2026年被视为AI商业化成败的关键节点，高盛提出十大行业主题：从生成式AI与代理式搜索的崛起，到广告、电商与社 交电商的深度融合；从高位资本开支与云回报考验，到自动驾驶出行加速落地；再到空间计算、互动娱乐、健康科技与企 业在增长和投入间的再平衡，共同构成新一年的产业主线。 硬·AI 作者 | 张雅琦 编辑 | 硬 AI 在经历了多年的基础设施建设后，2026年将成为AI技术验证的关键年份，市场焦点将从单纯的资本支出转 向应用层的实际效用与商业化回报。 高盛全球投资研究团队Eric Sheridan等分析师于10日发布了2026年互联网行业展望报告。报告提出，随 着通用大语言模型（LLM）代理和聊天机器人使用率在2025年的提升，2026年或将成为消费者计算领域 的转折点。 市场关注的焦点将从单纯的付费订阅模式，转向包括广告、电商及代理功能在内的多元化变现途径。高盛 强调，AI技术必须在基础设施、模型平台及应用层持续交付动力并实现资本回报，以避免落入Gartner技 术成熟度曲线中的"幻灭低谷期"（Trough of Disillusionment）。 资本开支方面，尽管市场对"过度建设"存有疑虑，但高盛预计 ...

在经历了多年的基础设施建设后，2026年将成为AI技术验证的关键年份，市场焦点将从单纯的资本支 出转向应用层的实际效用与商业化回报。 据追风交易台消息，高盛全球投资研究团队Eric Sheridan等分析师于10日发布了2026年互联网行业展望 报告。报告提出，随着通用大语言模型（LLM）代理和聊天机器人使用率在2025年的提升，2026年或 将成为消费者计算领域的转折点。 市场关注的焦点将从单纯的付费订阅模式，转向包括广告、电商及代理功能在内的多元化变现途径。高 盛强调，AI技术必须在基础设施、模型平台及应用层持续交付动力并实现资本回报，以避免落入 Gartner技术成熟度曲线中的"幻灭低谷期"（Trough of Disillusionment）。 资本开支方面，尽管市场对"过度建设"存有疑虑，但高盛预计AI相关的资本投入在2026年仍将保持高 位。报告数据显示，过去一年市场对亚马逊、谷歌和META在2025及2026财年的资本开支（Capex）预 期分别上调了约46%和80%，两年合计增加约2500亿美元。这些在AI开发上的巨额投入能否转化为可见 的利润回报，将是未来12至18个月左右投资者情绪和公司业 ...

Core Insights - The article discusses the increasing involvement of billion-dollar quantitative private equity firms in the artificial intelligence (AI) sector, highlighting the establishment of Shanghai AllMind Technology Co., Ltd. by Nian Kong Technology to explore cutting-edge AI topics [1][2] - Several major private equity firms, including Jiukun, Kuande, Mingshi, and Heiyi, are actively investing in AI, driven by the success of DeepSeek and the potential to enhance their competitive edge and expand business boundaries [1][8] Company Developments - Nian Kong Technology founded AllMind on May 19, 2025, focusing on foundational research and engineering technologies related to general large language models (LLMs) [2][3] - The company aims to address current challenges in large models, such as improving logical reasoning capabilities and reducing hallucination issues, while also exploring applications in various fields like new materials and pharmaceutical research [3][5] Academic Collaborations - AllMind collaborated with Shanghai Jiao Tong University to write a paper titled "Step-wise Adaptive Integration of Supervised Fine-tuning and Reinforcement Learning for Task-Specific LLMs," submitted to NIPS on May 20, 2025 [2][3] - The paper introduces a new training framework that alternates between supervised fine-tuning (SFT) and reinforcement learning (RL), demonstrating superior performance over traditional methods [6][5] Industry Trends - Kuande Investment has established an AI talent recruitment initiative, focusing on developing general-purpose AI technologies, while also launching an independent AI research lab [8] - Jiukun Investment has previously published research on AI, successfully replicating DeepSeek-R1's work and identifying issues related to language mixing that affect reasoning capabilities [8] - Mingshi Fund has integrated AI into its entire investment research process, enhancing efficiency in factor mining, model optimization, and risk control since establishing its AI lab in 2021 [8][9] - Heiyi Asset has implemented AI throughout its investment research process, utilizing large language models to analyze complex data and improve prediction accuracy [9] Future Outlook - The quantitative private equity sector is expected to see increased investment in AI, including talent acquisition and infrastructure development, leading to a competitive landscape for AI professionals [9]