思朗科技十年磨一剑，用 MaPU 与 " 天穹 " 超算重构全球科学智能科研新范式 人工智能、工业智能已迈入深耕落地的成熟期，科学智能（AI for Science，AI4S）则悄然崛起为科技突破的 " 隐形引擎 "。 AI 赋能基础科研，打破 "猜假设、做实验" 的传统模式，为微观探索、新药研发等前沿领域开辟全新方向 。 2025年11月，特朗普签署行政令启动"创世纪计划"，以"曼哈顿计划2.0"的格局推动AI与科研深度融合，通过整合资源重构 科研全流程，让AI从辅助工具升级为科学发现核心驱动力。这一理念的源头，可追溯至2018年科研界的反思。彼时AI在科 研中仅零散作为工具，难以突破"被动辅助"瓶颈，无法触及核心规律探索。基于此，科研界萌生让AI深度参与科学发现的 构想，" AI for Science "概念正式诞生，中文定名为" 科学智能 "。此后，全球范围内包括加州理工、谷歌 DeepMind 等国 际顶尖科研机构，以及我国多所高校和阿里、华为等科技企业纷纷布局，相继成立相关实验室，科学智能的内涵也不断升 级，从最初提升科研效率，逐步拓展到参与科研全流程。2024 年诺贝尔物理、化学奖中 AI 元素 ...

Core Insights - Shanghai Silang Technology Co., Ltd. (Silang Technology) has filed for IPO guidance with the Shanghai Securities Regulatory Bureau, with Guotai Junan Securities as the advisory institution [1] - The company, founded in 2016, focuses on the design and application of high-performance domestic computing chips, specifically for AI in scientific research, and holds complete intellectual property rights for its trillion operations per second algebra processing unit (MaPU) [1] - The MaPU architecture combines the efficiency of ASICs with the programmability of CPUs/GPUs, achieving performance that can exceed traditional architectures by several times in specific scientific computing tasks [1] Company Developments - Based on the MaPU architecture, Silang Technology has developed a series of communication baseband products and the "Tianqiong" 3D scientific computer, which is the first fully domestically developed 3D scientific computer in China [2] - The "Tianqiong" computer offers significant advantages in low communication latency and computational efficiency, providing 2 to 4 orders of magnitude acceleration compared to traditional 2D supercomputers [2] - The company has established a commercial supercomputing center in Xiaogan, Hubei Province, which is the first operational center for the "Tianqiong" in China, launched in 2023 [3] Industry Context - The distinction between "supercomputing" and "intelligent computing" is highlighted, with supercomputing focusing on large-scale calculations for scientific research, while intelligent computing is tailored for AI training and inference [2][3] - The integration of AI with physical, chemical, and biological research is emphasized, showcasing the need for high-performance chips in supercomputing to meet the demands of extensive computational requirements [3] - Silang Technology's MaPU architecture has also made strides in the communication sector, with its UCP baseband chips being adopted in 5G small base stations by major telecom operators and participating in low Earth orbit satellite communication projects [4]