Group 1 - The overall A-share market showed a slight upward trend on July 11, with major indices mostly rising, particularly in sectors like brokerage, biotechnology, insurance, and heavy machinery [1] - The recently popular Cloud Computing 50 ETF (516630) saw a slight increase, with top-performing holdings including Dongfang Guoxin, Fanwei Network, Zhongke Xingtou, and others [1] - xAI released its latest AI model, Grok4, which achieved a comprehensive score of 73, surpassing OpenAI's o3-pro (71) and Google's Gemini 2.5 Pro (70), indicating it is currently the strongest AI model [1] Group 2 - The competitive landscape for large models is taking shape, with companies increasing investments to become significant players in the field [2] - xAI plans to launch a professional coding model in August, a multimodal intelligent agent in September, and a video generation model in October, aiming for comprehensive AI coverage within three months [2] - OpenAI is set to release GPT-5 in the summer, while Google is testing the next-generation Gemini 3.0 series models, indicating ongoing competition and demand for computing power [2]

4.7 日后，中美两国股市都迎来了一轮上涨： 纳斯达克累计涨幅 32.9% 。恒生科技指数 ETF （ 513180 ）累计涨幅 11.57% 。上证涨了 12.16% 。 A 股整体的上涨幅度没有美股大，从结构上看，也有 很大的差别。 逻辑是对的： 训练芯片以 H100 和 H200 为首，从 2.28 日以后其主力云上价格是往下得： 分析里面的原因：还是基础大模型的迭代变慢所致，在 transformer 架构下的基 础模型迭代明显变慢了，去年 11 月我们还能说是因为算力不够的原因，大 模型迭代受阻。但是从今年 3 月 B200 开始交付， 5 月以后情景看，基础模型的升级就是变慢了。而且从我们跟踪的行业趋势看，预训练美国大厂们还没 放弃，但是基本都把大模型迭代的重点全力投向了 RL 强化学习路径，马上要出来的 GPT5 也是这方面的集大成者。 RL 后训练强化学习阶段，老黄也认为遵 循 scaling law 法则，但比起预训练的暴力增参的大幅需求， RL 的需求显然要小一些。所以从 2 月全球一窝蜂的 转向强化学习这个路径后， H100 和 H200 的租赁价格是明确下降的，这里面也有地区影响，比如 ...

Core Viewpoint - The AI industry is experiencing significant growth, with major companies like xAI and Nvidia leading the way in advancements and market valuation [3][4]. Group 1: Company Developments - xAI has secured $10 billion in funding, raising its valuation to $113 billion, with new products like Grok 4 and Grok 4 Code set to launch soon [3]. - Nvidia's market capitalization reached $3.84 trillion, reflecting strong demand for its AI hardware, particularly the GB200 model [3]. Group 2: Industry Trends - The AI sector is witnessing a substantial increase in token consumption, with Google’s monthly token usage projected to grow from 9.7 trillion in April 2024 to 480 trillion by April 2025, a 50-fold increase [3]. - Microsoft Azure AI infrastructure processed over 100 trillion tokens in Q1 2025, a fivefold increase from the previous year, indicating robust demand for AI services [3]. - ByteDance's Volcano Engine reported a daily average token usage of 16.4 trillion in May 2025, a 137-fold increase compared to May 2024 [3]. Group 3: Market Outlook - The overall fundamentals for AI hardware have improved significantly, with new software models like Deepseek R2 and GPT-5 expected to launch, indicating ongoing growth in the AI industry [4].

DeepSeek-R2，终于要来了？ 大模型竞技场秘密上线了一个叫 steve 的神秘模型， 在对话中透露自己来自DeepSeek 。 克雷西 发自 凹非寺 量子位 | 公众号 QbitAI 不过，网友们并不满足于知道steve的厂商，开始讨论起了steve的具体身份。 有人说是R2，有人说是V4，还有人说是旧版本的升级…… 考虑到奥特曼的GPT5和马斯克的Grok4也在热传，如果steve真的是R2，那这个七月就真的太热闹了。 神秘steve引发猜测 在大模型竞技场的公开页面里，并不能找到steve的身影，不过 在前端代码当中，还是能发现有steve的痕迹 。 | ... DevTools - Imarena.ai/?mode=direct | | | | | | | --- | --- | --- | --- | --- | --- | | ir LD 源代码/来源 Lighthouse 元素 应用 | 控制台 | 内存 | 网络 | 性能 | 隐私与安全 | | | | | | | | | > ...

硅谷模型大厂变化：对预训练和 Capex 的影响？ 20250702 摘要 Meta 积极挖角 OpenAI、谷歌和 Anthropic 等公司人才，尤其集中在 多模态处理、后训练和推理模型领域，旨在提升 LLAMA 模型竞争力， 加速下一代大模型研发。 Meta 通过收购 Skill AI 加强数据筛选能力，弥补参数量不足，同时加速 集群建设，学习 OpenAI 模式，综合提升数据、人力和算力，推动下一 代模型发展。 OpenAI 的 O1 模型核心团队成员流失，任宏宇、赵胜佳和余佳慧等关 键人物被挖走，对 OpenAI 造成一定冲击，促使其加快发展步伐。 AI 人才薪酬大幅上涨，顶尖人才年薪高达一亿美元，显示科技公司对 AI 人才的争夺白热化，扎克伯格通过高薪全面压制竞争对手。 2025 年下半年，科技公司将回归预训练阶段，Meta、谷歌和 OpenAI 采取不同策略：Meta 侧重数据，谷歌优化架构，OpenAI 坚持大集群战 略。 Q&A 近期硅谷的变化对全球大模型 AI 演绎节奏有何影响？ 近期硅谷的变化显著加快了全球大模型 AI 演绎的节奏。今年（2025 年）下半 年，美国将进入新一轮的大模型迭 ...

Microsoft TodayMSFTMicrosoft$476.84 -1.20 (-0.25%) 52-Week Range$344.79▼$480.69Dividend Yield0.70%P/E Ratio36.79Price Target$515.68Add to WatchlistMicrosoft Corporation NASDAQ: MSFT stock is up 23% in the last three months and is now at an all-time high of around $478. At times like this, it’s normal for investors to consider taking some profits. Some technical signals point to the potential that the rally is getting tired. However, MSFT stock is one of the clear winners in the market this year.Get Microso ...

Group 1 - The core viewpoint of the news is the merger plan between Haiguang Information and Zhongke Shuguang, where Haiguang Information will absorb Zhongke Shuguang through a share exchange, leading to the latter's delisting [2] - Zhongke Shuguang is currently the largest shareholder of Haiguang Information, holding a 27.96% stake prior to the merger [2] - Following the merger, shareholders of Zhongke Shuguang will receive 0.5525 shares of Haiguang Information for each share they hold, indicating a potential 21.5% price increase for Zhongke Shuguang based on pre-suspension closing prices [2] Group 2 - The computer ETF (159998) has seen a significant inflow of funds, totaling 337 million yuan over 17 days, marking a historical high in both scale and share count [2] - The cloud computing ETF (517390) has attracted over 200 million yuan this year, focusing on leading companies in the AI sector [3] - The A-share market is expected to return to a structural bull market, with potential catalysts in AI, embodied intelligence, and national defense industries [3] Group 3 - The computing power leasing industry is experiencing high demand, as evidenced by large orders and procurement plans from various companies, indicating a high level of industry prosperity [4] - Domestic computing power is transitioning from "single-point breakthroughs" to a collaborative development across the entire industry chain, enhancing overall performance and stability [4] - The AI sector shows significant growth potential and investment value, particularly driven by domestic substitution and technological advancements [4]

截至2025年6月10日 13:22，中证数字经济主题指数(931582)下跌1.99%。成分股方面，深信服(300454)领跌7.14%，用友网络(600588)下跌6.22%，金山办公 (688111)下跌5.22%，龙芯中科(688047)下跌4.98%，拓荆科技(688072)下跌4.59%。数字经济ETF(560800)下跌1.98%，最新报价0.74元。流动性方面，数字经 济ETF盘中换手2.74%，成交2143.73万元。（以上所列示股票为指数成份股，仅做示意不作为个股推荐。过往持仓情况不代表基金未来的投资方向，也不代 表具体的投资建议，投资方向、基金具体持仓可能发生变化。市场有风险，投资需谨慎。) 申万宏源指出，A股重回结构牛行情，大概率需要等待科技重磅产业催化兑现。国内AI、具身智能、国防军工都具备成为结构牛核心产业趋势的潜力。下半 年值得重点关注的产品包括：DeepSeekR2、GPT5和特斯拉Optimus量产版本。但在重磅产业催化前，科技也是震荡市，重点把握高性价比区域的反弹和主 题机会。鼓励并购重组和创投一级市场联动形成共振，第三次大的新经济一二级联动行情正徐徐展开。从一级市场创投融 ...

长按即可参与 作者： 门捷列夫学徒 来源：雪球 自DS出现以后，我看网上没有很好的一篇文章去总结目前中美ai叙事的差异以及背后的算力需 求。简单来说： China Ai叙事：DS开源带来的LLM平权以及由此延伸的各种ds蒸馏小模型端侧（包括一体机）、 边缘侧的ai应用落地，大有全面开花之势。 China算力：H20和国产Asic就行，甚至端侧还有更简单的方案。当然杰文森悖论还是适用的，大 厂加大数据中心投入。 北美Ai叙事：继续大力出奇迹通往AGI，各种多模态高阶模型推出，万亿参数大模型继续走起。 北美算力：高端NV卡仍然供不应求，GPUs are melting。。。 芯片来跑。 所以各位小伙伴期待的国内版文生图可能就要等等了，哪怕推出了估计也会卡得要死。主要是还 是卡不行。 其实你看tx的微信就是聊天，所以llm对于他们暂时就够用。而zj因为是视频起家，肯定对于高端 卡的需求是更大的。前面没说的是Ai视频生成更需要500+tflops和48G显存，更需要NV的高端 卡。 这也是为什么我们直到现在都没看到国内的 csp 推出多模态的Ai应用 。 现在的gap在于：显然北美这条路出成果没这么快，而国内ai叙事 ...