研究据点wep

据点损失中……

[小盖 @xiaogaifun]

字节对大厂 AI Coding 的反思，好真实。

字节技术副总裁洪定坤的分享，我来回看了好几遍，很有启发。

字节在 AI Coding 方面的实践还是很有代表性的，推荐所有做研发的同学都可以看看。应该会感同身受。

我看完记了一整页笔记，分享给大家。

我甚至觉得可以把这个分享理解为字节在 AI Coding 上的一些真实反思。

根据自己的理解，我把这个分享里对我有启发的几个判断展开来聊一聊。

其中会夹杂很多我自己的感触，想看原文的可以直接去找演讲全文。

反思一：AI 代码贡献率不该是 KPI

AI Coding 基本上都已经逐步进入了各个公司的生产流程。

很多人都在说自己的业务有 90% 的代码是 AI 生成的，乍一听，感觉很恐怖。

但其实，AI 对研发的提效没有外界想象的那么高。

洪定坤举了 TRAE 团队的例子。TRAE 本身就是做 AI 工具的，所以这个团队对 AI Coding 的采用非常积极。

过去半年里，他们有 94% 的代码都是 AI 贡献的。但人均需求吞吐率只提升了 60%，也就是之前的 1.6 倍。

这就有疑惑了，AI 写代码的速度至少是人的 10 倍以上，如果 90% 以上的代码都是 AI 产出的，效率至少应该提升 5 倍或者 10 倍吧？为什么只提高了 1.6 倍？

字节得出来的结论是，单一的指标，比如 AI 代码占比，根本没有办法代表真实的生产力。

如果把 AI 代码贡献率当成 KPI，结果就是大家都去优化 AI 的生成量，而不是优化交付能力。

看起来 AI 用了很多，但系统的效率并没有真正变好。

那为什么 90% 的代码都是 AI 写的，人效才提了 1.6 倍？一个很重要的原因是，写代码只是整个研发流程的一个环节。

写之前要理解需求、写 Spec，写完之后要验证功能、提交发布，这些环节如果还是传统方式，光把写代码加速了，整体效率提不上去。

洪定坤把字节在这方面的尝试叫做系统化的 AI Development，核心意思就是 AI 不能只管写代码，得让它进入更多的研发环节，整条链路都跑通，效率才能真正上来。

前两天出去的时候还跟别人争论这件事。现在还有不少公司在追踪员工到底用了多少 Token，说白了，这是在追踪过程。

更应该关注的是，用了这个工具之后，从结果层面去看，交付到底有没有变得更快、更可靠。

一个团队天天说自己 AI 工具用得贼溜，消耗了多少 Token，但没有什么有效的产出，那这到底是好事还是坏事。我觉得这是一个值得每个管理者思考的新问题。

反思二：功能正确≠工程可用

Vibe Coding 的理想状态是就像聊天一样，用自然语言表达自己的需求，最后做出来想要的产品。

如果哪里不对，再用自然语言和 AI 沟通，让它修改。这是过去一年 Vibe Coding 风靡的思路。

对于不太复杂的应用，这种方式完全没问题。我自己做的一些项目基本上就是这个流程跑下来的。

但只要是做生产��的软件，无论公司大小，流程肯定不是这样。

因为公司里必然有老代码，有一套约束。怎么复用已有的组件，安全和权限怎么处理，性能怎么保证，还有兼容性、可维护性。

正经写过工程代码的人都清楚，Vibe Coding 描述的那个状态是比较理想化的，更适合做小项目和验证想法。

真正的程序员虽然也在 Vibe Coding，但流程跟理想状态不一样。

字节内部做了一个实验来验证这个判断：三个模型，三个 Agent 框架，两两组合成 9 种方案，针对同一个需求，每组跑 100 次，总共 900 次。

结果发现，AI 在功能正确率上表现还不错，所有组合都超过了 80%，也就是说，AI 把功能写对的能力已经过了及格线。

但无论哪个组合，生成代码的工程质量都不太好。比如 UI 不对，没有复用组件，性能有问题，结构不符合规范。

这些问题大家在用 AI 写代码的时候应该都碰到过。

现在所有上了牌桌的 Coding 模型，都已经过了 Opus 4.6 这个级别的临界点，模型可以自主写代码了。

这个时候影响 AI Coding 成败的绝对不是裸模型，而是裸模型加上 Harness 的能力。

这个判断本身不算新鲜。

但我最受触动的是字节对 Harness 的理解。

他们的反思是，整个行业好像还是把 Harness 等同于 Agent 框架，诸如用 single agent 还是 multi agent，包含哪些角色，角色之间怎么配合。

这些当然重要，但字节在实践中发现，真正决定 AI Coding 能不能落地的，反而是更基础、更工程化的东西。

洪定坤把它叫做基建。

比如上下文工程有没有做好，架构的约束够不够清晰，团队的知识能不能有效沉淀下来，过去的技术债有没有梳理清楚。

这些看起来不那么性感的工作，反而直接影响 AI Coding 的效果。

实验数据也验证了这一点。同样的模型和框架组合，把这些基建结合进去之后，功能正确率直接从 80% 提升到了接近 90%，工程质量得分，也从之前 40 到 60 分的不及格水平，普遍提升到了 80 分左右。

基建做不好的话，可能的后果是，Vibe Coding 感觉快了，但实际整体可能更慢。工程的债，迟早得还。

反思三：代码门槛下降之后，团队怎么协同

洪定坤分享里有一个例子让我印象很深刻。

产品经理有个需求，发现还得等研发排期，就说那我自己来吧，用 AI 三下五除二就把功能给实现了。

确实这个功能不复杂。做完之后产品经理把代码给到研发，说我已经把代码写完了，现在你只需要帮忙把功能上线就行。

研发一看，不行。你这代码能跑，但不符合上线的规范，有权限问题、安全问题。

产品经理就很委屈，你们没时间做这个需求，现在我都做完了又不让上线。可研发看到的其实是代码质量的问题。

所以这里面就有一个需要所有人正视的事情，虽然代码的生成门槛虽���下降了，这并不代表系统的复杂度也下降了。

真实的业务系统里，代码要放到已有的架构里，要跟已有的模块配合，还要考虑各种各样的问题。

绝对不是谁写出来就能直接上线的。不然肯定会出问题。

怎么让不同角色的人用同一套工具和规范做出符合要求的代码，这是接下来大家需要去解决的。

字节的思路是在内部尝试工具化。比如把内部实践直接产品化，沉淀到 TRAE 里面，开放给所有人。

其实说白了就是工具化。

我看朋友圈有好多大佬也都在转这篇文章，应该还是有挺多共鸣的。

我感觉这一次分享多少也是一些拨乱反正吧。因为过去一段时间确实有很多听起来很离谱的言论，有些人会疯狂地炫耀自己使用了多少 Token，会认为这就代表着 AI Native......

强烈推荐大家看看原文。字节跳动的公众号就有。

[ruanyf @ruanyf]

SQLite 数据库的作者 Richard Hipp 解释，为什么他的项目一律不接受外部的 PR。

PR 不是免费的。你实际上是对我提要求：你开发了这个很棒的功能，然后希望我帮你维护它、帮你编写文档、帮你测试，并在接下来的二十五年里一直为你维护它。

Linus 曾说过一句名言：Free 既可以指免费啤酒，也可以指言论自由。但还有另一种 Free：免费的小狗。“瞧，我这儿有只免费的小狗送给你。” 你明白我的意思了吧？

提交一个 pull request 就相当于有人送你一只小狗。一天下来，你的小屋里就多了一只小狗。你不能把它扔掉——你有道义上的责任照顾它，直到它自然死亡。

我可不要任何免费的小狗。https://t.co/kD8Kri3V9K

[Vonng @RonVonng]

表妹在闲鱼上被骗了2300，报警炸出来一个诈骗团伙，骗了近千人了。这个手法确实很有迷惑性，利用千问 CDN 挂假闲鱼，在支付宝 App 里的浏览器会直接放行，全程在阿里 App 内完成。写了篇文章，剖析了一下诈骗手段，并嘲笑了一下支付宝和阿里。链接见评论区。

[研究据点wep @lospeeJD]

@xczyuan 巧姐儿：你继续说

[研究据点wep @lospeeJD]

@NullableX 我觉得反而证明了北京对金融的态度

[研究据点wep @lospeeJD]

@anyeweiguang 哎，头部学校报名和入学比例巨大，到最后还是看房本。只有一个办法可以打压学区房，那就是教育资源轮转。

[研究据点wep @lospeeJD]

@yihong0618 龙虾非常重要，增加了很多同学部署、调试和与客户解释问题的能力，经此一役，我觉得不少人都可以去干售前了。

[研究据点wep @lospeeJD]

@Shenyilusi @fxtrader 打游戏为什么不上7900XTX

[研究据点wep @lospeeJD]

@maybeanormalppl 吃喝比住还是重要的吧，我没有离家出走的经验，找地方趴一晚上的经验有一点

[联合国教科文组织 @UNESCO_chinese]

侨批是一份 #世界记忆 。

它是书信与汇款合一的华侨家书，盛行于潮汕等地，在19和20世纪被大规模使用。它既是维系亲情的纽带，也是中国历史的见证者。

#世界记忆 项目是教科文组织专门负责 #文献遗产 保护与传播的旗舰项目。《侨批档案-海外华侨银信》已列入世界记忆名录➡️https://t.co/0EkZ3JDGtg

[Macro_Lin ｜ 市场观察员 @LinQingV]

长鑫存储扩产的设备供应链全景
长鑫现在合肥两座、北京一座，三座12英寸厂。合肥一厂11万片月产能，二厂8万片，北京厂7万片，加起来接近30万片，全部满产。两年涨了三倍，2024年初差不多10万片，2025年底冲到28到30万，2026年目标稳在30万。按晶圆片数算全球DRAM占比已经接近15%，但按销售额算只有3.97%，片数堆上来了，单价还在追赶。

工艺端，G4在16纳米已经量产，DDR5良率从2024年底的80%升到现在的90%区间。G5对应15纳米，2026年底量产。HBM2去年下半年就上了，比外界预期提前两年，主要供华为昇腾910。HBM3的前段晶圆产能主要在合肥和北京现有基地内部腾出来，目标月产6万片，约占30万片总产能的20%。后段的堆叠和封装由上海新建的HBM封测厂承接，2026年底投产。

产能还在继续扩。上海新厂分两期，Phase 1产能10万片，2027年初量产，Phase 2同样10万片，2028年初量产。加上现有的30万片，远期产能目标是奔着50万片以上去的。华为也在建一条10万片规模的产线，专门配合长鑫做昇腾系列所需的HBM3和LPDDR5X，2026年下半年开始全面量产。

IPO募资295亿，DRAM技术升级拿了130亿，量产线升级75亿，前瞻技术研发90亿。多家券商纪要提到其中约200亿会直接变成设备采购订单，叠加2024年712亿的存量Capex节奏，是国产设备厂未来两年���确定的基本盘。

长鑫正在进行的大规模扩产，设备采购需求非常可观。每个品类的国产化进展差异很大。

光刻是最大的对外依赖，国产化率连5%都不到。主力机器是ASML的NXT:1980Di，每小时出275片，覆盖到16纳米节点，目前还能正常采购。被荷兰管制的是NXT:2050i及以上，每小时295片，2023年9月起要许可证，2024年1月起对华许可基本停发。修正一个流传很广的说法，275到295片是1980Di到2050i的代际跳变，不是1980系列内部的迭代。长鑫往15纳米以下走，这道墙绕不开。上海微电子的28/14纳米DUV还在攻关，光刻这一环短期无解。

刻蚀占设备投资25到30%，是国产化最深的核心工艺。北方华创的ICP在长鑫产线上市占超过50%，中微的CCP介质刻蚀机做到50比1以上深宽比，专门用于HBM的TSV深硅通孔。但存储节点100比1深宽比的极端工艺，孔洞必须互相平行规整，任何偏差直接砸良率，Lam和东京电子在这个区间仍然是主力。

薄膜沉积占约25%，品类分得细。PECVD做绝缘层，PVD做金属互连，ALD做电容器的高k介电层。拓荆科技的PECVD已经批量进入长鑫DDR5和LPDDR5产线，北方华创的PVD覆盖铝铜溅射和氮化钛溅射。但DRAM电容器核心的高k ALD设备，应用材料和ASM的位置很难动，拓荆的ALD在验证爬坡中还没有大规模上量。整体看PECVD和PVD的国产化率高一些，ALD最低。

CMP只占设备投资5到7%。华海清科占国产CMP装备销售90%以上份额，12英寸Universal-300已经导入长鑫。新变量是中微4月29日刚过会的收购杭州众硅，6抛光盘架构是国际首创，效率比主流4盘��案高一截。国产CMP从单点供应正式进入双供。

清洗国产化率30到40%，盛美上海的SAPS/TEBO兆声波清洗覆盖FinFET和DRAM的16到19纳米制程。热处理40到50%，北方华创立式氧化炉打头，激光退火端莱普科技两年内国内市占从3%涨到16%，跟长鑫和长存共同开发匹配DRAM工艺的设备。这几个环节已经跑通了。

最薄弱的三个环节是量测检测、涂胶显影和离子注入。量检测国产化率个位数，KLA全球占51到54%，精测电子进了长鑫12英寸产线，中科飞测也在部分环节往里切，但高端光学检测和电子束检测差距仍然明显。涂胶显影国产化率仅4%，东京电子在大陆市占超过90%，芯源微是唯一量产替代，目前只在封装端站住了，前道关键层还进不去。离子注入同样个位数，万业凯世通累计交付40多台。弹性最大的三个环节，也是短期内最动不了的三个环节。

长存三期产线2026年一季度国产设备占比首次过50%，目标100%。长鑫设备国产化率已突破45%，但仍低于长存三期产线的水平。根源在工艺。DRAM的电容刻蚀和光刻精度对先进DUV的依赖远高于3D NAND。NAND可以靠堆层数绕过光刻瓶颈，DRAM要正面硬刚，结构性差异，跟意愿无关。

外部约束在收紧。4月22日美国众议院外交事务委员会投票通过了MATCH法案，路透社称之为"国会史上最大规模的半导体出口管制立法审议"。法案把长鑫、中芯国际、长江存储、华为、华虹一起列入covered facility，禁止ASML的DUV浸没式光刻机对长鑫出口，禁止盟国为既有设备提供维保，要求荷兰和日本在150天内对齐美方规则。主要推手是美光。一旦通过，NXT:1980Di这条最后的灰色通道就堵死了。长鑫本身还没进BIS实体清单，但设备进口其实早就在2022年10月那波18纳米以下DRAM工艺限制的笼子里。

45%的国产化率意味着长鑫在两条腿走路，海外设备保良率，国产设备保供应链安全。MATCH法案的压力反而在加速这个进程。长鑫每往国产设备多切一个百分点，对应的就是北方华创、中微、拓荆、华海清科、盛美这些公司实打实的订单增量。上海新厂分两期共20万片的增量产能，加上现有产线的持续技术升级，未来两到三年国产设备厂面对的是一个确定性极高的需求窗口。

从更长的时间尺度看，长鑫的扩产不只是一家公司的资本开支计划。它每走通一个工艺节点，整条12英寸国产设备平台就多一次被验证、被复用的机会。刻蚀和CMP已经证明了这条路径，薄膜沉积和清洗正在跟进，量测和涂胶显影是下一个要啃的硬骨头。这个验证循环一旦转起来，国产设备��竞争力会以远超线性的速度积累。

[中华人民共和国商务部MOFCOM @MOFCOM_China]

Q: Recently, MOFCOM issued the Overall Plan for Developing National Demonstration Zones for the Innovative Development of Trade in Services, and identified 17 localities, including Beijing, Tianjin, and Shanghai, to take the lead in establishing demonstration zones. Could you share the overall considerations behind the development of these demonstration zones? (MOFCOM Regular Press Conference May 14, 2026)

A: The development of national demonstration zones for innovative development of trade in services is a major decision and plan made by the CPC Central Committee and the State Council. MOFCOM, together with relevant departments, has formulated the overall plan, which specifies the overall objectives, main tasks, and pathways for developing the demonstration zones. Building on the previous pilot programs for innovative development of trade in services, we will focus on demand-driven approaches, breakthroughs in reforms, technology empowerment, and opening up and cooperation. We will continue to develop                                a number of major open cooperation platforms for trade in services that rank among the top in scale, gather high-end factors, demonstrate outstanding innovation capabilities, facilitate trade-industry linkages, and exert significant spillover effects. We will also cultivate a group of internationally competitive service export enterprises and key service industry clusters, and establish a more robust institutional system and business environment for innovation-driven development, thereby leading and driving nationwide innovative development in trade in services.

We will start the development of demonstration zones in 17 regions, including Beijing, Tianjin, Shanghai, Chongqing, Dalian, and Xiamen, and by 2035, the total number of demonstration zones is expected to reach around 35. Specifically, we will carry out differentiated exploration in areas such as pursuing integrated innovation in industrial and supply chains, fostering the competitiveness of market entities, facilitating the efficient supply of production factors, advancing high-standard opening up, and building domestic and international promotion systems. At the same time, we will guide local governments to leverage their resource endowments and geographical advantages to promote the coordinated development of trade in services, investment, consumption, and industries.

Moving forward, MOFCOM will work with relevant departments and local authorities to implement the plan, and will support and guide other eligible cities to actively apply for the establishment of demonstration zones, so as to continuously promote the high-quality development of trade in services.

[中华人民共和国商务部MOFCOM @MOFCOM_China]

MOFCOM Spokesperson’s Remarks on the Determination That the EU’s Relevant Practices in Foreign Subsidy Investigations Constitute Improper Extraterritorial Jurisdiction

Q: On May 15, the Ministry of Justice of the People’s Republic of China issued an announcement determining that the EU’s cross-border investigative practices against Chinese entities under its Foreign Subsidies Regulation (FSR) constitute “improper extraterritorial jurisdiction.” What is MOFCOM’s comment on this?

A: China has always opposed the EU’s abuse of unilateral tools such as the FSR to suppress Chinese enterprises. Recently, the EU has not only increased the frequency and scope of its investigations targeting Chinese companies, but also escalated its probes into enterprises, including Nuctech, to in-depth investigations. Furthermore, it has compelled Chinese banking institutions to cooperate with these investigations, unreasonably demanding extensive access to a large amount of information within China that is unrelated to the investigations. The normal investment and operation of multiple Chinese companies and banking institutions in Europe have been severely and adversely affected.

As early as January 2025, MOFCOM concluded through investigation that the relevant EU practices under the FSR constitute trade and investment barriers. MOFCOM called on the EU side to correct these practices and advocated for the proper management of differences through dialogue. However, the EU side has remained obstinate and gone further down the wrong path.

In response, following a comprehensive investigation, the Ministry of Justice, in conjunction with MOFCOM, has determined that the cross-border investigative practices taken by the EU against certain Chinese entities during its FSR probe into Nuctech constitute improper extraterritorial jurisdiction measures. This determination was made in accordance with the Regulations of the People’s Republic of China on Countering Foreign States’ Unlawful Extraterritorial Jurisdiction. Consequently, it has been ordered that no organization or individual shall execute or assist in the execution of such measures.

China reiterates its hope that the European Commission will promptly correct its erroneous practices, cease its unjustified suppression on Chinese enterprises, stop abusing the FSR investigation tool, and provide a fair, just, and predictable business environment for Chinese enterprises investing and operating in Europe. China has always advocated for managing differences through dialogue and consultation, and hopes that the EU side will pull with China in the same direction to resolve the issue through friendly consultations. At the same time, we will closely follow the EU’s subsequent actions and take necessary measures to resolutely safeguard national security and the legitimate rights and interests of Chinese enterprises. (Released on May 16)

[研究据点wep @lospeeJD]

@chenshaoju 要有别的方案，我们和美国完全不同，地产可能触底了

[Casemiro @Casemiro]

There are places we pass through in life… and there are places that become part of who we are.
Manchester will forever be my home.

To the city, the club, and every supporter, my sincerest thank you. These past four years have been unforgettable, filled with moments my family and I will carry with us for the rest of our lives. There simply aren’t enough words to describe the happiness and warmth we’ve felt here.

Thank you for every cheer, every memory, and for making us feel at home from the very first day.

Forever a Red Devil ❤️

[研究据点wep @lospeeJD]

@POI74103174 这下面的，长得都不像牛肉了。

[Evan A. Feigenbaum @EvanFeigenbaum]

It is noteworthy to me that official Chinese statements, before and after the visit, have been calling (aspirationally) for "predictable" and "stable" U.S.-China relations. If this was the goal, then I think we are going to need to distinguish “predictability” from “stability." I notice that a lot of people in the media, here on social media, and among the commentariat are either conflating the two or else using them in the same sentence, as in: “President's Trump's visit has brought greater predictability and stability” to relations between Washington and Beijing. But I'm sorry, these two words are NOT the same thing.

We might now get greater predictability— which is an achievement with a president as mercurially unpredictable as Trump. But we should have no illusions that this has “stabilized” U.S.-China relations because all of the things that made them unstable in the first place are still with us: (1) clashing security concepts, (2) obvious differences of political system and ideology, (3) contending economic interests, (4) technology competition as emerging and foundational technologies, from AI enabled applications to quantum, become inherently dual use, and (5) combustible domestic politics where there is, frankly, little benefit on either side in advocating for a "stable" relationship.

In analytical terms, I suspect Beijing is pleased, not least because China will settle for a more predictable relationship with Donald Trump. But we are nowhere near “stable.” And I thought much of the visit was well-done performance art, directed at specific domestic audiences, on both sides.

Incidentally, I am quite sure that Xi Jinping did not, in fact, say that “America is the hottest country in the world,” as the President told the press gaggle on Air Force One. So we should take much else that Xi is alleged to have said with a heaping dose of salt.

[Kyle Chan @kyleichan]

It's interesting to scan through MIIT's list of 67 pilot projects for high-tech industrialization to get a sense of what tech China thinks is important:

High-End Functional and Intelligent Materials

1. Ultra-high energy-density dielectric materials and devices — Tsinghua University
2. Data-driven design and manufacturing of copper alloys for integrated-circuit lead frames — University of Science and Technology Beijing
3. High-sensitivity entropy-regulated amorphous-alloy stress-impedance strain gauge for seismic resistance and disaster prevention in buildings — Nanjing University of Science and Technology
4. Noble-metal reduction technology for selective hydrogenation — Zhejiang University
5. Complete preparation technology for liquid-cooling thermal-management materials — Juhua Group
6. Single-phase immersion liquid-cooling solution for data centers — Juhua Group
7. 450 km/h high-speed train traction motor integrating new electromagnetic materials — National High-Speed Train Qingdao Technology Innovation Center
8. Aviation fuel coalescence-separation device — South China University of Technology
9. Inorganic two-dimensional material membranes for efficient hydrogen separation — South China University of Technology

Advanced Structural and Composite Materials

10. Preparation technology for sound-absorbing honeycomb and composite materials — China Aviation Manufacturing Technology Research Institute
11. Wear-resistant, fatigue-resistant, corrosion-resistant rails and frogs for the Sichuan–Tibet Railway — China Academy of Railway Sciences Corporation
12. Long-term performance-retention technologies for structural concrete in complex environments on the Sichuan–Tibet Railway — China Academy of Railway Sciences Corporation
13. High-performance shotcrete technology for complex environments on the Sichuan–Tibet Railway — China Academy of Railway Sciences Corporation
14. Crack-resistance improvement technologies for structural concrete in complex environments on the Sichuan–Tibet Railway — China Academy of Railway Sciences Corporation
15. Powder-making technology and applications for recycling coarse high-temperature alloy powder — AECC Beijing Institute of Aeronautical Materials
16. Large-tonnage carbon-fiber composite cables — University of Science and Technology Beijing
17. Multi-layer gradient cold-spray repair and nano-hard reinforcement composite plating for continuous-casting molds — Ansteel Group Beijing Research Institute
18. Key technologies and application development for ultra-high-stiffness magnesium-matrix composites — Harbin Institute of Technology

New Displays and Strategic Electronic Materials

19. Sub-6GHz GaN radio-frequency devices — CETC 13th Research Institute
20. High-frequency, high-power laser modulator technology — Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences

Rare-Earth New Materials

21. Preparation technology for high-temperature-resistant cobalt-based permanent magnet materials — China Jiliang University
22. Heavy-rare-earth-free high-coercivity sintered NdFeB technology — Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences

High-Performance Manufacturing Technologies and Major Equipment

23. 6-inch semi-insulating SiC crystal-growth furnace and 6/8-inch compatible SiC epitaxial furnace — NAURA
24. New MOCVD equipment for Micro-LED — Advanced Micro-Fabrication Equipment Inc. China
25. Precision forming technology for large aerospace thin-walled aluminum-alloy integrated cylindrical sections — Shanghai Aerospace Precision Machinery Institute
26. High-temperature-resistant, corrosion-resistant transmission-system bearings — Luoyang Bearing Group
27. High-performance seals for aviation hydraulic systems — Guangzhou Mechanical Engineering Research Institute
28. Key manufacturing technology for ultra-large seamless titanium cathode rollers — Xi’an Taijin New Energy Technology

Intelligent Sensors

29. Advanced sensors, core components, and manufacturing processes for spacecraft control systems — Beijing Institute of Control Engineering
30. Flexible intracranial implantable multimodal sensing and modulation system for multiparameter brain monitoring — Aerospace Information Research Institute, Chinese Academy of Sciences
31. Self-powered sensor technologies for human health monitoring — Beijing Institute of Nanoenergy and Nanosystems
32. High-sensitivity MEMS magnetic sensing components and sensors — State Grid Smart Grid Research Institute
33. Miniature high-performance accelerometers — Beijing Aerospace Xinghua Technology
34. Rocket sensors — Long March Rocket Technology
35. New broadband ng-resolution triaxial accelerometer — Tianjin SIASUN Robot & Automation
36. Automotive-grade high-precision integrated navigation sensors — Hebei Meitai Electronic Technology
37. Series of sensors for deep-sea environmental observation and resource exploration — Shenyang Institute of Automation, Chinese Academy of Sciences
38. Multi-parameter differential-pressure flowmeter — Shenyang Zhongke Bowei Technology
39. Quantitative sensing-interface model and analytical instrument technology based on resonant cantilever beams — Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences
40. High-performance X-ray sensors — iRay Technology
41. Electromagnetic sensors — Xinlian Superconductor Shanghai
42. Thin-film getter structure with a micro-heater and its manufacturing method — Shanghai New Micro Technology R&D Center
43. High-performance acoustic sensing elements and sensors — Wuxi Weigan Semiconductor
44. New high-performance MEMS gas sensors — Suzhou Huiwen Nanotechnology
45. MEMS sensor mass-manufacturing platform — XINLIAN Integrated Circuit Manufacturing
46. Development and application of diamond quantum magnetic sensors — University of Science and Technology of China
47. In-situ continuous temperature sensors and measurement systems for molten steel — Maanshan Iron & Steel
48. High-performance laser gas-sensing components — Shandong Science & Technology Innovation Group
49. Key technologies and applications for biosensor sensitive elements — Shandong Kanghua Biomedical Technology
50. Wireless passive temperature sensors based on polymer-derived ceramic metamaterials — Zhengzhou University
51. High-precision printing technology and equipment for ultrafine fiber surfaces — Huazhong University of Science and Technology
52. Complete sensor set for high-speed rail vehicle health-monitoring systems — CRRC Zhuzhou Institute
53. Industrializable mass-producible automotive-grade solid-state LiDAR for autonomous driving — RoboSense
54. Sensing-computing integrated room-temperature infrared imaging detection technology — Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences
55. Nanometer-precision displacement-measurement grating sensors — Xi’an Jiaotong University
56. Distributed thin-film sensors for highway infrastructure — AVIC Electromechanical Measurement Instrument Xi’an

Industrial Software

57. Core components of an industrial internet operating system for discrete industries — Beijing Institute of Technology
58. Data-driven closed-loop performance analysis, regulation, and optimization technology and software for manufacturing processes — University of Science and Technology Beijing
59. Intelligent analysis and decision-making system for full-process industrial data in discrete manufacturing — Beihang University
60. Distributed time-series data management system Apache IoTDB — Tsinghua University
61. MEC-based edge control and real-time simulation theories and methods — Shenyang Institute of Automation, Chinese Academy of Sciences
62. Cloud-based service-oriented MES and intelligent management-control platform system — Beijing Xiaomi Mobile Software
63. Domestic isogeometric-analysis software ADIGA — Dalian University of Technology
64. Distributed factory industrial interconnection platform — Shanghai Aircraft Manufacturing
65. Industrial interconnection platform for personalized customization industries — Guangzhou MINO Equipment
66. End-edge-cloud interconnection integration technology and system for OT/IT convergence — Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences
67. Industrial interconnection platform for large-scale manufacturing industries — Gree Electric Appliances of Zhuhai

[研究据点wep @lospeeJD]

@XianyuLi 小额收账一定��二维码强于NFC。对于商户来说，不需要额外设备投入，对于付款人来说，完全掌握自己付款金额，更加安心。对于定额项目，比如说门票，闸机，NFC一定更方便。这两个一起用就是了，有什么好比的。

[Chinese Embassy in US @ChineseEmbinUS]

#China and the #US have agreed on a new vision of building a constructive China-U.S. relationship of strategic stability.🇨🇳🇺🇸