2025年2月01日，星期六，第1170期

“DeepSeek效应”首先应从西方资本利益内部矛盾进一步尖锐计划的角度加以观察

【媒体报道】

2月1日，华为云发文《没有硝烟的战争：守护DeepSeek的83小时保卫战》

2024年1月25日凌晨，DeepSeek服务器集群遭恶意攻击，国内顶尖科技企业团结起来，开展了一场83小时的保卫战。包括360、华为云等多家企业参与其中，各展所长，通过技术手段锁定攻击源、搭建防护层、提供情报支持等，最终攻击流量下降97.0%，成功守住了我国AI产业高地。

【讨论纪要】

●一场基于中国与美帝国主义之间不可调和矛盾下的“网络战争”

1月27日，受中国人工智能初创公司——深度求索公司（DeepSeek）冲击，美国人工智能主题股票遭抛售，美国芯片巨头英伟达股价历史性暴跌，纳斯达克综合指数大幅下跌。

截至当天收盘，英伟达公司股价下跌16.97%，市值一日内蒸发近6000亿美元，创美国历史上任何一家公司的单日最大市值损失。英伟达的暴跌对整个市场造成了冲击，博通公司股价下跌17%，超威半导体公司（AMD）股价下跌6%，微软股价下跌2%。此外，人工智能领域的衍生品，如电力供应商也受到重创。美国联合能源公司股价下跌21%，Vistra的股价下跌29%。

此次美国股市震荡的核心原因是DeepSeek的最新突破，严重动摇了美国科技行业的全球霸权（技术本位制）地位。

2024年底，DeepSeek发布了新一代大语言模型V3，并宣布开源。测试结果显示，该模型在多项评测中表现优于主流开源模型，且具有成本优势。2025年1月，DeepSeek在世界经济论坛2025年年会开幕当天发布了最新开源模型R1，再次引发全球关注。

据介绍，R1模型在技术上实现了重要突破——用纯深度学习方法让AI自发涌现出推理能力，在数学、代码、自然语言推理等任务上，性能比肩OpenAI的o1模型正式版，且训练成本仅为560万美元，远低于美国科技巨头的数亿美元乃至数十亿美元投入。

美国科技股出现大规模暴跌后，爆火海内外的DeepSeek遭到大规模网络恶意攻击。1月28日，深度求索官网显示，其线上服务受到大规模恶意攻击。网络安全公司奇安信称DeepSeek此次受到的网络攻击IP地址都在美国。

对方举国动员，倾尽全力。我们也是全国动员。在我们的观察与评估中，这是典型的网络战争，且这场战争不可避免。这也是中美之间，或者更准确的说是中国与美帝国主义之间不可调矛盾的具体表现之一。值得一提的是，在这次“DeepSeek攻防战”中，华为提供的“底层（软、硬件）保护”起到了核心关键作用。

此外，我们也对DeepSeek选择与华为进行全面合作表示欢迎。

●“打不过就加入”的英伟达、微软和亚马逊与DeepSeek的“头号黑粉儿”——马斯克

值得一提的是，美国科技巨头的数亿美元乃至数十亿美元投入很大一部分集中在硬件上，也就是英伟达的算力卡上，这使得英伟达通过“卖铲子”编织的神话就此破灭，导致了英伟达股票的暴跌。此外，OpenAI在算法上编织的神话也就此破灭，人们意识到OpenAI此前不过是在“讲故事”。英伟达和OpenAI的股票大跌是美国股市在“DeepSeek效应”打击下暴跌的一个缩影，而这次美国股市巨震的影响之大，已经涉及到美国经济，尤其是金融维稳问题。这对于为纽交所敲开盘钟并公开表示“股市就是一切”的美国总统特朗普来说是无法接受的。这或是美国在“DeepSeek效应”显现之初，举全国之力对DeepSeek进行绞杀的主要原因所在。

值得注意的是，在“DeepSeek攻防战”以攻击流量下降97.0%，中方完胜告一段落后，美国，乃至整个西方内部，对DeepSeek的态度出现了明显分化：

第一，因“deepseek效应”股市大跌的英伟达1月31日宣布，DeepSeek-R1模型已作为NVIDIA NIM微服务预览版在英伟达面向开发者的网站上发布。英伟达网站介绍称，DeepSeek-R1模型是最先进、高效的大型语言模型，在推理、数学和编码方面表现出色。1月30日报道，与美国开放人工智能研究中心（OpenAI）是密切合作伙伴的微软宣布，目前已将DeepSeek-R1正式纳入Azure AI Foundry，成为该企业级AI服务平台的一部分；

第二，同样在1月31日，微软作为OpenAI的最大股东，决定将DeepSeek-R1模型纳入其AI服务。微软强调，DeepSeek-R1模型已通过“严格的红队测试与安全评估”，并经历“模型行为自动化检测与广泛的安全审查，以降低潜在风险”。

微软也在声明中提到，后续微软仍会持续评估DeepSeek-R1模型，并可能进行一定程度的调整与优化，以提升其准确度及审查机制；

第三，也是在1月31日，亚马逊表示，DeepSeek-R1模型现在已可以在Amazon Web Services上使用。甚至该公司首席执行官安迪·贾西告诉用户“尽管用”。

在我们的观察与评估中，在这些世界知名的美国高科技公司“突然”愿意和DeepSeek开展合作的背后，是美国，乃至整个西方，在亲身经历他们最后的“杀手锏”——“网络战”也无法打赢中国的情况后，他们选择了“打不过就加入”。而这大家首先不妨从围绕本次美国总统大选及后续发展，西方资本利益内部矛盾随着特朗普的正式上位，不但没有弱化，反而进一步尖锐激化的角度加以观察。

有趣的是，在英伟达、微软和亚马逊纷纷宣布与DeepSeek进行合作的同时，作为曾经OpenAI的投资人、如今“X-AI”掌舵者的马斯克虽然暂时没有在X平台就DeepSeek进行公开点评，但在“别人的评论区”中却频繁挤眉弄眼——只要有人质疑DeepSeek，马斯克就“捧哏”。比如，1月25日，ScaleAI掌门人亚历山大·王在接受媒体采访时说，DeepSeek有5万块英伟达H100芯片，只不过他们不能往外说。对此，马斯克立刻随后评论：“明显就是啊！”。不难看出，马斯克似乎成为了DeepSeek的“头号黑粉儿”。

●“DeepSeek效应”在西方资本利益内部产生的强烈分化作用

需要强调的是，在“DeepSeek效应”出现之初，之所以美国，乃至西方上下一致在第一时间对DeepSeek发难，其中一个主要原因在于，DeepSeek宣布开源（包括商用范畴都可以拿过去用），威胁到了他们所有人。

以微软为例，他们所谓的“商业模式”不能做到开源的（在客观上，这种狭隘的盈利模式已经导致全球科技研发进度被拖慢甚至停滞）。类似“W-i”联盟的核心目的在于通过“讲故事”去“刺激股市”，如果搞成了开源，那还怎么“讲故事”？显然，DeepSeek的成功案例让他们讲的故事穿帮了，于是便对DeepSeek群起而攻之。

DeepSeek的做法让人们立刻想起了微软和安卓之争，其中安卓之所以生出，而微软在手机操作系统上一败涂地，主要原因之一就在于安卓是基于Linux开发的，而Linux就是开源的。在我们看来，DeepSeek宣布开源显然就是要效法Linux——走技术版本的“群众路线”或“统一战线”。对此，很多中小公司表示热烈欢迎。这让诸如微软这样的美国高科技公司，尤其是企图在AI算力模型方面搞垄断的西方高科技公司的处境尴尬。

让诸如微软、英伟达这样的美国高科技公司感到尴尬、甚至感到威胁的除了“DeepSeek效应”外，恐怕还有“非技术因素”，且是主要因素。也就是，在围绕本次美国总统大选及后续发展，西方资本利益内部矛盾随着特朗普的正式上位，不但没有弱化，反而进一步尖锐激化的背景下，微软、英伟达等美国高科技公司为代表的，仍代言西方资本利益的美国资本利益中的一部分，甚至包括西方资本利益中的一部分因“政治因素”，不得不“求生”且寄希望于寻求“外援”。更何况，基于“打不过就加入”的层面，与DeepSeek的合作，并不耽误他们继续赚钱。这一点在西方资本利益内部产生了强烈的分化作用。值得一提的是，在与DeepSeek进行合作的问题上，微软、英伟达、亚马逊等美国高科技公司“有选择”，但马斯克“没有选择”。

这一点大家可以参考《三国演义》中，鲁肃对孙权说过的话：

鲁肃说，刚才他们（张昭等“主降派”）说的话，您可千万不要听，全东吴的人都能投降曹操，但将军您却不能！孙权听了鲁肃的话后，不禁问道：为什么？

鲁肃答道，我们投降了曹操，曹操会根据我们的能力、地位，让我们担任相应的职务，以后做得好，一样会继续升迁，可孙将军您若投降曹操，曹操会怎么安排您呢？也就是说，东吴的大臣们投降曹操还能继续做大臣，可孙权这个“土皇帝”投降了曹操，曹操还会让孙权继续当“土皇帝”吗？答案不言自明。

可以说，马斯克对DeepSeek的攻击，某种意义上很大程度上毁了他的人设，但马斯克和孙权一样，不能选择“投降”。在西方资本利益的圈子中，马斯克是一个“另类”般的存在。其是产业资本和金融资本的结合体，更有“诗与远方”。如果马斯克选择“投降”，就等同于西方资本利益彻底放弃了科技霸权。而构成西方资本利益金融霸权的重要基石之一就是科技霸权。

●如果这一传闻最终被证实，或意味着美国总统特朗普真的急于访华，甚至不排除紧急访华的可能性

在接下来的讨论中，我们不妨就美国内政问题简单补充两句。

当地时间1月29日，一架客机在华盛顿里根国际机场降落过程中，与一架UH-60“黑鹰”直升机相撞后坠河。美航空公司表示，飞机上有60名乘客和4名机组人员，共64人。美国防官员则表示，与客机相撞的“黑鹰”直升机上有3名士兵。

值得注意的是，由于这起发生在当地时间29日的空难所在地点的敏感性（距离美国国会、白宫和五角大楼距离都很近）和匪夷所思，加之美国内政问题波谲云诡，一时间，各种传闻满天飞。

美国防部长海格塞斯1月31日在接受美国媒体采访时表示，失事UH-60“黑鹰”直升机当时正在进行“政府连续性计划”（也称“末日计划”）演习。但他拒绝透露更多细节，称自己不想“涉及任何机密”。

有媒体报道称，这架UH-60“黑鹰”直升机隶属于陆军第12航空营，负责在发生灾难或美国政府遭到袭击时，迅速将美国高级官员输送至安全地点。其中可能涉及美军的一个地下掩体——乌鸦岩山综合体。资料显示，“乌鸦岩山综合体”始建于1950年代，坐落在戴维营的总统官邸附近，被外界描述为“地下五角大楼”，是国防部高级人员在必要时工作的主要后备地点。美国战略与国际研究中心的国防分析师坎西安表示：“政府连续性计划是五角大楼最严密的机密之一。计划内容包括谁将撤离、如何撤离以及撤离到哪里。”

有趣的是，目前阶段，“政府连续性计划”最可能的主要受益者、美国现任总统特朗普似乎对该计划并不了解很多。特朗普本周在被要求就“政府连续性计划”发表评论时表示，“我不知道这指的是什么。”

尽管目前对于这起诡异的空难猜测很多，但有一点可以肯定，那就是与围绕本次美国总统大选及后续发展，西方资本利益内部矛盾随着特朗普的正式上位，不但没有弱化，反而进一步尖锐激化密切相关。而就在这起空难事件发生前几天，美国总统特朗普签署关于公布与美国前总统约翰·肯尼迪、前参议员罗伯特·肯尼迪以及民权领袖马丁·路德·金遇刺事件有关机密文件的行政令。

不难想象的是，在这一背景下，尤其在对肯尼迪遇刺案真相或有“些许了解”的美国前国务卿蓬佩奥随时可能因安全保护待遇被取消出现“意外”的情况下，已经被“揭开盖子”的肯尼迪遇刺案之后续发展已经让美国内部，或者说代言西方资本利益的美国资本利益内部的某些人感到了空前恐惧。这意味着，围绕本次美国总统大选，虽然1月20日已过，特朗普也已顺利上位，但美国内部，或代言西方资本利益的美国资本利益内部的“恶斗”不仅没有减弱，反而更加显得“刀刀见骨”！所以，不排除这起诡异空难事件是对特朗普发出强烈警告的可能性。

无独有偶，俄罗斯国家杜马（议会下院）主席沃洛金当地时间周三（1月29日）对美国记者塔克·卡尔森爆料的“拜登政府曾试图刺杀俄总统普京”一事作出回应，他表示必须对此进行彻底调查。另外，他已就此事于周四（1月30日）向美国国会、联合国提交了相关声明。

报道介绍，本周早些时候卡尔森在其节目中称拜登政府曾“试图刺杀”普京，但他没有给出具体细节以及任何证据支持他的说法，比如这件事可能发生的时间和地点，或关于所谓“刺杀企图”的其他情况。

在我们的观察与评估中，这种消息大概率是“特朗普们”爆料的，如果是这样，说明美国总统特朗普在动用外交资源以寻求“外援”，且首先将其服务于“内部问题”。可以说，特朗普和微软、英伟达、亚马逊等在应对“DeppSeek效应”过程中的心态非常类似，都有强烈的寻求“外援以求生”之意。所以，对国际社会而言，分化他们的最好办法就是瞄准他们眼前的切实利益矛盾点“猛踹”。

透过“DeppSeek效应”，特朗普深刻感受到了美国金融维稳之举步维艰。在美国军事霸权、科技霸权均已摇摇欲坠的情况下。特朗普必须做的一件事就是尽快访华。对于早已是投资货币的美元而言，如果不能讲好“故事”，投资就是0。有趣的是有，恰在此时，有传闻称，Tiktok正准备关闭美国服务，彻底退出美国。如果这一传闻最终被证实，或意味着美国总统特朗普真的急于访华，甚至不排除紧急访华的可能性。

●汉字（汉语）在人工智能的层面，有特殊的优势！DeepSeek的成功已经初步证明了这一点

最后需要补充的是，中国对西方的冲击不仅有“DeepSeek效应”上，还有春晚的“机器人扭秧歌”，其主要看点就是机器人与人之间协同的非常好。如果把这种“人机协调性”应用于军事，也许距离我们看到PLA获得“单兵版‘忠诚僚机’”的出现并不遥远，而这些完全建立在中国完全自主独立研发的软、硬件架构基础之上。

有人说，当今世界的“巅峰对决”，是中国的中国人与美国华裔之间的对决。在我们看来，这种说法有一定道理，显然双方都使用了某种“中国思维”。但两者之间最本质的区别在于，美国的华裔大部分是精致的利己主义者，这意味他们使用的“中国思维”徒有其表，已经脱离了“中国思维”的核心精华。这一点在中国“重庆山火”与美国“加州山火”的对比中中战线的淋漓尽致。且这种对比层级越高，规模越大，就越明显！

值得一提的是，Deepseek的成功，除了团队的智慧与努力之外，一个非常重要的因素是中华文明的基因——Deepseek的算法中，核心是“汉字（汉语）”逻辑：

第一，高信息密度与语义凝练。

汉字以单字为基本单位，兼具象形、会意、形声等多维表意功能(如“明”=日+月)，单个字符即可承载复杂语义，在文本生成、信息压缩任务中效率显著。例如，DeepSeek的模型在生成中文时，能以更短的token长度表达同等信息量，降低算力消耗。此外，汉语的语法灵活性(如省略主语、无时态变化)要求模型更擅长捕捉上下文隐含逻辑。这种特性反而推动了Al在长程依赖建模、语义消歧等核心能力上的进化。

第二，结构化思维的天然映射。

汉字的笔画、部首组合(如“树”=木+对)蕴含层级化特征，与深度学习中的图神经网络(GNN)、注意力机制存在内在契合。DeepSeek在字形生成、手写体识别任务中，已验证了基于汉字结构的预训练方法能提升模型收敛速度。汉语中大量成语(如“刻舟求剑”)将抽象哲理浓缩为四字结构，这种“文化压缩编码”为知识图谱构建和常识推理提供了高价值语料库。

第三，跨模态融合的桥梁作用。

汉字的象形属性(如“山”字形似山峰)使其成为连接视觉与语言的天然媒介。DeepSeek在多模态模型中，通过汉字字形嵌入(Glyph Embedding)技术，让模型同时学习文字符号的视觉特征与语义，显著提升图文互生成任务的准确性。

所以，东方时事解读很多年前就明确强调，汉字（汉语）在人工智能的层面，有特殊的优势！DeepSeek的成功已经初步证明了这一点。最后，期望DeepSeek的AI能与华为为代表的中国软件、硬件系统充分融合，将DeepSeek的进一步发展建立在中国自己的软、硬体系上，这才会前景不可限量！

【相关话题】

第7750期-串讲：特朗普宣誓就职期间美元意外暴跌、特朗普宣布美国退出世卫组织、英伟达机器人业务团队到访北京国地中心（2025-1-21）
第7764期-东方点评，欧央行行长拉加德称，是时候将幻想破灭的美国人才引进到我们这来了（2025-1-26）
第7769期-外媒刷屏“中国AI已追上美国” ，如何看DeepSeek“搅动”硅谷？（2025-1-27）

声明：具体内容如有出入，请以“东方时事解读”音频为准。

Saturday, February 1, 2025, Issue No. 1170

"DeepSeek Effect"It should first be observed from the perspective of the further intensification of internal contradictions among Western capital interests.

[Media Coverage]

On February 1st, Huawei Cloud published an article titled "A War Without Smoke: The 83-Hour Battle to Protect DeepSeek".

In the early morning of January 25th, 2024, DeepSeek's server cluster was maliciously attacked. Top Chinese tech companies united to launch an 83-hour defense battle. Companies including 360, Huawei Cloud, and others participated, each contributing their expertise through technical means to locate the source of the attack, establish protective layers, provide intelligence support, etc. Ultimately, the attack traffic decreased by 97.0%, successfully defending China's AI industry high ground.

【Discussion Summary】

●A "Cyber War" Based on Irreconcilable Contradictions Between China and US Imperialism

On January 27th, due to the impact of DeepSeek, a Chinese AI startup, US AI-themed stocks were sold off, and shares of NVIDIA, the American chip giant, experienced a historic plunge. The Nasdaq Composite Index also fell sharply.

By the close of trading that day, NVIDIA's shares had fallen by 16.97%, wiping out nearly $600 billion in market value in a single day, marking the largest one-day market value loss for any company in US history. NVIDIA's plunge had a ripple effect on the entire market, with Broadcom's shares falling by 17%, AMD's shares dropping by 6%, and Microsoft's shares declining by 2%. Additionally, derivatives in the AI field, such as power suppliers, were also hit hard. Shares of U.S. utility company Vistra fell 29%, and those of another power provider, United Energy, tumbled 21%.

The core reason for this US stock market turmoil was that DeepSeek's latest breakthroughs seriously shook the global hegemony (technology-based system) status of the US tech industry.

At the end of 2024, DeepSeek released its new generation of large language model V3 and announced it would be open-source. Test results showed that this model outperformed mainstream open-source models in multiple evaluations and had cost advantages. In January 2025, DeepSeek released its latest open-source model R1 on the opening day of the World Economic Forum 2025 Annual Meeting, attracting global attention once again.

According to reports, the R1 model achieved significant breakthroughs in technology – using pure deep learning methods to enable AI to spontaneously emerge with reasoning capabilities. In tasks such as mathematics, code, and natural language reasoning, its performance was comparable to OpenAI's o1 model official version, and its training cost was only $5.6 million, far lower than the hundreds of millions or even billions of dollars invested by US tech giants.

After the large-scale plunge in US tech stocks, DeepSeek, which had become popular both domestically and internationally, suffered a massive malicious cyber attack. On January 28th, DeepSeek's official website showed that its online services were subject to a large-scale malicious attack. The cybersecurity company QiAnXin stated that the IP addresses of the network attacks on DeepSeek were all located in the United States.

They mobilized the whole country and spared no efforts. We also mobilized the whole country. In our observation and assessment, this is a typical cyber war, and this war is inevitable. It is also one of the specific manifestations of the irreconcilable contradictions between China and the US, or more accurately, between China and US imperialism. It is worth mentioning that in this "DeepSeek defense battle", the "underlying (software and hardware) protection" provided by Huawei played a core and crucial role.

Furthermore, we welcome DeepSeek's decision to comprehensively cooperate with Huawei.

●NVIDIA, Microsoft, and Amazon's "If You Can't Beat Them, Join Them" Approach, and Musk as DeepSeek's "Top Hater"

It is worth noting that a significant portion of the hundreds of millions or even billions of dollars invested by US tech giants was focused on hardware, specifically NVIDIA's computing power cards. This caused the myth spun by NVIDIA through "selling shovels" to collapse, leading to the plunge in NVIDIA's stock. Additionally, the myth woven by OpenAI in algorithms also collapsed, as people realized that OpenAI had previously been just "telling stories". The sharp decline in the stocks of NVIDIA and OpenAI is a microcosm of the plunge in the US stock market under the impact of the "DeepSeek effect", and the impact of this US stock market turmoil is so significant that it has affected the US economy, especially financial stability. This is unacceptable to US President Trump, who rang the opening bell for the NYSE and publicly stated that "the stock market is everything". This may be the main reason why the US, at the initial manifestation of the "DeepSeek effect", mobilized the whole country to strangle DeepSeek.

Notably, after the "DeepSeek defense battle" concluded with a 97.0% reduction in attack traffic, marking a complete victory for China, attitudes towards DeepSeek within the US and the entire West have significantly diverged:

First, NVIDIA, whose stock plummeted due to the "DeepSeek effect", announced on January 31st that the DeepSeek-R1 model had been released as a preview of NVIDIA NIM microservices on NVIDIA's developer-facing website. NVIDIA's website describes the DeepSeek-R1 model as the most advanced and efficient large language model, excelling in reasoning, mathematics, and coding. On January 30th, Microsoft, a close partner of the US-based OpenAI research center, announced that it had officially integrated DeepSeek-R1 into Azure AI Foundry, making it part of the enterprise-level AI service platform.

Second, also on January 31st, Microsoft, as the largest shareholder of OpenAI, decided to incorporate the DeepSeek-R1 model into its AI services. Microsoft emphasized that the DeepSeek-R1 model had passed "rigorous red team testing and security assessments" and undergone "automated detection of model behavior and extensive security reviews to mitigate potential risks". Microsoft also mentioned in its statement that it would continue to evaluate the DeepSeek-R1 model and might make certain adjustments and optimizations to enhance its accuracy and review mechanisms.

Third, again on January 31st, Amazon stated that the DeepSeek-R1 model is now available on Amazon Web Services. Even the company's CEO, Andy Jassy, told users to "feel free to use it".

In our observation and assessment, behind these world-renowned US tech companies' "sudden" willingness to cooperate with DeepSeek, is the fact that after personally experiencing their last "trump card" – "cyber warfare" – also failing to defeat China, they chose to "if you can't beat them, join them". This can first be observed from the perspective of the further intensification of internal contradictions among Western capital interests with the official ascension of Trump in the context of the current US presidential election and subsequent developments.

Interestingly, while NVIDIA, Microsoft, and Amazon have announced their cooperation with DeepSeek, Musk, the former investor of OpenAI and current head of "X-AI", although he has not publicly commented on DeepSeek on the X platform, has frequently made sarcastic remarks in "other people's comment sections" – whenever someone questions DeepSeek, Musk "plays along". For example, on January 25th, when Alexander Wang, the head of ScaleAI, said in an interview with the media that DeepSeek had 50,000 NVIDIA H100 chips but couldn't disclose it, Musk immediately commented, "Obviously!". It is not difficult to see that Musk seems to have become DeepSeek's "top hater".

● The Profound Divisive Effect of the "DeepSeek Effect" Within Western Capital Interests

It bears emphasizing that one of the main reasons why the United States, and indeed the West as a whole, unanimously confronted DeepSeek at the first opportunity when the "DeepSeek Effect" emerged was because DeepSeek announced its open-source nature (including for commercial use), posing a threat to all of them.

Take Microsoft as an example. Their so-called "business model" does not accommodate open-source practices (objectively, this narrow profit model has already slowed or even stalled global technological research and development progress). The core purpose of alliances like the "W-i" is to "stimulate the stock market" through "storytelling." If everything becomes open-source, how can they continue to "tell stories"? Clearly, DeepSeek's success stories have exposed the falsehoods in their narratives, prompting a concerted attack against it.

DeepSeek's approach immediately reminds people of the competition between Microsoft and Android. One of the main reasons why Android emerged and Microsoft failed miserably in the mobile operating system market is that Android is based on Linux, which is open-source. In our view, DeepSeek's announcement of open-sourcing is clearly an attempt to emulate Linux—taking the "mass line" or "united front" approach in technology development. Many small and medium-sized companies have warmly welcomed this move. This has put U.S. high-tech companies like Microsoft, especially those in the West aiming to monopolize AI computing models, in an awkward position.

Apart from the "DeepSeek Effect," what makes U.S. high-tech companies like Microsoft and NVIDIA feel awkward, or even threatened, are perhaps the "non-technical factors," which are the main factors. That is, amid the current U.S. presidential election and its aftermath, with the internal contradictions within Western capital interests becoming increasingly acute instead of weakening following Trump's ascension, some U.S. high-tech companies, represented by Microsoft and NVIDIA, which still represent a portion of Western capital interests, and even some within Western capital interests due to "political factors," are forced to "seek survival" and pin their hopes on seeking "external assistance." Furthermore, based on the principle of "if you can't beat them, join them," cooperating with DeepSeek does not prevent them from continuing to make money. This has had a profound divisive effect within Western capital interests. It's worth mentioning that while U.S. high-tech companies like Microsoft, NVIDIA, and Amazon have "choices" regarding cooperation with DeepSeek, Musk "has no choice."

This can be likened to what Lu Su said to Sun Quan in "Romance of the Three Kingdoms":

Lu Su said, "You must not listen to what they (the "surrender faction" led by Zhang Zhao) just said. Everyone in all of Wu can surrender to Cao Cao, but you, General, cannot! Sun Quan asked, "Why?" Lu Su replied, "If we surrender to Cao Cao, he will assign us positions based on our abilities and status. If we perform well in the future, we will continue to be promoted. But if you, General Sun, surrender to Cao Cao, how will he treat you? In other words, the ministers of Wu who surrender to Cao Cao can continue to serve as ministers, but if Sun Quan, the 'local ruler,' surrenders to Cao Cao, will Cao Cao allow you to continue as the 'local ruler'?" The answer is self-evident.

It can be said that Musk's attack on DeepSeek has, to some extent, tarnished his reputation. However, like Sun Quan, Musk cannot choose to "surrender." Within the circle of Western capital interests, Musk is an "outlier." He is a combination of industrial and financial capital, with a vision for the future. If Musk chooses to "surrender," it would be tantamount to Western capital interests completely abandoning their technological hegemony. And technological hegemony is one of the important cornerstones of Western capital interests' financial hegemony.

●If this rumor is eventually confirmed, it could imply that U.S. President Trump is really eager to visit China, and even the possibility of an emergency visit cannot be ruled out.

In the following discussion, let's briefly add a few words on U.S. domestic politics.

On January 29, local time, a passenger plane crashed into a river after colliding with a UH-60 "Black Hawk" helicopter during the landing process at Ronald Reagan Washington National Airport. The airline stated that there were 60 passengers and 4 crew members on board, totaling 64 people. A U.S. defense official said that there were 3 soldiers on the "Black Hawk" helicopter that collided with the passenger plane.

It's worth noting that due to the sensitivity of the location of this air disaster on January 29 (close to the U.S. Capitol, the White House, and the Pentagon) and its bizarre nature, coupled with the turbulent domestic politics in the United States, rumors have been flying everywhere.

U.S. Defense Secretary Hagel said in an interview with U.S. media on January 31 that the crashed UH-60 "Black Hawk" helicopter was conducting a "Continuity of Government" (also known as the "Doomsday Plan") exercise. However, he refused to disclose more details, saying he did not want to "get into any classified information."

According to media reports, the UH-60 "Black Hawk" helicopter belonged to the 12th Aviation Battalion of the Army and was responsible for rapidly transporting senior U.S. officials to safe locations in the event of a disaster or an attack on the U.S. government. This may involve a U.S. military underground bunker - the Raven Rock Mountain Complex. According to records, the "Raven Rock Mountain Complex" was built in the 1950s and is located near the presidential mansion at Camp David. It is described as the "underground Pentagon" and is the primary backup location for senior Defense Department personnel to work when necessary. Defense analyst Kancian from the Center for Strategic and International Studies said, "The Continuity of Government plan is one of the Pentagon's most tightly guarded secrets. The plan includes who will evacuate, how they will evacuate, and where they will evacuate to."

Interestingly, at this stage, President Trump, who is arguably the main beneficiary of the "Continuity of Government" plan, seems to know very little about it. When asked to comment on the "Continuity of Government" plan this week, Trump said, "I don't know what that refers to."

Although there are many speculations about this bizarre air disaster, one thing is certain: it is closely related to the fact that the internal conflicts within Western capital interests have not weakened but have further intensified with Trump's official ascension to power, surrounding the U.S. presidential election and subsequent developments. And just a few days before this air disaster, U.S. President Trump signed an executive order to disclose classified documents related to the assassinations of former U.S. President John F. Kennedy, former Senator Robert F. Kennedy, and civil rights leader Martin Luther King Jr.

It is not hard to imagine that against this backdrop, especially with former U.S. Secretary of State Pompeo, who may have "some understanding" of the truth behind the Kennedy assassination, potentially facing an "accident" due to the revocation of his security protection, the subsequent developments of the uncovered Kennedy assassination have made certain individuals within the United States, or those representing Western capital interests within the U.S. capital interests, feel unprecedentedly fearful. This means that although January 20 has passed and Trump has smoothly taken office, the "fierce battles" within the United States, or within the U.S. capital interests representing Western capital interests, have not weakened but have become even more "cutthroat"! Therefore, the possibility cannot be ruled out that this bizarre air disaster was a strong warning to Trump.

Similarly, Volodin, the Chairman of the Russian State Duma (lower house of parliament), responded on Wednesday (January 29) to U.S. journalist Tucker Carlson's revelation that the Biden administration had attempted to assassinate Russian President Putin. He stated that a thorough investigation must be conducted. In addition, he submitted relevant statements to the U.S. Congress and the United Nations on Thursday (January 30).

According to reports, earlier this week, Carlson claimed on his show that the Biden administration had "attempted to assassinate" Putin, but he did not provide specific details or any evidence to support his claim, such as the time and location of the alleged "assassination attempt" or other circumstances regarding the so-called "attempted assassination."

In our observation and assessment, this kind of news is most likely leaked by "Trump and his allies." If this is the case, it indicates that U.S. President Trump is using diplomatic resources to seek "external assistance" and primarily using it to address "internal issues." It can be said that Trump, as well as Microsoft, NVIDIA, Amazon, and others, share a similar mindset in dealing with the "DeppSeek effect," all having a strong desire to seek "external assistance to survive." Therefore, for the international community, the best way to divide them is to "kick hard" at their immediate and tangible conflicting interests.

Through the "DeppSeek effect," Trump has deeply felt the precariousness of U.S. financial stability. With U.S. military and technological hegemony both teetering, one thing Trump must do is to visit China as soon as possible. For the U.S. dollar, which has long been an investment currency, if it cannot tell a good "story," investment is zero. Interestingly, at this moment, there are rumors that TikTok is preparing to shut down its U.S. services and completely withdraw from the United States. If this rumor is eventually confirmed, it could imply that U.S. President Trump is really eager to visit China, and even the possibility of an emergency visit cannot be ruled out.

●Chinese characters (and the Chinese language) have unique advantages at the level of artificial intelligence! The success of DeepSeek has preliminarily proven this point.

Lastly, it should be added that China's impact on the West is not only seen in the "DeepSeek effect," but also in the "robot twisting yangko" at the Spring Festival Gala, where the main highlight was the excellent coordination between robots and humans. If this "human-machine coordination" is applied to the military, perhaps we are not far from seeing the PLA acquire a "single-soldier version of the 'loyal wingman'," and all of this is built on a software and hardware architecture that is fully independently developed by China.

Some say that the "ultimate showdown" in today's world is between Chinese people in China and Chinese Americans. In our view, there is some truth to this statement, as both sides clearly use a certain kind of "Chinese thinking." However, the most fundamental difference between the two lies in the fact that most Chinese Americans are exquisite egoists, meaning that the "Chinese thinking" they employ is superficial and has deviated from the core essence of "Chinese thinking." This point is vividly demonstrated in the comparison between the "Chongqing wildfire" in China and the "California wildfire" in the United States, and the higher the level and the larger the scale of the comparison, the more evident it becomes!

It is worth mentioning that in addition to the wisdom and hard work of the team, a very important factor in the success of DeepSeek is the gene of Chinese civilization – the core of DeepSeek's algorithm lies in the logic of "Chinese characters (and the Chinese language)":

Firstly, high information density and semantic condensation.

Chinese characters take single characters as the basic unit, combining pictographic, ideographic, and phonetic functions (e.g., "明" = sun + moon), with a single character able to carry complex semantics, making it highly efficient in text generation and information compression tasks. For example, when generating Chinese text, DeepSeek's model can express the same amount of information with shorter token lengths, reducing computational power consumption. Furthermore, the flexibility of Chinese grammar (e.g., omitting subjects, no tense changes) requires the model to be better at capturing the implicit logic in the context. This characteristic has instead promoted the evolution of AI in core capabilities such as long-range dependency modeling and semantic disambiguation.

Secondly, the natural mapping of structured thinking.

The stroke and radical combinations of Chinese characters (e.g., "树" = wood + pair) contain hierarchical features, which have an inherent fit with graph neural networks (GNNs) and attention mechanisms in deep learning. In tasks such as character shape generation and handwritten character recognition, DeepSeek has verified that pre-training methods based on the structure of Chinese characters can improve model convergence speed. A large number of idioms in Chinese (e.g., "carving a boat to find a sword") condense abstract philosophies into four-character structures, and this "cultural compression coding" provides a high-value corpus for knowledge graph construction and common sense reasoning.

Thirdly, the bridging role in cross-modal fusion.

The pictographic nature of Chinese characters (e.g., the character "mountain" resembles a mountain peak) makes them a natural medium connecting vision and language. In multimodal models, DeepSeek uses glyph embedding technology to enable the model to learn both the visual features and semantics of written symbols simultaneously, significantly improving the accuracy of image-text generation tasks.

Therefore, East News Analysis has emphasized for many years that Chinese characters (and the Chinese language) have unique advantages at the level of artificial intelligence! The success of DeepSeek has preliminarily proven this point. Finally, we hope that DeepSeek's AI can fully integrate with China's software and hardware systems represented by Huawei, and that the further development of DeepSeek can be built on China's own software and hardware system, which will have unlimited prospects!

[Related Topics]

Episode 7750 - Recap: The US dollar unexpectedly plummets during Trump's inauguration, Trump announces US withdrawal from WHO, NVIDIA's robotics business team visits Beijing Guodi Center (January 21, 2025)
Episode 7764 - East News Commentary: ECB President Lagarde says it's time to bring disillusioned American talent to us (January 26, 2025)
Episode 7769 - Foreign media flood with "China's AI has caught up with the US," how to view DeepSeek "stirring up" Silicon Valley? (January 27, 2025)

Disclaimer: In case of any discrepancies in the specific content, please refer to the 'Eastern Current Affairs Interpretation Audio' for the most accurate information.

原文作者公众号：

广州市贯日翻译服务有限公司为东方时评-衍射传媒/衍射咨询提供翻译支持

翻译请联系http://www.en-ch.com/chcontact.htm

手机微信13924166640

广州市越秀区环市东路世界贸易中心大厦南塔24楼 020-86266990