80s
AI智能体能自主生成并运行量子计算代码_我的网站
A | 科技日报讯 （记者张佳欣）氛围编程（Vibe Coding）正从传统软件开发进入量子计算领域。
B |
With just one day to go before Chinese humanoid robotics pioneer Unitree Robotics opens subscriptions for its initial public offering (IPO) on Shanghai's STAR Market, investor attention has extended far beyond the company itself, highlighting a broader capital-market rush into embodied intelligence as the sector moves from technological breakthroughs toward commercialization.
As of press time, discussions surrounding the IPO have pushed the topic to the top of trending lists across multiple stock trading platforms. The debate has focused on Unitree Robotics' disclosed short-term financial performance, as well as the potential opportunities for investors from China's policy push to include embodied intelligence and humanoid robots among key areas of medium- and long-term technology and industrial development.
Experts said the wave of IPOs among robotics companies, including Unitree Robotics, represents a shift that allows robot makers to engage more directly with public markets and investors. It also brings companies at different stages of development and pursuing different technological paths into a common valuation framework, promoting market comparisons and competition.
At the same time, the capital-market boom is likely to accelerate a shakeout, separating companies capable of building sustainable business models from those that struggle to turn technological advances into commercial returns.
On Thursday, Unitree Robotics released its announcement on the IPO of shares and listing on the STAR Market. According to the announcement, the offering price was set at 150.80 yuan per share, valuing the company at approximately 61 billion yuan ($9.04 billion) upon listing.
Secondary-market professionals told the Global Times that demand for Unitree Robotics' IPO subscription could be extremely high, meaning the chances of individual investors securing shares are likely to be very limited.
Robotics is currently one of the few categories that can translate AI capabilities into tangible assets with relatively comparable valuation metrics, creating a unique advantage in valuation narratives. This scarcity is one of the key factors drawing the attention of both large and small investors, Tian Feng, former dean of SenseTime's Intelligence Industry Research Institute, told the Global Times on Sunday.
Unlike large language model companies, whose capabilities are often assessed through benchmarks and demonstrations that are difficult for ordinary secondary-market investors to directly evaluate, robots offer visible and verifiable evidence through tasks such as grasping, walking and dexterous manipulation, Tian said. This tangible demonstration of "embodied" intelligence naturally creates a valuation premium in market narratives, he added.
Tian cautioned that extreme subscription figures, such as RobotHoenix's 14,855-fold oversubscription on its listing and LDRobot's 127.62 percent first-day share price surge, were driven by a combination of institutional advantages and sector scarcity rather than serving as evidence of the commercial validation of robotics products. He noted that there is no direct correlation between oversubscription multiples and the certainty of product deliveries or commercialization prospects.
Beyond Unitree Robotics, a broader financing boom in China's robotics industry has also gained significant momentum, with the Hong Kong Stock Exchange emerging as the main battleground for this latest IPO wave.
Citing industry data provider Wind, the China STAR Market reported that five robotics-related companies have gone public in Hong Kong so far in 2026. Meanwhile, 51 companies with businesses spanning industrial robots, service robots and embodied intelligence solutions are currently in the IPO pipeline in Hong Kong, accounting for 12.7 percent of all companies awaiting listing.
According to data from IT industry research platform ITjuzi.com, financing in China's embodied intelligence sector totaled 93.5 billion yuan in the first half of 2026, surpassing 90 billion yuan and marking a fivefold increase from the same period in 2025. The number of financing deals reached 322, up 137 percent year-on-year.
Tian said that while policy support and clearer commercialization pathways are boosting confidence in the sector, the latest IPO wave should also be viewed through a more cautious lens. The surge comes partly as the primary market's ability to continue funding embodied intelligence companies faces limits, prompting firms to seek new liquidity channels through public markets, rather than solely reflecting a dramatic leap in industrial maturity.
As emerging technology sectors increasingly become the focus of public attention, more objective and professional analysis is also needed to help investors assess companies' technological capabilities, commercialization prospects and long-term value, rather than simply following market enthusiasm, Chinese experts noted.
。据英国《自然》杂志日前消息，法国量子计算初创企业Pasqal研究人员开发出一种人工智能（AI）智能体，能根据英文自然语言指令生成量子计算代码，并自动在真实量子计算机上运行。研究人员表示，这种量子氛围编程有望降低量子计算实验的技术门槛，让更多人能够开展量子计算研究。 所谓氛围编程，是指用户只需要用自然语言告诉AI工具，希望软件实现什么功能，AI便会生成相应程序，并根据用户反馈进行修改和运行。
C | 目前，Anthropic公司的Claude等前沿大语言模型已展现出理解量子计算的能力，不少研究人员也开始用这类模型辅助编写量子计算代码。
D | Pasqal研究团队则让这些前沿大语言模型熟悉其量子计算机的技术规格，进一步探索大语言模型能否实现量子领域的氛围编程。 团队重点探索了量子计算最具潜力的应用之一，即量子模拟。量子模拟是指调控量子计算机，使其表现得类似于另一种物理系统，例如催化剂或具有特殊磁性的材料。
E | 量子计算机运行这样的模拟，可预测材料的性质，而相关计算可能会超出经典计算机的能力。 Pasqal团队对AI智能体进行了3项测试，每项测试都选取一篇描述某种物理现象的论文，并要求智能体编写和运行代码，以验证这一现象是否可通过量子计算机进行模拟。 Pasqal首席技术官伊克·亨里耶说，要完成这一“转换”，通常需要同时掌握材料物理学和量子计算知识的专业团队，而AI智能体现在可协助完成这一过程。 不过，这并不意味着量子计算已能完全交给AI。
F | 在其中一项测试中，智能体需要团队提供较多指导，才能最终得到符合物理规律的实现方案。在涉及实验物理准确性的关键环节，人类研究人员仍不可或缺。 目前，其他研究团队也在探索利用大语言模型辅助量子计算。纽约大学阿布扎比分校应用物理学家努海拉·伊南及其同事开发了一个定制版开源大语言模型LLaMA，使其能生成量子计算代码，但目前还不能自主运行这些代码。
G | 不过，这一工具已帮助她和同事将一些原本需要4天完成的工作缩短至1天。
Current article:http://www.yuejinhuapennuochoudengying.cyou/q0m/20260826/513533.docx
Published on:17:15:25