1. Report Industry Investment Rating No relevant content provided. 2. Core Views of the Report - Generative AI is reshaping industries, but its security issues are becoming a key bottleneck for sustainable development. Future AI security will trend towards security left - shifting, system - and intelligence - based defense, and an open and shared - responsibility ecosystem [142][144] - Volcano Engine positions itself as a trusted and secure infrastructure provider for AI cloud - native, offering safe and compliant AI services and sharing security responsibilities with users [27][46] 3. Summary by Directory 3.1 Introduction - **Industrial Trajectory and Inflection Point**: The capabilities of foundational models are expanding rapidly, and enterprises are shifting from single - point trials to platform - based construction, requiring unified management of model services, data governance, etc. [16][17] - **Core Issues and Challenges in Generative AI Security**: There are risks in the model, data, and application layers, and governance and compliance need to be embedded in products [19][21][23][24] - **Volcano Engine's AI Security Proposition**: It aims to be a trusted and secure infrastructure provider for AI cloud - native, building AI security capabilities in technology, governance, and the ecosystem [27] 3.2 Generative AI Security Risks - **Regulatory and Compliance Risks**: Global regulatory bodies are strengthening laws and regulations for AI. Enterprises need to comply with relevant requirements in different regions [31][32][33] - **Data Privacy Risks**: There are risks in data collection, storage, training, and usage stages, and internal human factors can also cause risks [36][37][38] - **Generative AI Security Risks**: Risks exist in AI infrastructure, models, platforms, and intelligent agents, and along the "AI infrastructure → large model → intelligent agent" chain [40][41][42] 3.3 Volcano Engine's Generative AI Service Security Assurance System - **Security Responsibilities in the Generative AI Wave**: Security responsibilities in generative AI scenarios are shared between users and service providers, including compliance, privacy, and security responsibilities [46] - **Compliance Qualifications and Certifications**: Volcano Engine's large models have completed relevant filings and evaluations, and it participates in standard - setting to promote industry security [61][62] - **Data Security and Privacy Protection Design Concept**: The key challenges in large - model data and privacy security are addressed. The Ark TrustAI System provides a comprehensive protection plan [65][67][72] - **Generative AI Security Technology Assurance System** - **AI Infrastructure Security**: It combines platform - based and enhanced security solutions, covering governance, product protection, threat intelligence, and more [76][80][84] - **AI Model and Platform Security**: Volcano Ark ensures model and user information security. Model security has principles and lifecycle management, and the platform has a secure architecture [92][93][103] - **AI Intelligent Agent Security**: It includes identity and permission management, tool management and access control, and in - depth defense and reinforcement [114][120][124] 3.4 Summary - **Generative AI Industry Security Outlook**: Future AI security will trend towards security left - shifting, system - and intelligence - based defense, and an open and shared - responsibility ecosystem [142][144] - **Volcano Engine's Commitment to Generative AI Security**: Volcano Engine is committed to providing a trusted, controllable, and compliant AI cloud - native base and collaborating with partners to address security challenges [142]

wo[Table_Title] Research Report 21 Dec 2025 中国电子 China (Overseas) Technology 火山引擎 FORCE 大会追踪（2）： Agent 规模化落地，方舟与企业底座升 级 Volcengine FORCE(2): Scaling Agents to Production: Upgrades to Volcano Ark and Enterprise Foundations 姚书桥 Barney Yao 吕小潼 Xiaotong Lyu barney.sq.yao@htisec.com xt.lyu@htisec.com [Table_yemei1] Flash Analysis [Table_summary] (Please see APPENDIX 1 for English summary) 风险: 事件 2025 年 12 月 18 日，火山引擎在 2025 冬季 FORCE 原动力大会上发布了面向多模态 Agent 场景优化的豆包大模型在 2025 FORCE 原动力大会开发者日（上海）上，火山引擎围绕智能体（Agent）规模化生产与 ...

Core Insights - The article highlights the launch of Doubao Model 1.8 by Volcano Engine, which features enhanced multimodal agent capabilities and a 256K ultra-long context for handling complex tasks [2][3][5] - Volcano Engine's "AI Savings Plan" aims to optimize user costs, offering savings of up to 47% on AI usage [3][17] - The company emphasizes the importance of expanding the AI market rather than competing for existing market share, predicting a potential market growth of tenfold in the coming year [4] Model Capabilities Upgrade - Doubao Model 1.8 shows significant improvements in multimodal understanding, particularly in long video comprehension and security monitoring scenarios [5] - The model's context management allows companies to tackle complex tasks and support decision-making processes [5] - New image generation model Doubao-Seedream-4.5 offers capabilities such as multi-image combinations, creative photography, and virtual try-ons [5] Video Generation Enhancements - The Seedance series includes two versions: Seedance-1.0-Lite focuses on cost and speed, while Seedance-1.0-Pro delivers cinematic quality and native sound effects [7] Application Scenarios - Doubao Model has been integrated into smart hardware and voice assistants, covering daily communication, professional services, and online searches [9] Ecosystem Development - Volcano Engine introduced "Volcano Ark" inference outsourcing service, supporting major open-source models for seamless deployment [11] - The Viking series products enhance user input quality and facilitate the rapid construction of knowledge and memory bases for models and agents [13] - The company launched an enterprise-level AI Agent platform, AgentKit, which has been adopted by leading clients [15] Cost Optimization Plan - The "AI Savings Plan" allows users to join once and benefit from cost reductions across various models, with flexible payment options [17] - The initiative is expected to enhance performance and reduce costs, particularly for video generation models, and is seen as a potential investment opportunity in the AI application landscape [17]

Core Insights - The article discusses the launch of the Seedance 1.5 Pro model, highlighting its advanced capabilities in video and audio synchronization, particularly in Chinese and dialect outputs, and emotional expressiveness [3][12][36]. Group 1: Video and Audio Synchronization - Seedance 1.5 Pro achieves film-level audio-visual synchronization, allowing for accurate lip-syncing and multi-scene synchronization, significantly reducing production time [13][16][18]. - The model can generate up to 12 seconds of video, enabling the creation of short advertisements with precise dialogue and sound effects [18][19]. Group 2: Language and Dialect Capabilities - The model excels in multilingual outputs, including English, Japanese, Korean, and Spanish, but stands out for its proficiency in Chinese dialects, particularly Cantonese [21][23]. - Seedance 1.5 Pro can seamlessly switch between various Chinese dialects, allowing for realistic interactions between characters from different regions [25][26]. Group 3: Emotional Expressiveness - The model has significantly improved its emotional expressiveness, allowing for varied performances based on the same line of dialogue, enhancing the overall storytelling experience [27][30]. - It can integrate sound effects, music, and visual elements to create immersive video content, streamlining the production process [33][34]. Group 4: Future Developments - An anticipated feature is the draft sample capability, which allows users to preview lower-resolution drafts before finalizing high-resolution outputs, optimizing both time and cost [35]. - The advancements in Seedance 1.5 Pro represent a significant leap in AI video production, merging sound and visuals to create high-quality content suitable for professional use [37][38].

Core Insights - The AI market driven by large models is accelerating with a new metric, Token consumption, becoming a "real benchmark" for AI application deployment [1] - The IDC report reveals a staggering growth projection, with the volume of large model calls on public cloud in China expected to reach 536.7 trillion Tokens in the first half of 2025, a nearly 400% increase from 114 trillion Tokens in 2024 [1] - The market landscape is becoming clearer, with Volcano Engine holding a 49.2% market share, expanding its lead from 46.4% in 2024 [1] Market Dynamics - Volcano Engine leads the Chinese large model public cloud service market with a 49.2% share, followed by Alibaba Cloud at 27.0% and Baidu Smart Cloud at 17.0% [2] - A different report by Omdia shows Alibaba Cloud leading with a 35.8% share when considering the entire cloud service chain, indicating a shift from infrastructure competition to deepening model applications [2] Token Consumption as a Metric - The choice of "Token call volume" as a core statistic reflects a rethinking of evaluation standards in the AI industry, focusing on actual model usage rather than just computational supply [3] - Token consumption is closely tied to application deployment, showcasing a more sustainable and exponentially growing model for the AI industry [4] Growth Catalysts - Two key technological breakthroughs have significantly impacted market growth: the first in July 2024, when the YoY growth rate for large model public cloud services exceeded 160% following cost reductions from the Doubao model [5][6] - The second breakthrough occurred in February 2025, marked by the popularity of the DeepSeek-R1 inference model, indicating a shift from model training to inference services [6] Volcano Engine's Competitive Edge - Volcano Engine's rapid growth in the large model business is attributed to its strategic, technological, and scale advantages [7] - The Doubao model family has a leading iteration speed in the industry, covering multiple modalities including text, image, audio, and video [8] - The performance of Volcano Engine's MaaS platform, "Volcano Ark," has been significantly enhanced, with output rates for the DeepSeek-R1 model being 2.6 times that of some competitors [9] Industry Penetration - The AI cloud service market is expanding from the internet sector into traditional industries, with Volcano Engine serving major clients across various sectors, including automotive and finance [10] - The market is expected to have hundreds of times growth potential, with multi-modal models and Agent applications driving future growth [11] Future Trends - Volcano Engine is continuously upgrading its products and services, recently launching several new models and a "smart model routing" service to balance performance and cost [11] - The daily Token consumption has surpassed 30 trillion, reflecting a growth of over 80% since May 2025 [11] - The competition in the "Tokens economy" will favor those who provide the best performance at the lowest cost, shaping a more mature ecosystem in the AI cloud market [12]

Core Insights - The AI cloud market in China is experiencing significant growth, with a notable increase in advertising from major cloud service providers at airports in cities like Beijing, Shanghai, Shenzhen, and Hangzhou [1][2] Market Growth - According to IDC, the volume of large model calls in China's public cloud is expected to grow nearly 400% by the first half of 2025 [2][19] - Omdia predicts a 148% growth in the AI cloud market in China by 2025 [2] Market Leaders - Alibaba Cloud holds the largest market share in the AI cloud sector at 35.8%, followed by Volcano Engine at 14.8%, Huawei Cloud at 13.1%, and Tencent Cloud at 7% [8] - Volcano Engine has rapidly risen to become a challenger in the market, leveraging its MaaS growth [2][12] Advertising Claims - Alibaba Cloud claims to lead the AI cloud market, stating its share exceeds the combined total of the second to fourth players [4] - Volcano Engine asserts it holds a 46.4% share of the public cloud large model market [4][11] Statistical Discrepancies - The differing claims of market leadership stem from varying statistical methodologies used in reports by Omdia and IDC [5][12] - Omdia's report includes a comprehensive revenue model (IaaS, PaaS, and MaaS), while IDC focuses solely on external customer token usage in the MaaS layer [11][12] Competitive Dynamics - The competitive landscape is influenced by the aggressive pricing strategies initiated by DeepSeek, which has led to a significant price reduction in large model services [17][18] - DeepSeek's pricing adjustments have prompted other major players like ByteDance, Alibaba, Baidu, and Tencent to follow suit, resulting in a price war in the large model market [17][18] Future Projections - IDC forecasts that the market for generative AI software in China will continue to grow rapidly, potentially reaching 48.24 billion RMB by 2028 [21] - Omdia anticipates a compound annual growth rate (CAGR) of 26.8% for the AI cloud market from 2025 to 2030, with the MaaS layer expected to grow at a CAGR exceeding 72% [22]

Core Insights - The AI film international summit held during the 30th Busan International Film Festival sparked discussions on the profound impact of AI on the film industry, likening it to an industrial revolution, while many filmmakers believe the core essence of film creation cannot be replaced [1][2] - The introduction of the InnoAsia unit at the Busan International Film Festival's Asian Content and Film Market aims to explore how AI can reshape film expression [1] - The "Future Image Plan" showcased award-winning short films created using generative AI platforms, highlighting the capabilities of ByteDance's Seedance video generation model [1] Industry Trends - AI technology is penetrating the film industry in a "full-chain" manner, enhancing creators' freedom and efficiency across all production processes [2] - The acceptance and respect for AI in filmmaking have grown, with industry professionals noting that AI is approaching a "singularity" moment in its application [2] - AI is seen as a transformative force in film production, making previously uncontrollable elements like characters and lighting "precise and controllable" [2] Perspectives on AI in Filmmaking - The use of AI is expected to become a mainstream trend in the film industry, with producers increasingly adopting AI to save budgets and reduce production time [3] - Despite the advantages of AI, some voices argue that it cannot replace the unique essence of traditional filmmaking, particularly the irreplaceable qualities of human actors [3] - Future filmmaking may adopt a "dual-track" approach, combining traditional methods with AI to explore previously unimaginable stories [3]

Core Insights - DeepSeek-V3.1 has officially launched on Huoshan Ark, providing enterprise users with enhanced resource stability and speed [1] Group 1 - The new model achieves an ultra-low latency of 20-40ms in token processing time (TPOT) [1] - It offers the highest initial concurrency of 5 million transactions per minute (TPM) across the network [1]

Core Insights - The large model industry is entering the "Agent Year," with significant transformations occurring in the enterprise market, led by Volcano Engine's HiAgent intelligent platform, which has achieved the highest rankings in both contract amounts and numbers [2] - Industry data indicates that by the first half of 2025, the number of projects awarded for intelligent agent construction platforms reached 371, representing a 250% year-on-year increase, nearing two-thirds of the total from the previous year [2] - Volcano Engine has maintained its position as the leader in the number of contracts since the second half of 2024, with a strong growth momentum reflected in its dual championship in contract amounts and numbers in the first half of this year [2] Company Strategy - Volcano Engine's Vice President, Zhang Xin, emphasized that enterprises require definitive solutions that transform industry know-how into productivity rather than random technological attempts [2] - The HiAgent 2.0 platform features four core upgrades: the innovative Agent DevOps system for integrated development and operations, an industry template library that reduces customization cycles by 80%, a unified model application system for fine-tuning exclusive models, and the Canvas interactive portal for centralized management of hundreds of dispersed agents [2] Industry Trends - The business model in the industry is undergoing a fundamental shift, moving from pay-per-use and token-based pricing to a subscription model based on usage, with the ultimate goal being a revenue-sharing model focused on "marketing revenue and cost savings" [3] - Addressing concerns that agents do not enhance cloud vendor stickiness, Zhang Xin argued that the long-term memory and knowledge base of agents create natural barriers, with Volcano Engine building a cost-effective moat through its Doubao large model and Volcano Ark inference optimization [3] - A case was cited where replacing HiAgent allowed for a breakthrough in concurrent capabilities, exceeding the previous platform's limit of 200, thus meeting real production demands [3]

Core Insights - ByteDance's Volcano Engine has launched new AI models, including Doubao Image Editing Model 3.0 and Doubao Simultaneous Interpretation Model 2.0, enhancing its AI capabilities and services [2][7] - The Doubao model's daily token usage has surged to over 16.4 trillion, a 137-fold increase since its initial release, making it the leading public cloud model service in China with a market share of 46.4% [2][7] - The company aims for significant revenue growth, projecting over 12 billion in 2024 and doubling to over 25 billion in 2025, with a long-term goal of reaching 100 billion by 2030 [8] Product and Service Developments - The Doubao Image Editing Model 3.0 allows users to perform complex image editing tasks using natural language, making it suitable for various creative and marketing applications [8] - The Doubao Simultaneous Interpretation Model 2.0 reduces voice delay from 8-10 seconds to 2-3 seconds, enhancing real-time translation capabilities [8] - The Doubao model series has been upgraded to version 1.6, featuring improved performance in visual understanding, code reasoning, and cost efficiency for commercial applications [9] Market Position and Future Outlook - Volcano Engine holds a dominant position in the public cloud model service market, surpassing competitors like Baidu and Alibaba [7][8] - The company is focused on expanding its AI Agent capabilities, transitioning AI from a tool to an active assistant, which is crucial for commercial applications [7][8] - The HiAgent platform aims to enhance workplace efficiency by providing a digital employee interface that streamlines tasks and improves collaboration [14][16] Technological Innovations - The launch of the Responses API enhances multi-modal data management and reduces operational costs by 80%, significantly improving development efficiency [10] - The company is committed to open-sourcing various high-quality projects to foster innovation and collaboration within the AI ecosystem [12][18] - The Doubao team continues to develop multi-modal models, with plans to integrate the simultaneous interpretation model into consumer devices [17]