Core Insights - Morgan Stanley's report highlights China's ambition to become a global leader in artificial intelligence (AI) technology, driven by a robust ecosystem of talent, innovation, data, and infrastructure [1][3] - The report emphasizes the importance of applying AI to the "real economy" and commercializing AI products to enhance productivity in traditional industries [1][2] - China is focusing on market-driven AI applications, particularly in sectors like autonomous driving, smart manufacturing, and digital customer service, contrasting with the U.S. focus on broader consumer applications [1][2] Infrastructure Sector - China's AI GPU self-sufficiency is projected to increase from 34% in 2024 to 82% by 2027, with companies like Huawei and Cambricon leading innovations in chip development [5] - Lenovo's business segments are expected to benefit from the AI revolution, with a 60% year-on-year growth in its ISG segment for Q4 2024 [6] Data Center Sector - The data center industry is anticipated to see a significant increase in new bookings, growing from 2.1 GW in 2024 to 3.7 GW annually from 2025 to 2027, representing a 76% increase [8] - The rental pricing in China's data center sector has stabilized at lower levels, with improving return rates due to lower bank financing costs and faster client onboarding [8] Platform Sector - The rapid development of AI applications is expected to accelerate growth in China's IaaS/PaaS market, benefiting cloud service providers like Alibaba and Tencent [9] - Tencent's cloud business is projected to accelerate growth starting Q2 2025, as it reallocates resources to external cloud clients [9] Application Sector - In the 2C domain, AI applications are rapidly evolving, with platforms like WeChat leveraging user data to enhance user experience and drive profitability [11] - The 2B application speed is expected to surpass previous public cloud cycles, with a subscription model becoming prevalent for enterprise AI applications [11] Automotive and Robotics - The penetration rate of L2+ autonomous driving in China is expected to reach 25% by 2025, benefiting manufacturers like BYD and Geely [12] - By 2030, China's humanoid robot inventory is projected to reach 252,000 units, with significant growth anticipated in both commercial and household humanoid robots by 2050 [12] Energy and Quantum Computing - AI-driven data centers are expected to account for 10% of China's total electricity demand by 2035, with green energy initiatives gaining traction [14] - China's advancements in quantum computing, exemplified by the "Zuchongzhi 3" prototype, are set to provide new computational capabilities that will benefit AI and other industries [14] Conclusion - Despite U.S. chip restrictions, China's AI computing capabilities are advancing, with domestic semiconductor companies innovating rapidly to close the performance gap with U.S. counterparts [15]

在人工智能快速演进的浪潮下，大模型正加速重构各行业的技术底座，而 推理性能优化 正成为应对算力挑战、内存瓶颈与通信压力的关键突破口。 当前，大模型推理性能优化主要围绕 模型优化、推理加速与工程优化 在即将于 5 月 23 日 -24 日举办的 AICon 全球人工智能开发与应用大会·上海站 中，我们特别策划了《大模型推理性能优化策略》专题论坛，由阿里云公共云大模型技术服务负责人 王德山 担任专题出品人，现已确认多位业内实践者参与分享。以下为嘉宾阵容及即将带来的精彩议题简介～ 向乾彪 – 腾讯推理架构师 向乾彪在 GPU 推理加速拥有丰富经验。他的技术专长覆盖高性能异构计算及深度性能优化，并在实 践中不断突破前沿技术瓶颈。目前，向乾彪带领团队负责混元大语言模型的推理加速框架 【AngelHCF】 三大方向展开：通过模型量化、剪枝与蒸馏等手段降低计算复杂度、提升推理效率，例如 DeepSeek-R1-Distill-Qwen-32B 采用蒸馏策略，在保持高性能的同时显著压缩资源开销；依托 SGLang、vLLM 等高效推理引擎提升生成速度与系统吞吐能力；同时结合实际业务场景，合理规划 并发策略、优化 GPU 配置 ...