查看購物車
  • English
  • Japanese
  • Korean
封面
市場調查報告書
商品編碼
2064025

駕駛座與駕駛整合：中央網域控制器SoC與AI超級運算架構市場（2026年）

Cockpit-Driving Integration Central Domain Controller SoC and AI Supercomputing Architecture Research Report, 2026
出版日期: | 出版商: ResearchInChina | 英文 750 Pages | 商品交期: 最快1-2個工作天內

價格
簡介目錄

駕駛座與駕駛的融合，以及人工智慧超級運算的探索——單晶片解決方案正在車輛中迅速部署，而人工智慧超級運算架構正朝著完全領域融合的方向發展。

AI 超級運算架構的配置－雖然 OEM 廠商專注於全面整合，但一級供應商正透過軟體和硬體解決方案進入這一領域。

目前，中國主要汽車製造商正在建造由「一個中央運算平台+兩到三個ZCU」組成的EEA系統。大多數汽車製造商都在為其車輛配備統一的「中央大腦」和高效的「神經網路」，並在此基礎上逐步整合人工智慧應用。

長城汽車的GWM*ONE平台是首個以服務導向的架構（SOA）為核心基礎系統的原生AI平台。它創新地將車輛細分為300多個冗餘功能單元，透過全端式標準化服務合約系統，實現了車輛間的自由模型呼叫和跨場景的無縫協作。

該平台基於 Coffee EEA 4.0，整合了人工智慧作業系統和雙 VLA 模型，實現了車輛感知、決策和執行的統一管理。此外，全球首創的仿生運動控制系統打破了底盤、自動駕駛和動力傳動系統系統之間的界限，實現了車輛的協同控制。它還支援多種控制功能，包括±10°後輪轉向和蟹行功能集，即使車輛在以超過160公里/小時的速度直線行駛時一側兩個輪胎被扎破，也能保持穩定的煞車性能。

上汽IM的超級智慧體「IM Ultra Agent」由三大關鍵技術支柱構成：IM Fusion Nova駕駛座駕駛一體化全局融合架構、與Momenta合作開發的IM AD ZETA自動駕駛平台模型以及可量產車搭載的阿里巴巴Qwen模型。此外，還發布了全線控「靈曦數位底盤」作為執行平台。在底層異質運算架構層面，此架構將線傳、智慧駕駛AI和智慧座艙AI三大核心系統完全連接，建構了「世界大腦+敏捷小腦+執行體」的三層架構。

全球大腦：Qwen 模型負責理解使用者意圖、安排全球場景和協調多工處理。

敏捷小腦：IM AD ZETA 負責駕駛場景中的決策、風險預測和車輛運動控制；

執行主體：全數位線控線傳系統能夠準確、快速地將人工智慧決策轉換為車輛的實體動作。

大眾汽車集團正透過與小鵬汽車、地平線機器人等公司的合作，在本地部署全領域智慧體人工智慧：

CEA 1.0/1.3（2025-2026）：目前 CEA 1.0 及其改良版本 CEA 1.3 的核心在於完成從集中式架構到半集中式+區域架構的過渡，從而實現 800V 高壓、8295駕駛座晶片、高級自動駕駛和全局 OTA 等核心技術的部署。這些技術將涵蓋大眾、奧迪和斯柯達等多個品牌，涵蓋不同尺寸（從 A 級到 D 級）和動力傳動系統配置（純電動車、增程器汽車等）。預計 2026 年將推出大量新車型。

CEA 2.0/3.0（2027-2030）：CEA 2.0 將透過深度整合中央運算和中央閘道器，進一步提升整合度，並預裝 L4 級自動駕駛硬體，實現車輛與雲端的整合互動。 CEA 3.0 將採用原生 AI 平台模型，實現軟硬體完全分離，並支援按需訂閱和功能持續迭代。大眾汽車也在自主研發晶片。其晶片子公司 CARIZON 正在開發一款代號為「C7H」的晶片，該晶片的運算能力可在 600 至 2400 TOPS 之間靈活擴展。這款晶片針對多模態平台模型進行了最佳化，預計到 2028 年將在汽車領域大規模應用。

軟體和硬體供應商正圍繞人工智慧超級運算架構，在中央運算平台、AI作業系統、人工智慧代理和基礎模型等領域提案前瞻性解決方案。 ThunderSoft計劃於2026年預發布「Aqua Drive OS 2.0」，其定位為「原生AI車載操作系統」，並強調與AIBOX的深度整合。這不僅有助於基礎模型在汽車領域的大規模應用，還將支援人工智慧代理在各種場景下的部署。

全新升級的Aqua Drive OS 2.1基於NVIDIA Nemo Claw參考軟體堆疊，並針對高通驍龍8397/8797進行了全域最佳化。它基於標準化的L+A架構，全面實現了包括AI推理、BSP、中間件和代理管理在內的全鏈路功能。此外，還提案了整合的「AIOS + AIBOX」解決方案。 AIBOX搭載NVIDIA DRIVE AGX，可提供高達200 TOPS的AI運算能力和205 GB/s的傳輸頻寬，使其成為首款能夠在裝置上流暢運行70億參數基礎模型的產品。

2026年將是「物理人工智慧」的元年。屆時，車輛將透過攝影機、雷射雷達和觸覺感測器等「感測」手段獲取現實世界的信息，並將這些資訊與基於人工智慧的模型相結合，從而理解物理定律並做出決策，並透過電機和關節等「肢體」輸出一系列物理動作。最終，車輛將根據實際回饋不斷修正錯誤並自我進化。自動駕駛是實體人工智慧在智慧駕駛領域應用的關鍵場景。

QCraft的實體AI模型是基於一種名為「世界模型+強化學習」的整合架構。該架構由雲端世界模型和車載行為模型組成，實現了三大進步：增強推理能力、提高決策準確性、擴展泛化能力。

基於雲端的世界模型扮演著「創造者」的角色。它使用自然語言指令產生極端駕駛場景，創建大量在真實道路測試中不太可能遇到的長尾案例，並利用這些案例不斷訓練底層模型。

汽車世界模型扮演著「執行者」的角色，它是一個封裝了可變長度架構（VLA）和線上世界模型的元件。透過「世界模型+強化學習」的整合架構，多模態感知和即時軌跡產生緊密協作，完成了從預測到決策的即時封閉回路型。

QCraft全球模式的核心在於雲端與車輛的聯合訓練，這不僅涵蓋了90%的常規場景，還能有效減少剩餘10%邊緣場景的盲點。在此架構下，基於500 TOPS車規級運算平台的「Driven-by-QCraft MAX」自動駕駛解決方案，其效能可媲美數千TOPS的都市區NOA體驗。

AI 超級運算平台－「單晶片」解決方案將逐漸成為主流，國產晶片的採用速度也將加快。

車輛電子架構（EEA）正從分散式向域集中式，再向集中式演進。駕駛座與駕駛的融合是跨域融合的核心環節，「單晶片」解決方案則是駕駛座與駕駛融合的終極型態。業界經歷了兩個階段：「一盒兩板」（將兩塊板整合在同一網域控制器上）和「一盒一板」（將兩顆晶片整合在同一塊PCB上）。 2025年，單晶片解決方案終於開始量產。 NVIDIA Thor（2000 駕駛座算力）、Qualcomm SA8797P（320 TOPS高密度算力）以及Horizo​​​​n Robotics「Starry」等原生駕駛座與駕駛融合晶片相繼問世，使得在單晶片上同時支援L3/L4級自動駕駛、多螢幕與駕駛融合晶片相繼問世，使得在單晶片上同時支援L3/L4級自動駕駛、多螢幕空間與駕駛融合裝置以及未來的平台

2026年，高通SA8775P/SA8797和黑芝麻武當C1296等單晶片解決方案將進入量產週期。例如，基於驍龍8797的AutoLink World駕駛座及駕駛員整合控制器預計將於2026年開始量產。隨著小鵬、蔚來、理想汽車等新興OEM廠商快速將自研晶片解決方案（圖靈晶片、神機NX9031、Mach 100）整合到車輛中，以及地平線星空全面進軍生態系統，單晶片解決方案將在駕駛座及駕駛員整合市場佔越來越大的佔有率。

2025年10月，全新Arcfox aT5正式發布，成為全球首款基於高通SA8775P晶片的量產車型，該車型整合了駕駛座和駕駛功能。車輛搭載了AutoLink World的網域控制器解決方案，該方案與前置雷達、12個超音波雷達和7個攝影機（前置立體攝影機、環景顯示攝影機和後視攝影機）協同工作，實現了L2+高速公路NOA、整合式停車功能和HPA（高速公路自動駕駛輔助系統）。

Leapmotor D19 將於 2026 年 4 月發布，搭載兩顆高通驍龍 8797 晶片。其最大運算能力為 1280 TOPS，支援基於基礎模型的駕駛座和基於 VLA 的智慧駕駛功能。

本田汽車正式宣布與瑞薩電子就R-Car X5項目展開合作。下一代本田0系車型將升級至半集中式+區域架構。中央ECU將採用台積電3nm製程及多晶片技術的SoC。瑞薩電子第五代R-Car X5 SoC將與本田自主研發的AI加速器結合，目標是達到高達2000 TOPS的AI運算效能。

在中國，有黑芝麻武當C1296、地平線機器人「星光」和一汽紅旗1號等單晶片解決方案。

黑芝麻武當C1296採用7nm先進工藝，CPU運算性能約208K DMIPS，GPU運算性能約1.5 TFLOPS，NPU運算性能為76 TOPS。其「硬體隔離+虛擬機管理程式」架構首次實現了智慧駕駛座、自動駕駛、車身控制等功能域的硬體級資源整合和安全隔離。同時，它還提供了豐富的運算能力和介面，單晶片即可覆蓋從智慧駕駛座、駕駛/泊車一體化到車輛運算等多種核心場景。目前，該晶片已正式應用於東風汽車的「天元智慧駕駛座Plus」，並率先搭載於東風eπ007車型，全面支援基礎版、語音互動、L2+智慧駕駛和FAPA功能。

一汽紅旗1號晶片是一款多域融合晶片，整合了駕駛輔助、智慧駕駛座、車輛控制、通訊和安全五大功能域。與業界主流域融合晶片（如高通SA8775）相比，其邏輯處理能力提升21.7%，影像處理能力提升15.4%。該晶片採用獨立安全島設計，實現了硬體級隔離，符合ASIL-D功能安全等級，並滿足「國家密碼局認證的國產密碼演算法」二級資訊安全要求。目前，該晶片已成功研發完成，即將進入車輛測試階段。

地平線機器人推出的「星銳」：中國首款整合駕駛座與自動駕駛車輛代理晶片「星銳6P」，採用5nm工藝，擁有650 TOPS的運算能力和273 GB/s的頻寬，支援設備本地運行基礎模型。它還支援6至12塊螢幕的資訊顯示，滿足車上娛樂系統和多媒體顯示的渲染需求。採用「堡壘安全」物理分離架構，實現了駕駛座與自動駕駛的實體分離與獨立運作。結合車輛代理操作系統「卡卡爪」和HSD自動駕駛系統，為領先的AI超級運算架構奠定了基礎。基於「星銳6」的整合式駕駛座與自動駕駛車輛智慧解決方案計畫與奇瑞iCar V27一同全球首發。

AI 超級運算通訊架構－下一階段將引進汽車光纖通訊技術。

人工智慧超級運算架構對車載設備的互連提出了嚴格的要求，包括高頻寬、穩定性、可靠性、冗餘保護、低延遲以及確定性的汽車骨幹網路。目前，為了滿足大量資料傳輸的需求，骨幹網路正從CAN匯流排升級到Gigabit/萬兆乙太網路。未來，隨著人工智慧超級運算架構的演進，汽車光纖通訊將成為滿足汽車高頻寬通訊需求的有效且重要的解決方案。光纖通訊的廣泛應用以及擺脫銅線的趨勢在汽車產業日益明顯，汽車光子技術正步入產業化階段。

汽車光纖通訊是指以「光波」為資訊載體、光纖為傳輸介質，透過光訊號在光纖內傳輸數據，並利用電/光和光/電訊號的相互轉換實現資訊傳輸的通訊技術。這使得車載各種電控系統（ECU、感知器、顯示設備、計算單元等）之間能夠實現高速、即時、抗干擾的數據互連。

目前，汽車光纖通訊在中國境內外主要有兩種部署路徑：

1. 車載光纖乙太網路：基於光纖乙太網路技術的發展，採用點對點通訊模式。其核心是交換功能，使其適用於東西向流量主導的應用場景，並符合IEEE 802.3cz協定；

2. 車載PON：基於傳統光PON技術開發，採用點對多點通訊模式。其核心特徵是無源性。由於基於被動光分路器，無需開關或主動元件，因此非常適合以南北向交通為主的車載應用場景。

兩者通用在於，它們都是基於現有技術（PON 和乙太網路）的新技術，並且是針對汽車環境設計和最佳化的。然而，汽車乙太網路和汽車 PON 在功能和性能方面存在顯著差異。這些差異使得乙太網路本身難以滿足汽車通訊網路的所有需求。

目前，中國汽車製造商正在提案一系列與「汽車光纖乙太網路」相關的產品，並積極與一級供應商合作，開發用於汽車製造商實際車輛檢驗的系統解決方案。例如，理想汽車與恆興科技共同開發了一款汽車光纖通訊測試平台，該平台已通過A樣交付和B樣測試，計劃於2026年實現量產並應用於車輛。

2026年1月，AutoLink World與中基科技旗下子公司ReinOCS Technologies共同發表了「Deep Fusion EEA」解決方案。此方案由三個部分組成：基於高通驍龍8797的中央運算平台、基於AMD Versal AI Edge Gen 2的網域控制器，以及基於光PCIe通訊技術的高速光纖通訊骨幹網路。中央平台為AI提供強大的運算能力和靈活的調度能力，網域控制器確保功能整合和快速執行，而高速光纖通訊則突破了傳統線束的瓶頸，支援更高密度的感知和資料流傳輸。

在 Deep Fusion EEA 大會上，ReinOCS Technologies 提供了高速光纖通訊解決方案，展現了其具有競爭力的基礎技術能力：

適應各種場景：支援8K/60Hz超高清影像無損傳輸，滿足智慧駕駛座多螢幕整合以及與高解析度感測器（雷射雷達、高解析度攝影機等）進行自動駕駛資料整合的需求。最大傳輸距離達100米，可實現車內多區域設備的互聯互通。

適應惡劣環境：採用輕量化設計和電磁干擾抑制技術，即使在-40 度C至85 度C的惡劣汽車環境中也能穩定運作。這解決了傳統銅線電纜在複雜電磁環境下傳輸訊號衰減的問題。

獨特的光學和結構設計：模組和線束均能承受汽車環境中全頻寬的高加速度衝擊和持續振動，完全符合汽車線束和連接器的可靠性標準。即使在頻繁啟動和崎嶇路面等複雜工況下，也能確保穩定、連續的資料傳輸。

架構擴充性：模組化介面設計可無縫滿足未來高階自動駕駛（L4以上）的多感測器融合需求。這為OEM廠商提供了一種「硬體可升級、軟體可重複使用」的靈活解決方案，降低了車輛電子架構的長期升級成本。

目錄

產業概況 1：歐洲經濟區 OEM 的總體趨勢

產業概況2：中國乘用車市場人工智慧超級運算架構的整體趨勢

產業概況3：中國乘用車市場人工智慧代理的整體趨勢

第1章：駕駛座整合市場規模與滲透率預測

  • 中國乘用車中央運算架構的預測採用率
  • 中國乘用車中央運算架構的採用率（2023-2030 年預測）
  • 附表：中國乘用車中央運算架構滲透率（2023-2030 年預測）
  • 中國乘用車駕駛座與駕駛一體化解決方案的預測採用率
  • 中國乘用車駕駛座與駕駛一體化解決方案的分類邏輯
  • 中國乘用車整合式駕駛座駕駛解決方案的普及率（2023-2030 年預測）
  • 附表：中國乘用車駕駛座與駕駛一體化解決方案滲透率（2023-2030 年預測）
  • 中國乘用車中央運算平台的預測普及率
  • 駕駛座整合解決方案的演進：整合形式：單盒、單板、單晶片
  • 中央運算平台（單板單晶片）採用率：2025-2030 年預測
  • 2025年至2030年中央運算平台（單板多晶片）的採用率
  • 2025-2030年中央運算平台（整合多板系統）的採用率
  • 附表：中國乘用車駕駛座整合解決方案滲透率（按整合類型分類，2025-2030 年預測）

第2章：駕駛座與駕駛座椅整合及人工智慧超級運算架構的發展趨勢概述

  • 駕駛座駕駛整合架構設計：中央運算 + ZCU 架構
  • 汽車產業歐洲經濟區表單迭代過程
  • 汽車產業 EEA 表單 - 中央運算 + ZCU 架構
  • 車輛中心+區域架構的部署
  • 車輛中心+區域架構的發展方向
  • 原始設備製造商：駕駛座整合和人工智慧超級運算整合解決方案概述
  • 一級供應商：人工智慧超級運算融合與人工智慧盒子解決方案概述
  • 駕駛座整合和人工智慧超級運算整合趨勢

第3章：國內OEM廠商的AI超級運算架構設計策略

  • Great Wall Motor
  • GAC Group
  • SAIC Motor
  • Chery
  • Changan Automobile
  • Geely
  • BAIC
  • FAW
  • BYD
  • Dongfeng Motor
  • AIOS
  • Li Auto
  • NIO
  • XPeng
  • Leapmotor
  • Xiaomi Auto
  • Harmony Intelligent Mobility Alliance (HIMA)
  • 實體建築

第4章：海外OEM廠商的AI超級運算架構佈局策略

  • Tesla
  • Volkswagen
  • BMW
  • Daimler
  • Toyota
  • Honda
  • AIOS
  • SAIC-GM
  • Nissan

第5章：一級供應商提供的軟體和硬體解決方案

  • Neusoft Reach
  • SenseAuto
  • QCraft
  • DeepRoute.ai
  • Banma Intelligence
  • Thundersoft
  • Momenta
  • AFARI
  • ECARX
  • Full-Stack Technology Product Matrix
  • Multi-Domain Integration Products
  • Zenith Central Computing Platform
  • Autolink
  • Hangsheng Electronics
  • BICV
  • Zhuoyu Technology
  • Modelbest Technology
  • Rivotek
簡介目錄
Product Code: CL013

Cockpit-Driving integration and AI supercomputing research: The One Chip solution is rapidly installed in vehicles, and AI supercomputing architectures are moving towards full-domain integration.

AI supercomputing architecture layout: OEMs focus on full-domain integration, while Tier 1 suppliers enter the field with software and hardware solutions.

Currently, leading Chinese OEMs have built an EEA system of "1 central computing platform + 2-3 ZCUs". Most OEMs have equipped their vehicles with a unified "central brain" and efficient "neural network", and are gradually integrating AI applications on this basis.

Great Wall Motor's GWM*ONE platform, as the first native AI full-power platform, uses SOA as its core in its underlying system. It innovatively granulates a vehicle into more than 300 multiplexable functional units, and achieves free cross-vehicle model invocation and seamless cross-scenario collaboration through a full-stack standardized service contract system.

Based on Coffee EEA 4.0, the platform integrates the AI OS and dual VLA models to achieve unified management of vehicle perception, decision and execution. Moreover, its world's first bionic motion control system opens up the boundaries between chassis, intelligent driving and powertrain system, realizing vehicle coordinated control, and supports +-10° rear wheel steering and crab mode, and enables stable braking in straight line single-sided dual-wheel tire burst at 160+ km/h and other control function clusters.

SAIC IM's super agent, IM Ultra Agent, is composed of three major technical pillars at the bottom - the IM Fusion Nova cockpit-driving integration full-domain fusion architecture, the IM AD ZETA intelligent driving foundation model developed with Momenta, and Alibaba Qwen Model that is available in production vehicles. The fully wire-controlled Lingxi Digital Chassis was also released as an execution carrier. From the underlying heterogeneous computing architecture level, this architecture completely connects the three core systems of chassis-by-wire, intelligent driving AI, and intelligent cabin AI, and builds a three-layer architecture of "global cerebrum + agile cerebellum + execution body":

Global cerebrum: Qwen Model is responsible for understanding user intentions, global scenario scheduling, and multi-task coordination;

Agile cerebellum: IM AD ZETA is responsible for driving scenario decision, risk prediction, and vehicle motion control;

Execution body: A fully digital chassis-by-wire is responsible for accurately and quickly converting AI decisions into vehicle physical actions.

Volkswagen Group is deploying full-domain agent AI locally through cooperation with companies such as XPeng and Horizon Robotics:

CEA 1.0/1.3 (2025-2026): The core of the current CEA 1.0 and its iterative version CEA 1.3 is to complete the transformation from a centralized architecture to a quasi-central + zonal architecture, and realize implementation of core technologies such as 800V high voltage, 8295 cockpit chip, high-level intelligent driving, and full-domain OTA, covering multiple brands such as Volkswagen, Audi, and Skoda with varying sizes (from Class A to D) and powertrain forms (battery-electric, extended-range, etc.). A number of new cars will be launched intensively in 2026.

CEA 2.0/3.0 (2027-2030): CEA 2.0 will deeply integrate the central computing and central gateway to further improve the level of integration, pre-install L4 intelligent driving hardware, and enable vehicle-cloud integrated interaction. CEA 3.0 will be embedded in native AI foundation models to achieve complete decoupling of software and hardware, and support on-demand subscription of functions, and continuous iteration. Volkswagen also self-develops chips. Its chip company, CARIZON, is developing a chip code-named C7H whose computing power can be flexibly expanded between 600~2400TOPS. It is optimized for multi-modal foundation models and is expected to find large-scale automotive application in 2028.

Software and hardware suppliers have proposed forward-looking solutions around the AI supercomputing architectures in terms of central computing platforms, AIOS, AI agents, and foundation models. The Aqua Drive OS 2.0 Pre released by ThunderSoft in 2026 is directly positioned as an "AI native vehicle operating system", and emphasizes its in-depth collaboration with AIBOX, which can not only promote the large-scale automotive application of foundation models, but also support the implementation of AI agents in scenarios.

The newly upgraded Aqua Drive OS 2.1 is based on the NVIDIA Nemo Claw reference software stack and has been globally adapted on Qualcomm Snapdragon 8397/8797. Relying on the standardized L+A architecture, it fully implements full-link capabilities including AI reasoning, BSP, middleware and agent management. The "AIOS+AIBOX" integrated solution was also proposed. The AIBOX is equipped with NVIDIA DRIVE AGX, which can provide up to 200TOPS of AI computing power and 205GB/s transmission bandwidth, and is the first to smoothly run 7B foundation models on the device.

2026 is the first year of physical AI. Vehicles obtain information from the real world via "senses" such as cameras, LiDAR, and tactile sensors, combine it with AI foundation models to understand physical laws and make decisions, and then output a range of physical actions through "limbs" such as motors and joints. Finally, they continue to correct errors and evolve themselves from real feedback. Autonomous driving is an important scenario for implementation of physical AI in the field of intelligent driving.

QCraft's physical AI model is based on the "world model + reinforcement learning" unified architecture. It is composed of cloud world model + vehicle world behavior model, which can achieve three major upgrades: stronger reasoning, better decision, and wider generalization.

The cloud world model plays the role of "creator": using natural language instructions to generate extreme driving scenarios, batch production of long-tail cases that are difficult to encounter in real road tests, and used for continuous training of foundation models.

The automotive world model is the "executor", which is an encapsulated VLA and online world model. With the world model + reinforcement learning unified architecture, multi-modal perception and real-time trajectory generation are deeply coupled to complete a real-time closed loop from prediction to decision.

The core of QCraft's world model lies in the joint cloud-vehicle training, which can not only cover 90% of regular scenarios, but also reduce blind spots in the remaining 10% of edge scenarios. Under this architecture, the "Driven-by-QCraft MAX" intelligent driving solution, based on the 500TOPS automotive computing power platform, can benchmark against the urban NOA experience with thousands of TOPS.

AI supercomputing platform: The One Chip solution will gradually become mainstream, and domestic chips will be deployed more rapidly.

Vehicle EEAs are evolving from distributed to domain centralized and then to centralized forms. Cockpit-Driving integration is the core link of cross-domain fusion, and One Chip is the final form of cockpit-driving integration. The industry has passed through two stages: One Box/Two Board (two boards are integrated on the same domain controller) and One Box/One Board (two chips are integrated on the same PCB). In 2025, the mass production of One Chip solutions just began. Native cockpit-driving integration chips represented by NVIDIA Thor (2000TOPS computing power), Qualcomm SA8797P (320TOPS dense computing power), and Horizon Robotics "Starry" have been launched one after another, making it possible for a single chip to simultaneously support L3/L4 intelligent driving, multi-screen cockpit interaction, APA and other functions, acting as the core carrier of the future central computing platform.

In 2026, single-chip solutions such as Qualcomm SA8775P/SA8797 and Black Sesame Wudang C1296 will enter the mass production cycle. For example, AutoLink World's cockpit-driving integration controller based on Snapdragon 8797 is expected to be mass-produced in 2026; as the self-developed chip solutions (Turing chip, Shenji NX9031, Mach 100) of emerging OEMs such as XPeng, NIO, and Li Auto are quickly available in vehicles and Horizon Starry enters all ecosystems, One Chip solutions will take an increasing share in cockpit-driving integration.

In October 2025, new Arcfox aT5, the world's first production vehicle model with cockpit-driving integration based on Qualcomm SA8775P, was officially launched. It is equipped with AutoLink World's domain controller solution, which is connected to a front radar, 12 ultrasonic radars, and 7 cameras (front view stereo camera, surround view, rear view), realizing L2+ Highway NOA, and integrated parking and HPA.

The Leapmotor D19, launched in April 2026, is equipped with dual Qualcomm Snapdragon 8797 chips. With computing power up to 1280 TOPS, they can realize foundation model-based cockpits and VLA-based intelligent driving functions.

Honda Motor has officially announced cooperation with Renesas R-Car X5. The next-generation vehicle models of the Honda 0 Series will be upgraded to a quasi-central + zonal architecture. The central ECU is planned to be equipped with an SoC using TSMC's 3nm process and multi-die chiplet technology. Renesas' fifth-generation R-Car X5 SoC will be combined with Honda's self-developed AI accelerator, with the target AI computing power of up to 2000TOPS.

In China, there are One Chip solutions like Black Sesame Wudang C1296, Horizon Robotics "Starry" and FAW Hongqi No.1.

Black Sesame Wudang C1296: Based on the 7nm advanced process, the CPU computing power is about 208K DMIPS, the GPU computing power is about 1.5TFLOPS, and the NPU computing power is 76TOPS. With the "hardware isolation + Hypervisor" architecture, the resource integration and safe isolation of functional domains such as intelligent cockpit, intelligent driving, and body control are realized at the hardware level for the first time. Simultaneously, it possesses abundant computing power and interface capabilities, with a single chip covering multiple core scenarios ranging from intelligent cockpit and driving-parking integration to vehicle computing. Currently, it has been officially designated for Dongfeng's "Tianyuan Intelligent Cockpit Plus", and will be first installed in Dongfeng eπ 007, fully supporting foundation models and voice interaction, L2+ intelligent driving, and FAPA.

FAW Hongqi No.1: Defined as a multi-domain fusion chip, it integrates five functional domains: driving assistance, intelligent cockpit, vehicle body control, communication, and safety; compared to mainstream domain fusion chips in the industry (such as Qualcomm SA8775), its logic computing capability is improved by 21.7%, and its image processing capability is improved by 15.4%; it has a built-in independent security island, hardware-level isolation, supports ASIL-D functional safety, and meets the Level 2 information security requirements of "home-made cryptographic algorithms recognized by the National Cryptographic Administration of China". Currently, it has been successfully developed and is ready to enter the vehicle testing phase.

Horizon Robotics "Starry": As the first cockpit-driving integrated vehicle agent chip in China, Starry 6P uses a 5nm process, 650TOPS computing power and 273GB/s bandwidth to support local operation of on-device foundation models. It also supports information display of 6-12 screens to meet the rendering requirements of automotive entertainment and multimedia display. It adopts the Fortress-safe physical isolation architecture to achieve physical isolation and independent operation of the cockpit and intelligent driving. Combined with the vehicle agent operating system KakaClaw and the HSD intelligent driving system, it lays the foundation for AI supercomputing architecture in advance. The cockpit-driving integration vehicle intelligent solution based on Starry 6 will be first launched globally with Chery iCar V27.

AI supercomputing communication architecture: automotive fiber optic communication technology will be introduced in the next stage.

The AI supercomputing architecture also places stringent requirements on automotive equipment interconnection, requiring high bandwidth, stable and reliable, redundant protection, low latency, and deterministic on-board backbone networks. At present, the backbone network has been upgraded from CAN bus to Gigabit/10G Ethernet to meet the demand for massive data transmission. In the future, with the evolution of AI supercomputing architecture, automotive optical communication will become an effective way and key solution to support communications with high bandwidth requirements in automobiles. The trend of optical advancement and copper withdrawal in the automotive industry has emerged, and automotive optical communication technology is entering industrialization.

Automotive fiber optic communication refers to the communication technology that uses "light waves" as the information carrier and optical fiber as the transmission medium, transmits data in the optical fiber through optical signals, and realizes information transmission by electrical/optical and optical/electrical mutual conversion, thereby realizing high-speed, real-time, and anti-interference data interconnection between various in-vehicle electronic control units (ECUs, sensors, display devices, computing units, etc.).

Currently, there are two main routes for autonomous optical communication inside and outside China:

1. Automotive fiber optic Ethernet: Based on the development of fiber optic Ethernet technology, it adopts a point-to-point communication mode. Its core lies in switching. It is suitable for application scenarios dominated by east-west traffic and follows the IEEE802.3cz protocol;

2. Automotive PON: PON is developed based on traditional optical PON technology and adopts a point-to-multipoint communication mode. Its core feature is passivity. Built on passive optical splitters, it eliminates the need for switches and active components, making it well-suited for in-vehicle application scenarios dominated by north-south traffic.

The similarity between the two is that they are new technologies based on existing technologies - PON and Ethernet, and designed and optimized to adapt to the automotive environment. However, there are major differences in functions and performance between automotive Ethernet and automotive PON. These differences make it difficult for Ethernet to serve as an independent technology to cater to all requirements of automotive communication networks.

Currently, Chinese OEMs have proposed many products related to "automotive fiber optic Ethernet" and are actively collaborating with Tier 1 suppliers to develop system solutions for real-vehicle verification by OEMs. For example, Li Auto and Hinge Technology have jointly developed an automotive optical communication test bench, which has passed A-sample delivery and B-sample testing, with a goal of mass production and automotive deployment in 2026.

In January 2026, AutoLink World and ReinOCS Technologies, a subsidiary of Zhongji Innolight, jointly released Deep Fusion EEA, which consists of three parts: a central computing platform based on Qualcomm 8797, a domain controller based on AMD Versal AI Edge Gen 2, and a high-speed optical communication backbone network based on optical PCIe communication technology. The central platform provides high computing power AI and flexible scheduling, and the domain controller ensures function convergence and rapid execution. High-speed optical communication breaks through the bottleneck of traditional wiring harnesses, and the transmission supports higher density sensing and data flows.

In Deep Fusion EEA, the high-speed optical communication solution is provided by ReinOCS Technologies, demonstrating competitive underlying technical capabilities:

All-Scenario Adaptability: It supports lossless transmission of 8K@60Hz ultra-high-definition video, meeting the data interaction needs of multi-screen interaction in intelligent cockpits and autonomous driving high-resolution sensors (such as LiDAR, high-definition cameras), with a maximum transmission distance of up to 100 meters, covering vehicle multi-region device connections;

Extreme Environmental Adaptability: Utilizing lightweight design and anti-electromagnetic interference technology, it can operate stably in extreme automotive environments ranging from -40°C to 85°C, solving the signal attenuation problem of traditional copper cable transmission in complex electromagnetic environments;

Special Optical and Structural Design: Both modules and wiring harnesses can withstand high-acceleration shocks and continuous vibrations across the entire frequency band in automotive scenarios, fully meeting the reliability standards for automotive wiring harnesses and connections. Even in complex operating environments such as frequent vehicle starts and stops and bumpy road conditions, it can ensure the stability and continuity of data transmission;

Architectural Scalability: Adopting a modular interface design, it can seamlessly adapt to the multi-sensor fusion requirements of future high-level autonomous driving (L4 and above), providing OEMs with a flexible solution of " upgradeable hardware and iterable software," reducing the long-term upgrade costs of vehicle electronic architecture.

Table of Contents

Industry Summary 1: Overall Trend of OEM EEA

Industry Summary 2: Overall Trend of AI Supercomputing Architectures in China's Passenger Car Market

Industry Summary 3: Overall Trend of AI Agents in China's Passenger Car Market

1 Forecast for Cockpit-Driving Integration Market Size and Penetration Rate

  • 1.1 Forecast for Penetration Rate of Central Computing Architecture in Local Passenger Cars in China
  • Penetration Rate of Central Computing Architecture in Local Passenger Cars in China, 2023-2030E
  • Attached Table: Penetration Rate of Central Computing Architecture in Local Passenger Cars in China, 2023-2030E
  • 1.2 Forecast for Penetration Rate of Cockpit-Driving Integration Solutions in Passenger Cars in China
  • Classification Logic of Cockpit-Driving Integration Solutions for Passenger Cars in China
  • Penetration Rate of Cockpit-Driving Integration Solutions in Passenger Cars in China, 2023-2030E
  • Attached Table: Penetration Rate of Cockpit-Driving Integration Solutions in Passenger Cars in China, 2023-2030E
  • 1.3 Forecast for Penetration Rate of Central Computing Platforms in Passenger Cars in China
  • Evolution of Cockpit-Driving Integration Solution Integration Forms: One Box, One Board, One Chip
  • Penetration Rate of Central Computing Platforms (One Board One Chip), 2025-2030E
  • Penetration Rate of Central Computing Platforms (One Board Multi Chips), 2025-2030
  • Penetration Rate of Central Computing Platforms (One Box Multi Boards), 2025-2030
  • Attached Table: Penetration Rate of Cockpit-Driving Integrated Solutions in Local Passenger Cars in China by Integration Form, 2025-2030E

2 Summary of Cockpit-Driving Integration and AI Supercomputing Architecture Trends

  • 2.1 Cockpit-Driving Integration Architecture Design: Central Computing + ZCU Architecture
  • Iterative Process of Automotive EEA Forms
  • Automotive EEA Forms: Central Computing + ZCU Architecture (1)
  • Automotive EEA Forms: Central Computing + ZCU Architecture (2)
  • Deployment Forms of Vehicle Central + Zonal Architecture (1)
  • Deployment Forms of Vehicle Central + Zonal Architecture (2)
  • Deployment Forms of Vehicle Central + Zonal Architecture (3)
  • Deployment Forms of Vehicle Central + Zonal Architecture (4)
  • Development Directions of Vehicle Central + Zonal Architecture (1)
  • Development Directions of Vehicle Central + Zonal Architecture (2)
  • Development Directions of Vehicle Central + Zonal Architecture (3)
  • Development Directions of Vehicle Central + Zonal Architecture (4)
  • 2.2 OEMs: Summary of Cockpit-Driving Integration and AI Supercomputing Integration Solutions
    • 2.2.1 AI Supercomputing Architectures of OEMs
    • Summary of AI Supercomputing Architectures of OEMs (1)
    • Summary of AI Supercomputing Architectures of OEMs (2)
    • Summary of AI Supercomputing Architectures of OEMs (3)
    • Summary of AI Supercomputing Architectures of OEMs (4)
    • 2.2.2 Cockpit AI Solutions of OEMs
    • Summary of Cockpit AI Solutions of OEMs (1)
    • Summary of Cockpit AI Solutions of OEMs (2)
    • Summary of Cockpit AI Solutions of OEMs (3)
    • Summary of Cockpit AI Solutions of OEMs (4)
    • 2.2.3 Intelligent Driving AI Solutions of OEMs
    • Summary of Intelligent Driving AI Solutions of OEMs (1)
    • Summary of Intelligent Driving AI Solutions of OEMs (2)
    • Summary of Intelligent Driving AI Solutions of OEMs (3)
    • Summary of Intelligent Driving AI Solutions of OEMs (4)
    • 2.2.4 Chassis AI Solutions of OEMs
    • Summary of Chassis AI Solutions of OEMs (1)
    • Summary of Chassis AI Solutions of OEMs (2)
    • Summary of Chassis AI Solutions of OEMs (3)
    • Summary of Chassis AI Solutions of OEMs (4)
    • Summary of Powertrain AI Solution of OEMs (1)
    • Summary of Powertrain AI Solution of OEMs (2)
    • Summary of Powertrain AI Solution of OEMs (3)
  • 2.3 Tier 1 Suppliers: Summary of AI Supercomputing Fusion and AI Box Solutions
    • 2.3.1 AI Supercomputing Fusion Solutions of Tier1 Suppliers
    • Summary of Central Computing Platform Solutions of Tier1 Suppliers (1)
    • Summary of Central Computing Platform Solutions of Tier1 Suppliers (2)
    • Summary of Central Computing Platform Solutions of Tier1 Suppliers (3)
    • Summary of Vehicle Operating System Solutions of Tier1 Suppliers (1)
    • Summary of Vehicle Operating System Solutions of Tier1 Suppliers (2)
    • 2.3.2 AI Box Solutions of Tier1 Suppliers
    • Cockpit AI Box (1)
    • Cockpit AI Box (2)
    • Cockpit AI Box (3)
    • Cockpit AI Box (4)
    • Summary of OEM Cockpit AI Box Products (1)
    • Summary of OEM Cockpit AI Box Products (2)
    • Cockpit AI Box Product Solution 1 (1-3)
    • Cockpit AI Box Product Solution 2
    • Cockpit AI Box Product Solution 3
    • Cockpit AI Box Product Solution 4
    • Cockpit AI Box Product Solution 5
    • Cockpit AI Box Product Solution 6
  • 2.4 Summary of Solutions of Chip Vendors
  • Cockpit-Driving Integration Central Domain Controller Soc Classification Logic
  • Central Domain Controller SoC Selection for Cockpit-Driving Integration Solutions (1)
  • Central Domain Controller SoC Selection for Cockpit-Driving Integration Solutions (2)
  • One Board One Chip Solutions (1)
  • One Board One Chip Solutions (2)
  • One Board One Chip Solutions (3)
  • Central Domain Controller Platform with One Board One Chip
    • 2.4.1 Qualcomm SA8797
    • Qualcomm SA8797P-based Cockpit-Driving Integration Domain Controller Solutions and Main Customers
    • Qualcomm SA8797P: Performance Features (1)
    • Qualcomm SA8797P: Performance Features (2)
    • Qualcomm SA8797P: Performance Features (3)
    • Qualcomm SA8797P-based Cockpit-Driving Integration Domain Controller Solution 1
    • Qualcomm SA8797P-based Cockpit-Driving Integration Domain Controller Solution 2
    • Qualcomm SA8797P-based Cockpit-Driving Integration Domain Controller Solution 3
    • Qualcomm SA8797P-based Cockpit-Driving Integration Domain Controller Solution 4
    • 2.4.2 Qualcomm 8775
    • Qualcomm SA8775P-based Cockpit-Driving Integration Domain Controller Solutions and Main Customers
    • Qualcomm SA8775P: Performance Features (1-4)
    • Qualcomm SA8775P-based Cockpit-Driving Integration Domain Controller Solutions
    • 2.4.3 Black Sesame Wudang C1296
    • Black Sesame Wudang C1296-based Cockpit-Driving Integration Domain Controller Solutions and Main Customers
    • Black Sesame Wudang C1296 (1-3)
    • Black Sesame Wudang C1296-based Cockpit-Driving Integration Domain Controller Solution (1-3)
    • 2.4.4 Horizon Robotics "Starry"
    • Horizon Robotics "Starry" (1)
    • Horizon Robotics "Starry" (2)
    • Horizon Robotics "Starry" (5)
    • Vehicle Agent Operating System (1)
    • Vehicle Agent Operating System (2)
    • Intelligent Driving Solution: HSD v1.6
    • 2.4.5 Renesas R-Car X5H
    • Gen5 Multi-Domain Fusion SoC (1)
    • Gen5 Multi-Domain Fusion SoC (2)
    • Gen5 Multi-Domain Fusion SoC (6)
    • Gen5 R-CAR (1)
    • Gen5 R-CAR (2)
    • 2.4.6 "Hongqi No.1" of FAW Group
    • Hongqi No.1
    • 2.4.7 NVIDIA Thor
    • NVIDIA DRIVE Thor-based Cockpit-Driving Integration Domain Controller Solutions and Main Customers
    • NVIDIA DRIVE Thor (1-4)
    • NVIDIA DRIVE Thor-based Cockpit-Driving Integration Domain Controller Solution (1-3)
    • 2.4.8 Qualcomm 8397
    • Qualcomm AI Intelligent Cockpit Chip (1)
    • Qualcomm AI Intelligent Cockpit Chip (2)
    • Qualcomm AI Intelligent Cockpit Chip Solution
    • 2.4.9 Mach 100 of Li Auto
    • Mach 100 (1-4)
    • 2.4.10 NIO Shenji NX9031
    • Shenji NX9031 (1-4)
    • 2.4.11 "Turing" AI Chip of XPeng
    • "Turing" Chip (1-4)
    • 2.4.12 SiEngine Longying No.2
    • SiEngine's Cockpit-Driving Integration Chip
    • SiEngine's AI Accelerator Chip
    • 2.4.13 Black Sesame Huashan A2000
    • Black Sesame Huashan (1)
    • Black Sesame Huashan (2)
    • Black Sesame Huashan (5)
    • Black Sesame Huashan Solution
    • 2.4.14 AXERA M97
    • AXERA's Cockpit-Driving Integrated Sensing Chip (1)
    • AXERA's Cockpit-Driving Integrated Sensing Chip (2)
    • AXERA's Cockpit-Driving Integrated Sensing Chip (3)
    • 2.4.15 SemiDrive X10
    • SemiDrive's AI Intelligent Cockpit Chip: X10 (1-4)
    • 2.4.16 MediaTek C-X1
    • MediaTek's AI Intelligent Cockpit Chip (1-4)
  • 2.5 Integration Trend of Cockpit-Driving Integration and AI Supercomputing
    • 2.5.1 Trend 1: MPAM Technology Improves the Anti-Interference Capability of Cockpit-Driving Integration SoCs
    • MPAM Technology (1)
    • MPAM Technology (2)
    • MPAM Technology (3)
    • MPAM Technology (4)
    • 2.5.2 Trend 2: Automotive Fiber Optic Communication Is Adapted to AI Supercomputing Architectures
    • AIDVs (AI-Defined Vehicles) (1)
    • AIDVs (AI-Defined Vehicles) (2)
    • Automotive Fiber Optic Communication Automotive Fiber Optic Communication Architecture Design
    • Comparison between Automotive Fiber Optic Ethernet vs. Automotive Optical PON (1-3)
    • Requirements for Applying Optical Communication Technology to Vehicles
    • Automotive Fiber Optic Communication: Key Technologies and Technical Challenges (1)
    • Automotive Fiber Optic Communication: Key Technologies and Technical Challenges (2)
    • Summary of OEM Automotive Fiber Optic Communication Layout Strategies
    • Tier 1 Automotive Fiber Optic Communication Architecture Solution 1-3
    • 2.5.3 Trend 3: Next-Generation Automotive Storage Technology Is Being Laid out
    • Automotive Storage Demand (1)
    • Automotive Storage Demand (2)
    • Development Process of Automotive Storage Technology
    • Automotive Storage Technology: LPDDR5X
    • Automotive Storage Technology: Next-Generation LPDDR6
    • Automotive Storage Technology: Next-Generation UFS 4.1
    • 2.5.4 Trend 4: AI Agents Enter the Automotive Cockpit Field
    • AI Agents (Agentic AI) (1-4)
    • AI Agents (Agentic AI) Enter the Automotive Cockpit Field
    • Cross-Domain Integration AI Agent Layout of OEMs (1)
    • Cross-Domain Integration AI Agent Layout of OEMs (2)
    • Cross-Domain Integration AI Agent Layout of Tier 1 Suppliers (1-4)
    • 2.5.5 Trend 5: Autonomous Driving Is the "Best Entry Point" for Physical AI
    • Physical AI (1)
    • Physical AI (2)
    • Physical AI (5)
    • Summary of Physical AI Layout Strategies of OEMs (1)
    • Summary of Physical AI Layout Strategies of OEMs (2)
    • Summary of Physical AI Layout Strategies of Tier 1 Suppliers

3 AI Supercomputing Architecture Layout Strategies of Domestic OEMs

  • 3.1 Great Wall Motor
  • AI Supercomputing Architecture (1-2)
  • Physical Architecture (1-3)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture Layout Strategy (3)
  • Foundation Models
  • Chassis AI
  • Vehicle Models with AI Supercomputing Architecture
  • 3.2 GAC Group
  • AI Supercomputing Architecture Layout (1)
  • AI Supercomputing Architecture Layout (2)
  • Physical Architecture (1-5)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture Layout Strategy (3)
  • Cockpit AI (1)
  • Cockpit AI (2)
  • Cockpit AI (5)
  • Intelligent Driving AI
  • Powertrain AI (1-3)
  • Chassis AI
  • 3.3 SAIC Motor
  • AI Supercomputing Architecture Layout (1-3)
  • AI Supercomputing Architecture (1-2)
  • AI Supercomputing Architecture Layout Strategy (1-3)
  • Intelligent Driving AI (1-2)
  • Cockpit AI (1-2)
  • Chassis AI (1-2)
  • Roewe's AI Supercomputing Architecture (1-3)
  • 3.4 Chery
  • AI Supercomputing Architecture (1)
  • AI Supercomputing Architecture (2)
  • Physical Architecture (1)
  • Physical Architecture (2)
  • Physical Architecture (3)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture Layout Strategy (5)
  • Intelligent Driving AI
  • Cockpit AI
  • Chassis AI
  • Powertrain AI
  • 3.5 Changan Automobile
  • AI Supercomputing Architecture Layout
  • AI Supercomputing Architecture (1)
  • AI Supercomputing Architecture (2)
  • Physical Architecture: SDA 2.0 (1-4)
  • AI Supercomputing Architecture Layout Strategy (1-4)
  • Intelligent Driving AI
  • Cockpit AI
  • Chassis AI (1-3)
  • 3.6 Geely
  • AI Supercomputing Architecture Layout (1-2)
  • Physical Architecture (1-2)
  • ZEEKR's Physical Architecture (1)
  • ZEEKR's Physical Architecture (2)
  • AI Supercomputing Architecture Layout Strategy (1-4)
  • Cockpit AI (1-3)
  • Intelligent Driving AI (1-3)
  • Power AI
  • Chassis AI (1-4)
  • Robotaxi's AI Supercomputing Architecture Layout
  • Robotaxi's Physical Architecture (1)
  • Robotaxi's Physical Architecture (2)
  • Robotaxi's Intelligent Driving AI
  • 3.7 BAIC
  • AI Supercomputing Architecture Layout (1-5)
  • Physical Architecture (1-2)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture (1-3)
  • 3.8 FAW
  • AI Supercomputing Architecture (1-4)
  • Physical Architecture
  • AI Supercomputing Architecture Layout Strategy (1-3)
  • Foundation Models
  • Intelligent Driving AI (1-2)
  • Cockpit AI (1-3)
  • Chassis AI (1-2)
  • 3.9 BYD
  • AI Supercomputing Architecture (1-3)
  • AI Supercomputing Physical Architecture (1-3)
  • AI Supercomputing Architecture Layout Strategy (1-3)
  • Foundation Models (1-2)
  • Intelligent Driving AI (1-2)
  • Cockpit AI
  • Chassis AI (1-2)
  • Power AI
  • 3.10 Dongfeng Motor
  • AI Supercomputing Architecture Layout (1)
  • AI Supercomputing Architecture Layout (2)
  • Physical Architecture (1)
  • Physical Architecture (2)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • Voyah's AI Supercomputing Architecture Layout Strategy
  • Foundation Models
  • AIOS
  • Cockpit AI (1-3)
  • Voyah's AI Supercomputing Architecture Layout (1)
  • Voyah's AI Supercomputing Architecture Layout (2)
  • Voyah's AI Supercomputing Architecture Layout (5)
  • 3.11 Li Auto
  • Physical Architecture (1-3)
  • AI Supercomputing Architecture Layout Strategy (1-3)
  • AIOS (1-4)
  • Intelligent Driving AI (1)
  • Intelligent Driving AI (2)
  • Intelligent Driving AI (3)
  • Intelligent Driving AI (1)
  • Intelligent Driving AI (2)
  • Intelligent Driving AI (3)
  • Cockpit AI (1)
  • Cockpit AI (2)
  • Cockpit AI (3)
  • Chassis AI
  • 3.12 NIO
  • AI Supercomputing Architecture
  • Physical Architecture (1)
  • Physical Architecture (2)
  • Physical Architecture (5)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AIOS (1-6)
  • Intelligent Driving AI (1-3)
  • Cockpit AI (1-3)
  • Chassis AI (1-5)
  • 3.13 XPeng
  • AI Supercomputing Architecture Layout (1)
  • AI Supercomputing Architecture Layout (2)
  • AI Supercomputing Architecture Layout Strategy (1-3)
  • Robotaxi AI Supercomputing Architecture Layout Strategy
  • AIOS (1)
  • AIOS (2)
  • Cockpit AI (1-3)
  • Intelligent Driving AI (1-4)
  • Robotaxi Chassis AI (1-4)
  • Robotaxi Intelligent Driving AI (1-2)
  • 3.14 Leapmotor
  • Physical Architecture
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • Intelligent Driving AI
  • Intelligent Driving AI
  • Chassis AI (1-4)
  • 3.15 Xiaomi Auto
  • Physical Architecture (1)
  • Physical Architecture (2)
  • AI Supercomputing Architecture Layout Strategy (1-3)
  • Cockpit AI
  • Intelligent Driving AI (1)
  • Intelligent Driving AI (2)
  • Chassis AI (1-2)
  • 3.16 Harmony Intelligent Mobility Alliance (HIMA)
  • Physical Architecture
  • AI Supercomputing Architecture Layout Strategy (1-5)
  • Intelligent Driving AI (1-5)
  • Cockpit AI (1-3)
  • Chassis AI (1-4)
  • Powertrain AI (1-4)

4 AI Supercomputing Architecture Layout Strategies of Overseas OEMs

  • 4.1 Tesla
  • Cybertruck's Physical Architecture (1)
  • Cybertruck's Physical Architecture (2)
  • Physical Architecture (1)
  • Physical Architecture (2)
  • AI Supercomputing Architecture Layout Strategy (1-3)
  • Intelligent Driving AI (1-3)
  • Cockpit AI (1-3)
  • Cybertruck's Chassis AI (1)
  • Cybertruck's Chassis AI (2)
  • 4.2 Volkswagen
  • AI Supercomputing Architecture Layout (1)
  • AI Supercomputing Architecture Layout (2)
  • AI Supercomputing Architecture (1)
  • AI Supercomputing Architecture (2)
  • AI Supercomputing Architecture (3)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture Layout Strategy (3)
  • AI Supercomputing Architecture Layout Strategy (4)
  • Intelligent Driving AI
  • Chassis AI
  • 4.3 BMW
  • Physical Architecture (1)
  • Physical Architecture (2)
  • Physical Architecture (6)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • Cockpit AI (1)
  • Cockpit AI (2)
  • Cockpit AI (5)
  • Power AI
  • 4.4 Daimler
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AIOS (1)
  • AIOS (2)
  • Cockpit AI
  • Intelligent Driving AI (1)
  • Intelligent Driving AI (2)
  • Chassis AI (1)
  • Chassis AI (2)
  • 4.5 Toyota
  • AI Supercomputing Architecture Layout
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture Layout Strategy (3)
  • Intelligent Driving AI (1-4)
  • Cockpit AI (1-3)
  • Chassis AI
  • 4.6 Honda
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AIOS
  • Intelligent Driving AI (1)
  • Intelligent Driving AI (2)
  • Cockpit AI (1-3)
  • 4.7 SAIC-GM
  • AI Supercomputing Architecture (1)
  • AI Supercomputing Architecture (2)
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • Cockpit AI (1)
  • Cockpit AI (2)
  • Cockpit AI (5)
  • Intelligent Driving AI
  • Chassis AI (1)
  • Chassis AI (2)
  • Power AI
  • 4.8 Nissan
  • AI Supercomputing Architecture Layout Strategy (1)
  • AI Supercomputing Architecture Layout Strategy (2)
  • AI Supercomputing Architecture Layout Strategy (3)
  • Intelligent Driving AI

5 Software and Hardware Solutions of Tier1 Suppliers

  • 5.1 Neusoft Reach
  • 5.2 SenseAuto
  • 5.3 QCraft
  • 5.4 DeepRoute.ai
  • 5.5 Banma Intelligence
  • 5.6 Thundersoft
  • 5.7 Momenta
  • 5.8 AFARI
  • 5.9 ECARX
  • Full-Stack Technology Product Matrix
  • Multi-Domain Integration Products
  • Zenith Central Computing Platform
  • Antora(R) 1000 Plus Central Computing Platform
  • Antora(R)1000 Pro Cockpit-Parking Integration Computing Platform
  • Antora(R)1000 Central Computing Platform (1)
  • Antora(R)1000 Central Computing Platform (2)
  • Automotive Cerebrum (ECARX Super Brain) Central Computing Platform (1)
  • Automotive Cerebrum (ECARX Super Brain) Central Computing Platform (2)
  • Automotive Cerebrum (ECARX Super Brain) Central Computing Platform (3)
  • Cross-Domain Software Solution (1)
  • Cross-Domain Software Solution (2)
  • 5.10 Autolink
  • Cockpit-Driving Integration Platform Planning
  • Cockpit-Driving Integration Platform (1)
  • Cockpit-Driving Integration Platform (2)
  • Cockpit-Driving Integration Platform (5)
  • AI Computing Platform (1)
  • AI Computing Platform (2)
  • AI Supercomputing Architecture (1-4)
  • 5.11 Hangsheng Electronics
  • Cockpit-Driving Integration Platform (1)
  • Cockpit-Driving Integration Platform (2)
  • Cockpit-Driving Integration Platform (3)
  • Apollo Computing Unit (AI)
  • Full-Stack Hardware and Software Solution
  • 5.12 BICV
  • Cockpit-Driving Integration Platform Planning
  • AI Box
  • Cockpit-Driving Integration Platform
  • 5.13 Lenovo Vehicle Computing
  • AI Box
  • Cockpit-Driving Integration Platform
  • 5.14 Zhuoyu Technology
  • 5.15 Modelbest Technology
  • 5.16 Rivotek