当英伟达从芯片企业向AI基础设施工厂完成深度蜕变，深耕推理市场的国产算力产业链，正站在发展的关键十字 路口。 " 推理拐点已至，中国算力产业链的机会在哪？ " 作者丨刘伊伦 编辑丨包永刚 2026英伟达GTC大会的行业热度持续发酵，黄仁勋擘画的万亿营收蓝图、性能跃升的Vera Rubin平台，加之软 硬件生态的全方位升级，推动整个AI算力产业迎来技术与产业的双重变革浪潮。 3月19日11:30，雷峰网推出GTC专题圆桌对话，特邀三位半导体与AI算力领域的重磅嘉宾，深度拆解2026 GTC 核心技术突破，直击国产算力面临的挑战与机遇。 直播时间： 03月19日 11:30（北京时间） 主持人:胡敏 雷峰网云计算主编 嘉宾: 徐凌杰 魔形智能CEO 嘉宾: 罗彤 芯动科技首席技术官 嘉宾:刘昊飞 嘉宾阵容 徐凌杰 魔形智能CEO 盛景嘉成创投创始合伙人 微信扫码预约直播 01 核心议题 本次圆桌对话紧扣GTC 2026最新动态，聚焦芯片、机架、模型与应用三大核心层面，拆解英伟达技术布局背后 的战略意图，探讨国产算力的应对之策与发展机遇： 02 1. GTC大会核心洞察：刷屏全网的GTC 2026，最强烈的产业感受 ...

Core Insights - Nvidia and Synopsys are expanding their strategic partnership to drive innovation across various industries, addressing challenges such as increased workflow complexity, rising development costs, and pressure on time-to-market [1][2] Group 1: Partnership Details - Nvidia is investing $2 billion to acquire 2.6% of Synopsys' common stock at a price of $414.79 per share [1] - The collaboration aims to leverage Nvidia's strengths in AI and accelerated computing alongside Synopsys' leadership in engineering solutions [1][2] Group 2: Key Initiatives - The partnership will accelerate Synopsys' product applications by utilizing Nvidia's CUDA-X libraries and AI-Physics technology, enhancing various computationally intensive applications such as chip design and physical verification [3] - Both companies will integrate Synopsys' AgentEngineer technology with Nvidia's Agentic AI technology stack to enable autonomous design capabilities in EDA and simulation workflows [3] - The collaboration will promote the use of digital twin technology to connect the physical and digital worlds, facilitating innovation in sectors like semiconductors, robotics, automotive, energy, industrial, and healthcare [3] - Joint market expansion efforts will be undertaken to promote local and cloud-based solutions, leveraging Synopsys' extensive sales force and existing customer base [3] Group 3: Non-Exclusive Agreement - The partnership is not exclusive, allowing both Nvidia and Synopsys to continue collaborating with a broader ecosystem in the semiconductor and electronic design automation (EDA) sectors [4]

Core Insights - Nvidia announced a strategic partnership with Synopsys, investing $2 billion to acquire a 2.6% stake, marking a significant move into the industrial design and engineering sectors [1][3][4] - The collaboration aims to integrate Nvidia's AI computing technology into the electronic design automation (EDA) field, transforming the design process from chips to systems [1][3] - Nvidia's CEO Jensen Huang emphasized the partnership's potential to tap into a trillion-dollar industrial market, far exceeding the consumer AI application space [3][7] Investment Details - Nvidia purchased approximately 4.8 million shares of Synopsys at $414.79 per share, reflecting a 0.8% discount from the previous closing price [4] - The investment positions Nvidia as Synopsys' seventh-largest shareholder, enhancing its influence in the semiconductor design software market [3][4] Strategic Collaboration - The partnership encompasses a multi-year collaboration that includes using Nvidia's CUDA-X libraries and AI physics technology to optimize Synopsys' computational applications [6] - Both companies will work on integrating Synopsys' AgentEngineer technology with Nvidia's AI technology stack to enhance EDA and simulation workflows [6] - The collaboration aims to develop advanced digital twin solutions for various industries, leveraging Nvidia's Omniverse and Cosmos technologies [6] Market Impact - Following the announcement, Synopsys' stock surged over 6.9% in early trading, while Nvidia's stock initially dipped but later rose by nearly 1.9% [1] - Huang highlighted the significant market opportunity in industrial applications, noting that spending on engineering software tools could be several billion dollars, with prototyping costs potentially 10 to 20 times higher [7] Industry Transformation - Huang described the partnership as a revolutionary shift in the design and engineering industry, moving from traditional CPU-based systems to GPU-accelerated solutions [8] - The collaboration is seen as a non-exclusive arrangement, allowing other chip manufacturers to benefit, distinguishing it from Nvidia's previous investments in AI that raised concerns about market monopolization [9][10]

Core Viewpoint - The software service index and related ETFs have shown positive performance, indicating a growing interest and investment in the software and AI sectors, particularly in the domestic market [1][4][5]. Group 1: Market Performance - As of September 17, 2025, the software service index increased by 0.95%, with notable gains from individual stocks such as Guidewire (up 11.13%) and Wealth Trend (up 5.65%) [1]. - The software ETF (159852) rose by 0.85%, with a cumulative increase of 5.58% over the past week as of September 16, 2025 [1]. Group 2: Liquidity and Scale - The software ETF had a turnover rate of 6.74% during the trading session, with a transaction volume of 311 million yuan [4]. - Over the past month, the ETF's average daily transaction volume was 579 million yuan, ranking first among comparable funds [4]. - The ETF's scale grew by 1.649 billion yuan over the past three months, also leading among comparable funds [4]. Group 3: Investment Opportunities - The top ten weighted stocks in the software service index account for 62.05% of the index, with notable companies including iFlytek, Tonghuashun, and Kingsoft [5]. - The report from Huatai Securities suggests a return to value and growth in the market, with a focus on AI as a key area for investment [5]. - The domestic AI computing power demand is expected to grow due to the development of local models and applications, presenting new opportunities for domestic computing businesses [5]. Group 4: Industry Developments - CrowdStrike announced the integration of NVIDIA's Nemotron model into its Charlotte AI to enhance threat detection and response capabilities [4]. - The first "AIDC Industry Development Conference" was held in Shanghai, focusing on establishing standardized guidelines for AI data center construction [4].