Group 1 - OpenAI is in informal talks with top Wall Street investment banks to prepare for its planned IPO in Q4 2026, having hired a senior financial team including Chief Accounting Officer Ajmere Dale [1] - The IPO preparation comes amid intense competition in the generative AI industry, with competitors like Anthropic reportedly raising over $10 billion and expected to initiate their own IPO process soon [3] - OpenAI aims to raise funds for R&D and strengthen its brand influence through the public market, solidifying its position as a leader in AI technology [3] Group 2 - OpenAI's unique governance structure poses potential obstacles for the IPO, as its complex equity framework, including special rights for strategic investors like Microsoft, needs restructuring to meet regulatory compliance [3] - Increasing legal scrutiny globally regarding copyright and data privacy issues related to AI-generated content may raise compliance costs and litigation risks [3] - CEO Sam Altman expressed mixed feelings about leading a public company, acknowledging that while an IPO could provide a more stable capital source and attract top talent, it may also impose short-term profit pressures that could divert from long-term research goals [4]

Core Insights - Yushu Technology announced the launch of the open-source model UnifoLM-VLA-0, which is designed for general humanoid robot operations [1][3] Group 1: Model Overview - UnifoLM-VLA-0 is a visual-language-action (VLA) large model under the UnifoLM series, aimed at overcoming the limitations of traditional visual-language models (VLM) in physical interactions [3] - The model evolves from general "text-image understanding" to a "embodied brain" with physical common sense through continued pre-training on robot operation data [3] Group 2: Performance and Capabilities - The model demonstrates significantly enhanced spatial reasoning abilities and reliable multimodal perception across various task scenarios [3] - It integrates text instructions with 2D/3D spatial details through continued pre-training, improving its spatial perception capabilities, particularly for operation tasks that require high instruction understanding and spatial awareness [3] - The model has been validated on real machines, achieving high-quality completion of 12 complex operational tasks with a single strategy [3]

Group 1 - The core point of the article is that Nvidia, Microsoft, and Amazon are negotiating to invest up to $60 billion in OpenAI, which would be the largest single funding round in the AI sector to date [1][4] Group 2 - Nvidia plans to lead the investment with $30 billion, as it is a key chip supplier for OpenAI, providing GPUs for training models like ChatGPT [4] - Microsoft, as OpenAI's largest strategic partner and cloud service provider, intends to invest less than $10 billion, continuing its support since 2019 while avoiding excessive dilution of its equity [4] - Amazon is looking to invest over $10 billion, potentially exceeding $20 billion, marking its first direct investment in OpenAI, which is seen as a strategic move to strengthen its AI ecosystem against Microsoft Azure [4] Group 3 - The funding round is perceived as a response to competitive pressures, especially after Anthropic raised $7.3 billion and other tech giants like Google and Meta accelerated their AI investments [4] - OpenAI's CEO Sam Altman has indicated that future funding needs could reach trillions of dollars to achieve the goal of Artificial General Intelligence (AGI), making this financing round just the first step in a long-term plan [4]

来源：科技日报 本报记者 张佳欣 这种生物化学反应为"沉没成本"提供了坚实的神经基础：因为你付出的足够多，所以获得回报的那一 刻，大脑给你的愉悦感冲击也更加猛烈。这种爆发式的快感会反过来重塑你的认知，认为它更值钱了。 但实际上，只是大脑在为你努力的汗水买单。 为了抢到一张心仪歌手的演唱会门票，你提前定好闹钟，在几个平台间疯狂切换，甚至不惜加价购 买"溢价票"，抢到票以后你会觉得无比喜悦；为了排到那家网红餐厅，你在寒风中瑟瑟发抖两小时，进 店后即便菜品味道一般，你也会自我安慰"来都来了，其实还行"；在一段早已千疮百孔的感情里，你一 边抱怨对方的冷漠，一边却因为"在一起这么多年"而迟迟不肯放手…… 这些生活场景背后的逻辑其实很古怪。我们对一件事物的估值，往往不取决于它本身的价值，而取决于 我们为其付出的代价。在经济学和心理学中，这被解释为"沉没成本"。即便客观收益已经微乎其微，只 要你投入了金钱、时间，甚至情绪，大脑就会像产生执念一样，还想继续投入。 为什么进化没有让我们变得更理智？近期，美国斯坦福大学医学中心科学家的最新研究，揭开了这背后 的神经科学奥秘。 在小鼠实验中，科学家设置了一系列关卡。小鼠需要通过反复 ...

来源：东方网 在数字化转型纵深推进的今天，数据已成为企业最宝贵的资产。然而，面对不断攀升的存储成本、日益 复杂的安全威胁以及运维管理的重重挑战，众多中小企业和开发者在"要不要上云""怎么安全上云"之间 踟蹰不前。在近期举办的"华为云中国区销售伙伴产品方案发布会第一期"中，华为云Flexus OBS凭借成 本最优、极致安全可靠、管理简单三大核心优势，成为中小企业数据上云首选。 成本最优：资源包灵活叠加，越用越省 传统云存储往往存在价格僵化、套餐互斥、用量越大成本压力越重等问题，让中小企业望"云"却步。 Flexus OBS直击这一痛点，通过创新的计费机制，真正实现"花得值、用得稳"。 Flexus OBS提供从1TB到200TB等多种规格的优惠资源包，支持自由组合、灵活叠加，且可直接抵扣标 准套餐包，综合成本降低20%。相比市场常见方案中"限购一个""资源包互斥"等限制，Flexus OBS显著 提升资源使用效率。配合阶梯价格策略，越用越省。无论是日常业务数据归档、网站静态资源托管，还 是突发流量下的内容分发需求，Flexus OBS都能让企业"用多少、付多少"，杜绝资源浪费，轻松实现降 本增效。 在存储成本持 ...

来源：人民日报客户端 0:00 ...

来源：科技日报 在铁磁体中，磁针均整齐指向同一方向，易于通过外磁场调控，因此常规磁性设备如机械硬盘等均以铁 磁材料作为存储单元。而反铁磁材料由于磁针方向相反、磁性抵消而很难被磁场调控，多被当作辅助铁 磁体的"配角"。 沈涵 科技日报记者 王春 29日，记者从复旦大学获悉，该校物理学研究团队利用自主开发的多模态磁光显微技术，发现一类特殊 的低维反铁磁体系能够在外磁场下像铁磁体一样展现出确定性的双稳态整体切换，并完善了经典的磁学 理论框架，用来描述该现象背后的物理机制。该研究揭示了低维层间反铁磁体磁化翻转的关键因素与独 特效应，推动反铁磁材料研究迈出从"有趣而无用"到"可读可写"的关键一步，为开发新一代低功耗、高 速运算芯片提供了新路径。相关成果1月29日在国际学术期刊《自然》上在线发表。 事实上，相比铁磁体，反铁磁材料更有助于开发更高密度、更快运行速度的磁性存储器，但要满足在保 持反铁磁态的基础上实现所有磁性层同时发生整体性双态切换这一条件。 该研究为反铁磁动力学基础研究以及技术应用带来了变革性突破，也为未来低维磁性材料集成到自旋电 子学以及光电子领域中开辟了新的途径。 复旦大学物理学系吴施伟团队基于多年的技术 ...

来源：央视新闻客户端 29日，商业航天器及应用产业链共链行动大会在上海举办。会上发布了多项重磅消息，包括发射成本有 望降低、谋划推动太空旅游、太空资源开发、太空交通管理等新领域的发展等。将如何理解这些信息？ 来看总台央视记者在现场带来的观察。 总台央视记者 李宁：这次大会聚集了商业航天产业链上的方方面面的企业，比如像火箭制造、卫星制 造、落地应用场景等。那么，大家主要关心哪些方面内容？ 愿景报告还提到偏远地区的通信实现全覆盖，天通与北斗终端进一步打通海洋沙漠等通信盲区。还有像 厘米级的定位，实现无盲区覆盖。这些对于喜欢旅游或者是一些无人区作业的工作人员来说，都是非常 好的消息。未来在沙漠、海洋等，这些无人区，就可以实现通信覆盖了。 参会企业工作人员：火箭，卫星制造，还有海外出口。 参会企业工作人员：我们更关注的是供需信息的发布，这对整个产业链的协同发展是非常重要的，我们 也希望能作一些贡献，整个公司现在都在布局。 在这次大会上有一个重磅的发布，就是航天科技集团牵头编制的中国商业航天重大供需信息正式发布。 这个清单的发布也向外界传递了一个非常明确的信号，就是邀请大家合作共赢。 在产业发展的愿景报告里面，有非常多 ...

来源：科技日报 科技日报记者 叶青 通讯员 华轩 你有没有过这样的体验？长时间独处后，明明没发生什么事，却莫名心慌、坐立不安甚至对社交产生恐 惧。这其实是"社交隔离"在悄悄改变你的大脑。 近日，华南理工大学医学院王卓副教授团队联合浙江大学、南方医科大学等单位，在国际期刊《细胞· 代谢》上发表原创性研究成果。该研究首次揭示了社交隔离会触发大脑特定区域的"铁堆积"，进而激活 一条全新的神经可塑性通路——研究团队将其命名为"铁可塑性"。这项发现，不仅解开了"孤独伤脑"的 长久谜题，更开辟了一条无需传统抗焦虑药物无创可逆的全新干预路径。 团队通过模拟人类长期独居的小鼠模型，发现了被单独饲养的小鼠，其大脑中一个叫 "腹侧海马" 的区 域（更偏好调控情绪的海马亚区）铁含量异常升高。 别小看这点"铁"——它可不是补品！过量的铁，就会像"错误信号"一样，激活一种名为α-突触核蛋白的 分子，导致神经元过度放电。这就像电路短路一样，不断向身体发送"危险！快逃！"的焦虑警报。更关 键的是，这种变化精准锁定情绪中枢，让大脑对"社交剥夺"有高度特异的应激反应。 过去，人们认为铁只是维持神经健康的"营养元素"，但该研究证明，在心理压力下 ...

Core Insights - The research team from the University of Science and Technology of China has developed the world's first quantum sensing network based on atomic nuclear spins, significantly enhancing the sensitivity for dark matter detection [1][3] - Dark matter, which constitutes approximately 26.8% of the universe, interacts through gravity but not electromagnetically, making it challenging to detect [1] - The team has established a distributed detection network using five ultra-sensitive quantum sensors, achieving detection precision that surpasses astronomical observations by 40 times in certain mass ranges of axions, a leading candidate for dark matter [3] Group 1 - The quantum sensors are equipped with advanced features that allow for the storage of fleeting signals in a nuclear spin coherence state for nearly a minute, greatly extending the detection window [1] - The team has developed a self-research quantum amplification technology that enhances weak signals by a factor of 100, making it easier to detect subtle interactions [1] - The breakthrough adds a new, more precise tool to humanity's arsenal for searching for dark matter, with potential future enhancements in sensitivity through global networking and space deployment [3] Group 2 - The sensors are deployed in Hefei and Hangzhou, synchronized via satellite for precise timing, forming a distributed detection network [3] - After two months of continuous observation, the team has provided the strictest constraints on dark matter models across a wide range of axion masses [3] - Future collaborations with gravitational wave observatories could further increase detection sensitivity by four orders of magnitude [3]