at Tesla with new self-driving platform
China automotive AI chipmaker Horizon Robotics targets EV cost cuts with integrated chip
Show more
Categories
Automotive IC EV
Tags
autonomous driving China chipmakers chips cost design Horizon Robotics market production software vehicle
Share this article
Other links
Related topics
Global supply chain: Emerging tech, businesses
Global supply chain: China market
Asia
Regions
BIZ FOCUS
Sep 14, 08:00
Hauman Unveils Micro LED Testing Breakthrough for AI Optical Interconnects
Monday 14 September 2026
Scaling Photonics: Assembly, Test and Fiber Automation at SEMICON Taiwan
Friday 11 September 2026
DAS Expands in Taiwan and Unveils World First at SEMICON
Friday 11 September 2026
Toward Technologies Redefines Relay Solutions at SEMICON Taiwan 2026
MOST-READ
7 DAYS NEWS
Google's v10 TPU opens door to AMD without sidelining Broadcom, MediaTek
Update: ABF substrate maker Unimicron searched in origin-labeling probe
Qualcomm's HBC play pulls Samsung and SK hynix into a new race beyond HBM
SK keyfoundry expands 8-inch capacity as China demand strains supply
Samsung, SK Hynix step up China NAND upgrades as US tool risks grow
Column: When digital trust breaks — quantum risk rewrites supply chain resilience
Wuliangye skips carmaking, turns liquor waste into silicon-carbon anodes
The 1,000-watt problem: a TSMC veteran on the brute force forging Nvidia's AI chips
eCloudvalley expands enterprise AI Agent portfolio
CXMT rides AI boom to 80% operating margin
Full list
Members only
×
Sorry, the page you are trying to open is available only for our paid subscribers.
Please login to read more
New users, please register first
Notification
×
决策记录链 · 只追加
每次领取产生一条记录,重新决策追加一条并以 supersedes 指向被替代的记录(ADR 0014)。
这条决策早于决策账本(2026-10-08 起逐条追加),没有追加记录。