中國下令撤銷Meta對人工智能新創公司的收購

Recently the New York Times reported the following:

China Orders the Unwinding of Meta’s Acquisition of an A.I. Start-Up

The impact of the ruling was not immediately clear, but it could send a chilling signal to Chinese tech founders seeking to team up with foreign companies.

NYT - By Meaghan Tobin - Reporting from Taipei, Taiwan

April 27, 2026

Updated 10:08 a.m. ET

The Chinese government said on Monday that it would require the unwinding of Meta’s acquisition of Manus, a Singapore-based artificial intelligence company with Chinese founders, in a move that could chill other Chinese entrepreneurs from seeking tie-ups with foreign partners.

Chinese officials had said in January that they were investigating whether Meta’s acquisition of Manus in December violated the country’s rules on foreign investment. They were also assessing whether the deal violated China’s requirements that companies obtain approval for the export of certain technologies.

The National Development and Reform Commission, a high-level ministry that oversees economic planning and plays a central role in setting China’s A.I. policy, said on Monday that it had decided to prohibit foreign investment in Manus, and instructed the parties involved to withdraw the acquisition.

It is not clear how such a transaction would be unwound. Meta has described the two teams as “deeply integrated.” Members of the Manus team have been working alongside Meta colleagues at the company’s office in Singapore, according to two people familiar with the operation who spoke on the condition of anonymity because they were not authorized to talk publicly.

In a statement, Meta said the transaction fully complied with law, adding “We anticipate an appropriate resolution to the inquiry.”

The Chinese government issued its decision just a few weeks before a planned meeting between President Trump and China’s leader, Xi Jinping.

The New York Times reported last month that officials from the Chinese agency had called in Meta and Manus executives to express concerns about the deal, and that Manus executives had been restricted from departing China, as part of an apparent effort to discourage Chinese A.I. executives from moving businesses offshore.

As companies in China and the United States race to develop cutting-edge artificial intelligence, the scrutiny could make it harder for other Chinese firms to attract funding from foreign investors. It could also signal to Chinese researchers not to follow the path Manus took, in which Chinese executives register companies outside China to sidestep regulations from both Washington and Beijing.

Manus is based in Singapore but was founded by Chinese engineers and had a Chinese parent company. The company was incorporated offshore and set up in China as a foreign-owned entity; it has affiliated offices in Beijing and Wuhan, China.

Many Chinese tech founders hope to attract Silicon Valley investors. But in recent years, they have increasingly found themselves needing to choose between targeting the Chinese market or moving their headquarters outside China to court foreign investors.

Jianggan Li, chief executive of Momentum Works, a consultancy in Singapore, said scrutiny like the Manus deal was facing “will make it increasingly hard for Chinese A.I. founders who started in China to sit on both sides or switch to the other side.”

“There are already a lot of uncertainties starting an A.I. start-up, and most founders are technologists but not politically savvy,” Mr. Li said.

Meta has been spending billions on A.I. researchers and data centers, and its acquisition of Manus formed a rare direct link between talent from both the United States and China.

In recent years, Chinese companies have made up a large share of Meta’s advertising revenue. Meta said on a call with analysts in 2024 that Chinese-based advertisers accounted for 10 percent of its revenue, almost double the amount two years earlier. China’s start-ups that offer games, short video apps and e-commerce have flooded Facebook and Instagram with ads as they look to establish a presence outside China.

Translation

中國下令撤銷Meta對人工智能新創公司的收購

該裁決的影響尚不明朗，但可能會對尋求與外國公司合作的中國科技創辦人發出寒蟬效應

中國政府週一表示，將要求Meta撤銷對新加坡人工智能公司Manus的收購。 Manus的創始人均為中國人。此舉可能會對其他中國企業家尋求與外國夥伴合作產生負面影響。

中國官員曾在1月表示，他們正在調查Meta去年12月對Manus的收購是否違反了中國的外商投資規定。他們也正在評估該交易是否違反了中國關於企業出口某些技術需獲得批准的規定。

國家發展和改革委員會是負責經濟規劃並在製定中國人工智能政策方面發揮核心作用的高級別部門，該委員會週一表示，已決定禁止外商投資Manus公司，並指示相關各方撤回收購。

目前尚不清楚這筆交易將如何解除。 Meta公司稱兩支團隊「深度融合」。據兩位知情人士透露，Manus團隊成員一直在Meta公司位於新加坡的辦公室與Meta的同事一起工作。這兩位知情人士因未獲授權公開談論此事而要求匿名。

Meta公司在一份聲明中表示，該交易完全符合法律規定，並補充說： “我們期待調查得到妥善解決。”

中國政府做出這項決定時，距離美國總統特朗普與中國國家主席習近平計劃會晤僅幾星期。

《紐約時報》上月報道稱，中國監管機構曾召見Meta和Manus的高層，表達對該交易的擔憂，並限制Manus高層離境，這顯然是為了阻止中國人工智能企業高管將業務轉移到海外。

隨著中美兩國公司競相開發尖端人工智能技術，這種審查可能使其他中國公司更難吸引外國投資者的資金。這也可能向中國研究人員發出警告，告誡他們不要效法Manus的做法 - 即中國高階主管在中國境外註冊公司，以避免華盛頓和北京的監管。

Manus總部位於新加坡，但由中國工程師創立，並有一家中國母公司。該公司在境外註冊成立，並在中國設立了外資企業；其在北京和武漢設有分支機構。

許多中國科技公司創辦人希望吸引矽谷投資者。但近年來，他們越來越需要在瞄準中國市場，或將總部遷至海外以吸引外國投資者之間做出選擇。

新加坡顧問公司 Momentum Works 的執行長Jianggan Li表示，像 Manus 收購案這樣受到的審查「將使在中國起步的人工智能創辦人越來越難以坐在兩邊，或者轉投他方」。

Li先生說： “創辦人工智能新創公司本身就存在諸多不確定性，而且大多數創始人都是技術專家，但不擅長政治。”

Meta 公司一直在人工智能研究人員和資料中心方面投入數十億美元，其對 Manus 的收購在美中兩國人才之間建立了一種罕見的直接聯繫。

近年來，中國公司在 Meta 的廣告收入中佔了很大份額。 Meta 在 2024 年與分析師的電話會議上表示，中國廣告商貢獻了 10% 的收入，幾乎是兩年之前的兩倍。中國的遊戲、短影片應用程式和電商新創公司紛紛在 Facebook 和 Instagram 上投放廣告，以期在中國以外地區拓展業務。

So, China has decided to prohibit foreign investment in Manus. Meta describes its team has “deeply integrated” with Manus. Many China’s start-ups are currently offering games, short video apps and e-commerce that flood Facebook and Instagram with ads as they try to establish a presence outside China. Apparent, this decision to prohibit Meta’s investment could chill other Chinese entrepreneurs from seeking tie-ups with foreign partners.

DeepSeek 的後續是將進一步擴大中國在開源人工智能領域的影響力 (2/2)

Recently the New York Times reported the following:

DeepSeek’s Sequel Set to Extend China’s Reach in Open-Source A.I. (2/2)

Chinese companies have embraced making their most advanced artificial intelligence models available to all.

The NYT - By Meaghan Tobin and Cade Metz - Meaghan Tobin reported from Taipei, Taiwan, and Cade Metz from San Francisco.

April 24, 2026

Updated 11:47 a.m. ET

(continue)

DeepSeek’s open-source models are central to this strategy. While many Western companies guard their most valuable models, China has embraced open source, and almost all of its top-performing systems are widely available.

Even so, Chinese A.I. firms face major hurdles. Three U.S. administrations have imposed export controls limiting access to advanced chips needed for cutting-edge A.I. systems. And firms in Silicon Valley continue to outspend Chinese rivals in the race for top A.I. talent.

China’s push into open-source A.I. has become a major economic advantage at home, according to a new study by a U.S. congressional advisory body. With few barriers to use, the systems have spread across industries such as robotics, logistics and manufacturing. The study found that these industrial applications generate real-world data that are used to improve A.I. systems.

This approach has allowed Chinese tech firms to capture global influence, as programmers and engineers around the world adopt their systems to build new products.

From Lagos to Kuala Lumpur, developers on tight budgets are turning to Chinese open-source models because they are cheaper to run and therefore easier to experiment with. Last May, Malaysia’s deputy minister of communications said the country’s sovereign A.I. infrastructure would be built on DeepSeek’s technology.

Chinese open-source models accounted for roughly one-third of global A.I. use last year, according to a study by OpenRouter, an A.I. model marketplace. DeepSeek was the most widely used, followed by models from Alibaba, the Chinese internet company.

That reflects a broader strategy. As Chinese companies expand abroad, making their systems open source helps them gain traction with coders by offering cheaper, more accessible tools.

“Open source is the soft power of technology of the future,” said Kevin Xu, the U.S.-based founder of Interconnected Capital, a hedge fund that invests in artificial intelligence technologies. Mr. Xu and his fund do not invest in DeepSeek.

Wei Sun, a principal analyst in A.I. at Counterpoint Research in Beijing, said DeepSeek’s success paved the way for China’s tech giants to release their A.I. systems publicly rather than closely guarding them.

Alibaba has since emerged as a leader. Its Qwen model family has surpassed one billion downloads. ByteDance, parent company of TikTok, has also shared some details of its technology after spending $11 billion on A.I. infrastructure in 2024.

“The A.I. generation of open-source builders from China was arguably the biggest A.I. story in 2025,” Mr. Xu said. “The progress of the models, the pace of the releases and the number of A.I. labs that both compete with each other but also seem to cheer each other on came fast and furious with no signs of slowing down.”

Translation

DeepSeek 的後續是將進一步擴大中國在開源人工智能領域的影響力 (2/2)

中國企業已開始積極向所有人開放其最先進的人工智能模型

（繼續）

DeepSeek 的開源模型是這項策略的核心。許多西方公司對其最有價值的模型嚴加保護，而中國則擁抱開源，幾乎所有性能卓越的系統都已廣泛開放。

即便如此，中國的人工智能公司仍面臨巨大的挑戰。美國三屆政府都實施了出口管制，限制取得尖端人工智能系統所需的高階晶片。此外，矽谷的公司在爭取頂尖人工智能人才方面的持續投入巨大資金，遠超中國競爭對手。

根據美國國會諮詢機構的一項最新研究，中國大力發展開源人工智能已成為其國內的一項重大經濟優勢。由於使用門檻低，這些系統已廣泛應用於機械人、物流和製造業等各個行業。研究發現，這些工業應用能夠產生真實世界的數據，用於改善人工智能系統。

這種模式使中國科技公司得以在全球擴大影響力，世界各地的程式設計師和工程師都在採用他們的系統來開發新產品。

從Lagos到吉隆坡，預算緊拙的開發者紛紛轉向中國開源模型，因為它們運作成本更低，因此也更容易進行實驗。去年五月，馬來西亞通訊部副部長表示，該國的自主人工智能基礎設施將基於DeepSeek的技術建構。

根據人工智能模型市場OpenRouter的一項研究顯示，去年中國開源模型約佔全球人工智能使用量的三分之一。 DeepSeek 的應用最為廣泛，其次是阿里巴巴，一間中國網路公司的模型。

這反映了一種更廣泛的策略。隨著中國公司向海外擴張，開源系統有助於它們透過提供更便宜、更易於使用的工具來吸引程式設計師。

總部位於美國的Interconnected Capital的創始人Kevin Xu表示:「開源是未來科技的軟實力」。Interconnected Capital是一家投資人工智能技術的對沖基金。Xu及其基金並無未投資 DeepSeek。

北京 Counterpoint Research 的人工智能首席分析師Wei Sun表示，DeepSeek 的成功為中國科技巨頭公開其人工智能系統鋪平了道路，而不是將其嚴加保密。

阿里巴巴此後崛起成為領導者，成為業界領導者。其 Qwen 模型系列的下載量已超過 10 億次。 TikTok 的母公司字節跳動在斥資 110 億美元用於人工智能基礎建設後，在 2024 年也分享了其技術的部分細節。

Xu先生表示： “中國的人工智能世代開源開發者可以說是 2025 年人工智能領域最大的故事”; “模型的進步、發佈速度以及人工智能實驗室的數量都絲毫沒有放緩的跡象，這些實驗室既快速又熱烈相互競爭，亦互相鼓勵。”

So, DeepSeek has released a preview of V4 and intends it to be open source. China’s push into open-source A.I. has become a major economic advantage. With few barriers to use, the systems have spread across industries. This approach will allow Chinese tech firms to gain global influence. Apparently, open source will be the soft power of technology in the future. I am wondering how will the US AI companies respond to this open-source challenge.

DeepSeek的後續是將進一步擴大中國在開源人工智能領域的影響力（1/2）

Recently the New York Times reported the following:

DeepSeek’s Sequel Set to Extend China’s Reach in Open-Source A.I. (1/2)

Chinese companies have embraced making their most advanced artificial intelligence models available to all.

The NYT - By Meaghan Tobin and Cade Metz - Meaghan Tobin reported from Taipei, Taiwan, and Cade Metz from San Francisco.

April 24, 2026

Updated 11:47 a.m. ET

When the Chinese start-up DeepSeek published details about one of its artificial intelligence models last year, it sent shock waves through the tech industry.

The company said it had built its system by spending far less on computer chips than American rivals like OpenAI and Anthropic. It marked the start of what became known as China’s “DeepSeek moment,” shorthand for the belief that Chinese A.I. companies were ready to showcase their technical capabilities to the world.

The DeepSeek moment reflected a shift in the global A.I. landscape. The change was about not only lower costs but also openness in how the technology is shared.

DeepSeek released its models as open source, which means others can freely use and modify them. By contrast, OpenAI and Anthropic kept their leading models proprietary. The episode demonstrated that an open-source system could perform almost as well as closed versions. In the months that followed, Chinese firms released dozens of other open-source models. By the end of 2025, these models made up a significant share of global A.I. use.

On Friday, DeepSeek released a preview of V4, its long-awaited follow-up model, which it intends to open source. The new model excels at writing computer code, an increasingly important skill for leading A.I. systems. It significantly outperformed every other open-source system at generating code, according to tests from Vals AI, a company that tracks the performance of A.I. technologies. China’s push into open-source A.I. has become a major economic advantage at home, according to a new study by a U.S. congressional advisory body. With few barriers to use, the systems have spread across industries such as robotics, logistics and manufacturing. The study found that these industrial applications generate real-world data that are used to improve A.I. systems. This approach has allowed Chinese tech firms to capture global influence, as programmers and engineers around the world adopt their systems to build new products.

DeepSeek released its new model just days after Moonshot AI, another Chinese start-up, introduced its latest open-source model, Kimi 2.6. While these systems trail the coding capabilities of the leading U.S. models from Anthropic and OpenAI, the gap is narrowing.

The implications are meaningful. Using A.I. to write code is faster and frees up human programmers to focus on bigger issues. It also means people can use DeepSeek’s latest release to power A.I. agents, which are personal digital assistants that can use other software applications on behalf of office workers, including spreadsheets, online calendars and email services.

As A.I. systems improve at writing computer code, they are also getting better at finding security vulnerabilities in software — a skill that is fundamentally changing cybersecurity. That means tools like DeepSeek’s can be used to both attack and defend computer networks.

Across tasks, DeepSeek V4 is on a par with Moonshot’s latest model. “They are basically neck and neck,” said Rayan Krishnan, the chief executive of Vals AI.

In the months leading up to DeepSeek’s latest release, foreign rivals moved to pre-empt another round of glowing headlines. Silicon Valley’s A.I. giants, Anthropic and OpenAI, said DeepSeek had unfairly piggybacked on their technology through distillation, a process in which engineers mimic a rival model by querying it millions of times and copying its behavior.

The competition to build the best-performing A.I. systems has transformed into a geopolitical power struggle. While Silicon Valley leaders at Anthropic and OpenAI warn that their technology would be dangerous in the hands of autocratic countries, China has invested billions to become an A.I. superpower, viewing the technology as a critical engine of economic growth.

(to be continued)

Translation

DeepSeek的後續是將進一步擴大中國在開源人工智能領域的影響力（1/2）

中國企業已開始積極向全世界開放其最先進的人工智能模型

去年，中國新創公司DeepSeek公佈了其人工智能模型之一的詳細信息，這在科技行業引起了巨大震動。

該公司表示，其係統在晶片上的投入遠低於OpenAI和Anthropic等美國競爭對手。這標誌著中國「DeepSeek時刻」的開始，這一時刻象徵著中國人工智能公司已準備好向世界展示其技術實力。

“DeepSeek時刻 ” 反映了全球人工智能格局的轉變。這項轉變不僅關乎降低成本，更關乎技術共享方式的開放性。

DeepSeek 將其模型開源，這意味著其他人可以自由使用和修改它們。相較之下，OpenAI 和 Anthropic 則將其領先的模式設為專有模式。這一事件表明，開源系統幾乎可以達到與封閉版本相同的效能。在接下來的幾個月裡，中國公司發布了數十個其他開源模型。到 2025 年底，這些模型在全球人工智能應用中佔據了相當大的份額。

上週五，DeepSeek 發佈了其期待已久的後續模型 V4 的預覽版，並計劃將其開源。新模型在編寫電腦程式碼方面表現出色，這對於領先的人工智能系統而言是一項日益重要的技能。根據追蹤人工智能技術性能的 Vals AI 公司的測試，V4 在程式碼生成方面顯著優於所有其他開源系統。

就在另一家中國新創公司Moonshot AI發佈其最新開源模型Kimi 2.6幾天後，DeepSeek也發佈了其新模型。雖然這些系統的編碼能力與美國領先的Anthropic和OpenAI模型相比仍有落後，但差距正在縮小。

發佈有重大意義。使用人工智能編寫程式碼速度更快，能夠使程式設計師騰出精力專注於更重要的問題。這也意味著人們可以使用DeepSeek的最新版本來驅動人工智能代理，這些代理是個人電子助理，可以代表辦公室員工使用其他軟體應用程序，包括電子表格、線上日曆和電子郵件服務。

隨著人工智能系統編寫程式碼能力的提升，它們在發現軟體安全漏洞方面也越來越出色 - 這項技能正在從根本上改變網路安全。這意味著像DeepSeek這樣的工具既可以用於攻擊電腦網絡，也可以用於防禦電腦網路。

在各項任務中，DeepSeek V4的表現與Moonshot的最新模型平起平坐。 Vals AI 的執行長 Rayan Krishnan 表示: 「他們基本上不分伯仲」。

在 DeepSeek 最新版本發佈前的幾個月裡，外國競爭對手試圖搶先一步，去阻止另一輪的引人注目的報道。矽谷的人工智能巨頭 Anthropic 和 OpenAI 聲稱，DeepSeek 透過「蒸餾」技術不公平地搭了他們的順風車。蒸餾是一種工程師透過數百萬次查詢競爭對手的模型並複製其行為來模擬其運行過程的方法。

建構最佳人工智能系統的競爭已經演變成一場地緣政治權力鬥爭。儘管 Anthropic 和 OpenAI 的矽谷領導人警告說，他們的技術落入專制國家手中將十分危險，但中國已投入數十億美元力圖成為人工智能超級大國，並將這項技術視為經濟成長的關鍵引擎。

（待續）

Note:

1.  Distillation (蒸餾) is basically a process in which engineers mimic a rival model by querying it millions of times and copying its behavior. In AI development, distillation (often called knowledge distillation) is a technique where a smaller AI model learns to imitate a larger, more capable model. The basic idea is that: A large, expensive model (“teacher”) generates outputs, while a smaller model (“student”) trains on those outputs. The student learns patterns, reasoning styles, and behaviors from the teacher. The result is a model that is faster and cheaper to run while keeping much of the teacher’s performance. (ChatGPT)

中國如何打造晶片產業，為何仍有所不足（2/2）

Recently the New York Times reported the following:

How China Built a Chip Industry, and Why It’s Still Not Enough (2/2)

More than a decade into Beijing’s push for self sufficiency, Chinese firms are producing fewer, lower-performing chips than their foreign competitors.

The NYT - By Meaghan Tobin reporting from Taipei, Taiwan ; Xinyun Wu contributed reporting from Taipei. Meaghan Tobin covers business and tech stories in Asia with a focus on China and is based in Taipei.

Feb. 14, 2026, 12:00 a.m. ET

(continue)

Huawei’s Pivot

In 2014, China was the world’s largest market for semiconductors. But 90 percent of the chips its companies used were made outside the country.

Concerned about that dependency, the State Council, China’s top governing body, approved a plan to spend billions and made a vow: China would be making every part of its semiconductor supply chain at home by 2030.

Policymakers had reason to be concerned about the risks that foreign technology posed to Chinese infrastructure. Earlier that year, documents provided by the former National Security Agency contractor Edward J. Snowden had disclosed that the U.S. government had monitored the communications of top executives at Huawei.

Then in 2017, President Trump fined the Chinese telecommunications giant ZTE for allegedly violating U.S. sanctions on Iran, crippling its business overnight. Although ZTE does not manufacture chips, the action gave China another lesson in its need for self reliance.

Next came Huawei. The first Trump administration embarked on a global campaign to get countries to stop using Huawei’s equipment in their telecommunications infrastructure. Huawei responded by offloading that business line and getting in step with Beijing’s self-sufficiency program.

“Huawei was unique in its capabilities and its alignment with China’s national goals,” said Kyle Chan, a fellow at the Brookings Institution who studies Chinese industrial policy. “Huawei’s experience was a microcosm of China’s broader experience: suddenly being cut off and now scrambling to build its own.”

Beijing also pushed foreign companies to turn over technology as a price of admission to the China market. Qualcomm, a San Diego tech giant, entered into a joint venture with Huaxintong Semiconductor in 2016. The Chinese government provided land and financing, and Qualcomm offered the technology and about $140 million in initial funding.

During this time, Huawei became one of China’s most popular smartphone makers. And it started working closely with chip factories to make chips for smartphones and A.I. systems.

Huawei has come out with a line of chips that are comparable to some of Nvidia’s older models. But analysts said those chips contained key components that foreign rivals like TSMC and Samsung had made.

Clouds and Clusters

The inability to get essential tools from ASML has been a major chokehold for Chinese chip makers. Since U.S. officials led an effort to lobby the Dutch government to block shipments to China, no Chinese company has been able to buy ASML’s most advanced tools.

Instead, Chinese chip makers have recruited engineers with experience using those machines at TSMC, the world’s top chip maker. And now, Chinese start-ups are trying to make their own chip manufacturing equipment.

A.I. systems require an immense amount of computing power to learn. China’s A.I. companies are trying to get the computing power they need by strapping together numerous less powerful chips. Huawei has taken such an approach, and the Chinese government has built what it calls “intelligent computing clusters” that are essentially state-run data centers.

But those clusters need a lot of chips. Experts and people who work in the industry say China’s most advanced chip maker, Semiconductor Manufacturing International Company, which does some work for Huawei, has struggled to produce enough chips. The chips it does produce are prone to defects and use more electricity than cutting-edge foreign ones. SMIC did not respond to a request for comment.

“Manufacturing volume is going to be an issue,” said Kendra Schaefer, a partner at Trivium China, a research and advisory firm.

Nonetheless, multiple Chinese A.I. researchers have reported breakthroughs in finding new ways to link chips together for maximum efficiency. Zhipu said last month that it had built its latest model entirely using Huawei’s chips and software.

So far, the efficiency gains have been limited and have not helped Chinese companies escape the fact that A.I. demands huge quantities of chips.

Another way China’s A.I. companies are getting the computing power they need is by paying cloud providers like Alibaba and Amazon for remote access to massive data centers stocked with powerful chips.

But the strategy is expensive.

Documents filed by Zhipu and Minimax, another Chinese A.I. start-up, with the Hong Kong Stock Exchange last month show that the two companies are spending a lot more buying cloud services than they are earning in revenue.

Translation

中國如何打造晶片產業，為何仍有所不足（2/2）

 北京推進晶片自給自足已逾十年，但中國企業生產的晶片數量和性能卻不如外國競爭對手

 （繼續）

華為的轉型

2014年，中國是全球最大的半導體市場。但其企業使用的晶片中，90%產自國外。

出於對這種依賴性的擔憂，中國最高決策機構國務院批准了一項數十億美元的計劃，並承諾：到2030年，中國將實現半導體供應鏈所有環節的自主生產。

中國政策制定者有理由擔憂外國技術對中國基礎設施構成的風險。同年早些時候，前美國國家安全局承包商Edward J. Snowden提供的文件披露，美國政府曾監聽華為高層的通訊。

隨後在2017年，特朗普總統以中國涉嫌違反美國對伊朗的製裁為由，對中國電信巨頭中興通訊處以罰款，一夜之間重創了其業務。儘管中興通訊並不生產晶片，但這項舉措再次讓中國認識到自力更生的必要性。

接下來是華為。特朗普第一屆政府發起了一場全球運動，要求各國停止在其電信基礎設施中使用華為的設備。華為的回應是剝離了該業務線，並配合北京的自給自足計劃。

布魯金斯學會研究中國產業政策的研究員Kyle Chan表示:「華為的獨特之處在於其能力以及與中國國家目標的契合度」;「華為的經歷是中國整體經歷的一個縮影：突然被切斷聯繫，現在正努力構建自己的體系」。

北京也要求外國公司交出技術，以此作為進入中國市場的准入條件。總部位於聖地牙哥的科技巨頭Qualcomm於2016年與Huaxintong半導體成立了一家合資企業。中國政府提供土地和資金，而Qualcomm則提供技術和約1.4億美元的初期資金。

在此期間，華為成為中國最受歡迎的智能手機製造商之一。它開始與晶片工廠密切合作，為智能型手機和人工智能系統生產晶片。

華為推出了一系列晶片，其性能可與英偉達的一些早期型號相媲美。但分析師指出，這些晶片包含的關鍵組件是由台積電和三星等外國競爭對手生產的。

雲端和集群

一直是困擾中國晶片製造商的一大難題是無法從ASML獲得關鍵工具。自從美國官員主導遊說荷蘭政府阻止向中國出口ASML設備以來，沒有一家中國公司能夠購買ASML最先進的設備。

取而代之的是，中國晶片製造商從世界頂級晶片製造商台積電（TSMC）招募了擁有相關設備使用經驗的工程師。如今，中國新創公司正嘗試自行研發晶片製造設備。

人工智能系統需要強大的運算能力才能進行學習。中國的人工智能公司正試圖透過將眾多性能較低的晶片組合在一起來獲得所需的運算能力。華為就採用了這種方法，而中國政府也建造了所謂的“智能計算集群”，這些集群本質上是國有數據中心。

但這些集群需要大量的晶片。專家和業內人士表示，中國最先進的晶片製造商 - 中芯國際(SMIC)，它為華為提供部分晶片製造服務 - 一直難以生產足夠的晶片。中芯國際生產的晶片缺陷率高，而且比國外最先進的晶片耗電量更大。中芯國際未對此置評。

研究顧問公司 Trivium China 的合夥人 Kendra Schaefer 表示:「產量將是一個問題」。

儘管如此，多位中國人工智能研究人員報告稱，他們在尋找將晶片連接起來以實現最高效率的新方法方面取得了突破。智普人工智能(Zhipu AI) 上個月表示，其最新型號完全使用了華為的晶片和軟件。

到目前為止，效率提升有限，也未能幫助中國企業擺脫人工智能需要大量晶片的困境。

中國的人工智能公司是獲取所需運算能力的另一種方式是向阿里巴巴和亞馬遜等雲端服務供應商付費，利用遠端去用配備有強大晶片的大型數據中心的服務。

但這種策略成本高。

智普(Zhipu AI)和另一家中國人工智能新創公司 Minimax 上個月向香港證券交易所提交的文件顯示，這兩家公司在購買雲端服務上的支出遠遠超過了它們的收入。

So, in the development of AI, one thing was holding back China: They needed more superfast semiconductors. China’s most advanced chip maker has struggled to produce enough chips. The chips it does produce are prone to defects and use more electricity than cutting-edge foreign ones. One solution for China’s A.I. companies is to get the computing power they need by paying cloud providers such as Alibaba and Amazon for their services, but the strategy is expensive. Apparently, China is facing a dilemma in its development in AI.

中國如何打造晶片產業，為何仍有所不足（1/2）

Recently the New York Times reported the following:

How China Built a Chip Industry, and Why It’s Still Not Enough (1/2)

More than a decade into Beijing’s push for self sufficiency, Chinese firms are producing fewer, lower-performing chips than their foreign competitors.

The NYT - By Meaghan Tobin reporting from Taipei, Taiwan ; Xinyun Wu contributed reporting from Taipei. Meaghan Tobin covers business and tech stories in Asia with a focus on China and is based in Taipei.

Feb. 14, 2026, 12:00 a.m. ET

At a conference at Tsinghua University in Beijing in January, a group of the most influential executives and founders working in artificial intelligence in China gathered to discuss the state of their industry. The mood was bullish. One of the companies in the room, which included people from Tencent, Alibaba and Zhipu AI, could soon lead the world, they agreed.

But one thing was holding them back: They needed more superfast semiconductors.

This year, Chinese chip makers are likely to produce a small fraction of the number of advanced chips made by foreign firms. Huawei, the telecommunications and electronics company leading China’s chip charge, has said it will need almost another two years to make chips that can perform as well as the current offerings from Nvidia of Silicon Valley.

“Even the national champion is fighting an uphill battle,” said Xiaomeng Lu, a director with Eurasia Group, a political consultancy and research group in Washington.

Still, while Chinese chip companies make fewer, slower chips — in large part because U.S. policies have prevented them from importing key tools — there is no shortage of momentum in the country’s A.I. industry.

While Washington’s export controls have slowed China’s chip development, they have added fuel to Beijing’s decade-long push to make strategic technologies like semiconductors and A.I. entirely at home.

Government and private money has been pouring into the development of Chinese artificial intelligence. Chinese tech stocks have made huge gains — Alibaba soared more than 94 percent last year. A stream of Chinese A.I. start-ups are going public. Last month, two of China’s most promising A.I. companies raised more than $1 billion in Hong Kong listings.

The gap between the money flowing into China’s A.I. sector and the reality that Chinese companies produce fewer chips than the country needs underlines the urgency of Beijing’s self-sufficiency efforts, and how much the Chinese A.I. industry still depends on foreign chips.

In December, President Trump extended China a lifeline when he allowed Nvidia to sell some of its advanced chips to Chinese companies, reversing years of U.S. policy. But whether China will get broad access to those chips remains an open question ahead of Mr. Trump’s planned visit to Beijing next month.

The Memory Chip Lag

The Chinese government’s push to make cutting-edge chips at home began more than a decade ago. And it has spent more than $150 billion on the drive.

China’s biggest tech companies, including Huawei, Alibaba and the TikTok parent company ByteDance, have started chip design businesses. Chip makers, many working with Huawei, are building dozens of factories and have hired top engineers from Taiwan and South Korea.

But the task of catching up has gotten progressively more difficult. While Chinese companies have been building their own supply chain for chip making, officials in Washington have tried to hold them back. Three presidential administrations have used export controls to keep Chinese companies from buying advanced chips and the tools to make them, over concerns the technology could fuel China’s economic and military power.

The restrictions have kept Chinese companies from buying equipment made by the Dutch company ASML that performs a crucial step in the chip making process. The lack of access to these machines, which are the size of school buses, is one reason Chinese companies are making chips that lag the performance of the top of the line from Nvidia.

Those are the kinds of chips that power artificial intelligence systems. Chinese companies will most likely make just 2 percent as many A.I. chips as foreign firms do this year, said Tim Fist, a director at the Institute for Progress, a think tank in Washington.

The production gap between Chinese and foreign manufacturers is especially big for memory chips, which are essential for the large calculations done by A.I.

Companies outside China will make 70 times as much memory storage capacity this year as Chinese chip makers will, Mr. Fist said.

The leading makers of memory chips are the South Korean conglomerates Samsung and SK Hynix. Taiwan Semiconductor Manufacturing Company, the world’s biggest chip producer, dominates production of the most advanced chips.

(to be continued)

Translation

中國如何打造晶片產業，為何仍有所不足（1/2）

北京推進晶片自給自足已逾十年，但中國企業生產的晶片數量和性能卻不及國外競爭對手。

今年1月，在北京清華大學舉行的會議上，一群中國人工智能領域最具影響力的企業主管和創辦人齊聚一堂，探討產業現況。會場氣氛樂觀。企業包括騰訊、阿里巴巴和智普人工智能在內的多家公司，其中一家預計他們很快就會引領全世界，與會者一致認同。

但有一點阻礙了他們：他們需要更多超高速半導體。

今年，中國晶片製造商生產的先進晶片數量可能僅為外國公司產量的一小部分。引領中國晶片產業發展的電信和電子公司華為表示，還需要近兩年時間才能生產出性能與矽谷英偉達現有產品相當的晶片。

華盛頓政治諮詢和研究機構歐亞集團的董事Xiaomeng Lu表示:「即使是國家的冠軍也面臨著一場艱苦的戰鬥」。

儘管中國晶片公司生產的晶片數量較少、速度較慢 - 這在很大程度上是因為美國的政策阻止了它們進口關鍵設備 - 但中國的人工智能產業並不缺乏動力。

雖然華盛頓的出口管制減緩了中國晶片的發展速度，但這卻為北京十年來致力於實現半導體和人工智能等戰略技術完全自主研發的努力注入了動力。

政府和私人資金一直湧入中國人工智能的研發領域。中國科技股漲幅驚人 - 阿里巴巴去年股價飆漲超過94%。大量中國人工智能新創公司紛紛上市。上個月，兩家中國最有前景的人工智能公司在香港上市，融資超過10億美元。

流入中國人工智能領域的資金，與中國企業晶片產量遠低於國內需求的現實之間存在巨大差距，凸顯了北京推進晶片自給自足的迫切性，以及中國人工智能產業對外國晶片的依賴程度。

去年12月，特朗普總統允許英偉達向中國企業出售部分先進晶片，此舉推翻了美國多年來的政策，為中國提供了一線生機。但中國能否廣泛取得這些晶片，在特朗普下個月訪華之前仍是個未知數。

記憶體晶片的滯後

中國政府推動自主研發尖端晶片的計劃始於十多年前，並已為此投入超過1500億美元。

包括華為、阿里巴巴和TikTok母公司位元組跳動在內的中國大型科技公司已開始涉足晶片設計業務。許多與華為合作的晶片製造商正在建造數十家工廠，並從台灣和韓國聘請了頂尖工程師。

但追趕的難度卻與日俱增。儘管中國企業一直在建造自己的晶片製造供應鏈，但華盛頓的官員卻試圖阻撓它們。三屆美國總統政府都曾動用出口管制，阻止中國企業購買先進晶片及其製造設備，理由是擔心這項技術會增強中國的經濟和軍事實力。

這些限制使得中國企業無法購買荷蘭ASML公司生產的設備，而這些設備正是晶片製造過程中至關重要的一環。無法獲得這些在體積上堪比一輛校車的機器，是中國企業生產的晶片性能落後於英偉達頂級晶片的原因之一。

而這些晶片正是人工智能系統的核心。華盛頓智庫「進步研究所」（Institute for Progress）主任 Tim Fist表示，中國企業今年生產的人工智能晶片數量很可能只有外國公司的2%。

中國製造商與外國製造商在記憶晶片領域的產量差距尤其巨大，而記憶晶片對於人工智能（AI）的大規模運算至關重要。

Fist先生指出，今年中國以外公司的記憶體晶片產量將是中國晶片製造商的70倍。

領先的記憶體晶片製造商是韓國的三星和SK海力士兩大企業集團。台積電（TSMC）是全球最大的晶片生產商，在最先進晶片的生產中佔據主導地位。

（待續）

根據《金融時報》報道，儘管美國實施限制，價值10億美元的英偉達AI晶片仍進入中國

Recently Yahoo News on-line picked up the following:

Nvidia AI chips worth $1 billion entered China despite US curbs, FT reports

Reuters - Reporting by Jaspreet Singh in Bengaluru; Editing by Arun Koyyur

Thu, July 24, 2025 at 5:27 a.m. PDT·2 min read

(Reuters) -Nvidia's advanced artificial intelligence chips worth at least $1 billion were smuggled to China in the three months after Washington tightened chip export controls, the Financial Times reported on Thursday.

The AI chip designer's high-end B200 processors, banned for sale in China, is widely available on a thriving Chinese black market for U.S. chips, the report said, citing sales contracts, company filings and multiple people with direct knowledge of the deals.

Nvidia told Reuters that building data centers with smuggled products is inefficient both technically and financially, as the company only offers service and support for authorized products.

The U.S. Department of Commerce, White House and Thai government did not immediately respond to requests for comment. Reuters could not independently verify the FT report.

In May, multiple Chinese distributors started selling B200s to suppliers of data centers that serve Chinese AI groups, according to the report.

The U.S. and China are battling for global dominance in AI and other cutting-edge technologies, triggering a tightrope walk for companies such as Nvidia between the world's two largest economies.

Nvidia last week said it would be allowed to resume sales to China after the Trump administration reversed an export restriction on the sales of chips such as H20. The curbs were imposed in April.

In the three months before that, Chinese distributors from Guangdong, Zhejiang and Anhui provinces sold Nvidia's B200s, as well as other restricted processors such as the H100 and H200, according to the report.

Southeast Asian countries have become markets where Chinese groups obtained restricted chips, the report said, citing industry experts.

The U.S. Commerce Department is discussing adding more export controls on advanced AI products to countries such as Thailand as soon as September, the report said.

Translation

根據《金融時報》報道，儘管美國實施限制，價值10億美元的英偉達AI晶片仍進入中國

（路透社）- 根據《金融時報》週四報道，在華盛頓加強晶片出口管制後的三個月內，至少價值10億美元的英偉達先進人工智慧晶片被走私至中國。

該報告引述銷售合約、公司文件以及多名直接了解相關交易的人士的話稱，在中國被禁止銷售的這家AI晶片設計公司的高階B200處理器，在蓬勃的中國的美國晶片黑市上隨處可得到。

英偉達向路透社表示，使用走私產品建造數據中心在技術和財務上都是低效的，因為該公司只為授權產品提供服務和支援。

美國商務部、白宮和泰國政府均未立即回應置評要求。路透社無法獨立核實《金融時報》的報導。

據報道，今年5月，多家中國經銷商開始向服務中國人工智能集團的數據中心供應商銷售B200處理器。

美國和中國正在爭奪人工智能和其他尖端技術領域的全球主導地位，這使得英偉達等公司在世界兩大經濟體之間著面對两難的挑戰。

英偉達上週表示，在特朗普政府撤銷對H20等晶片出口限制後，該公司將被允許恢復對華銷售。此限制措施於4月實施。

據報道，在此之前的三個月裡，來自廣東、浙江和安徽省的中國經銷商銷售了英偉達的B200處理器，以及其他受限的處理器，例如H100和H200。

報告引述業內專家的話稱，東南亞國家已成為中國企業取得受限晶片的市場。

報道稱，美國商務部正在討論最快於9月對泰國等國家加強對先進人工智能產品的出口管制。

So, Nvidia's advanced artificial intelligence chips worth at least $1 billion were smuggled to China in the three months after Washington tightened chip export controls and Southeast Asian countries had become markets where Chinese groups could obtain restricted chips. I am wondering what the US will do regarding this smuggling.

China quietly withdraws retaliatory tariffs on some US-made semiconductors, import agents reveal

 Recently CNN.co.jp reported the following:

中国、米国製半導体の一部で報復関税をひそかに撤回　輸入代理店が明らかに

2025.04.25 Fri posted at 15:08 JST

香港（ＣＮＮ） 中国が米国製半導体の一部に対する１２５％の報復関税をひそかに撤回したとみられる。広東省深圳の輸入代理店３社が２５日、ＣＮＮに明らかにした。

３社によると、この免除は半導体に適用される。代理店らはこの免除措置について２４日遅くに知ったという。公式発表は行われていない。

中国は今月１２日、トランプ米大統領が中国製品への関税を１４５％に引き上げたことを受け、米国の全製品に対する報復関税を１２５％に引き上げた。

中国政府は数カ月にわたり、自国には米国との貿易戦争の激化に耐える能力があるとの力強い姿勢と自信を示してきた。しかし、今回の免除は、国内で生産したり、他国から調達したりできない重要品目については関税を一部撤回する必要があることを示唆している。

中国当局は、この免除を公式に確認していない。税関当局は今回の免除について認識していないと述べた。

輸入代理店のサプライチェーン（供給網）マネジャーはＣＮＮに対し、２４日にメモリーチップを除く大半の半導体を含む８種類の集積回路への関税が免除され、ゼロになったことを知ったと語った。

深圳の税関当局は、この変更について一部の企業に通知したようだ。通知を受けた１社の従業員がＣＮＮに確認した。

中国の「財経」誌は２５日、上海に拠点を置く企業を含む、半導体を輸入する複数のテクノロジー企業の話をもとにこの免除措置について報じた。

Translation

China quietly withdraws retaliatory tariffs on some US-made semiconductors, import agents reveal

Hong Kong (CNN) China appeared to have quietly withdrawn retaliatory tariffs of 125% on some US-made semiconductors. Three import agents in Shenzhen, Guangdong Province, revealed to CNN on the 25th.

According to the three agents, the exemption applied to semiconductors. The agents learned of the exemption late on the 24th. No official announcement had been made.

On the 12th of this month, China raised retaliatory tariffs on all US products to 125% in response to US President Trump's increase in tariffs to 145% on Chinese products

For months, the Chinese government had been strong and confident that the country could withstand an escalating trade war with the US. However, the exemption suggested that some tariffs on critical items needed to be withdrawn as they could not be produced domestically nor sourced from other countries.

Chinese authorities had not officially confirmed the exemption. Customs officials said they were not aware of the exemption made this time.

A supply chain manager at an import agency told CNN that they learned it on Wednesday that, except memory chips, tariffs on eight types of integrated circuits that would include almost all the majority semiconductors had been waived and dropped to zero.

Customs officials in Shenzhen appeared to have notified some companies about the change, an employee at one company who received the notice confirmed to CNN.

China's Caijing magazine reported on the exemption on Thursday based on several technology companies that import semiconductors, including one Shanghai based company.

              So, China quietly withdraws retaliatory tariffs on some imported US-made semiconductors. Apparently, the tariffs war is beginning to impact on all the countries involved.

越南向中國徵收鋼鐵關稅抑制其激增出口

 Recently Yahoo News on-line picked up the following:

Vietnam Hits China with Steel Tariffs to Fight Surging Exports

Yihui Xie and Katharine Gemmell

Sun, February 23, 2025 at 11:41 p.m. PST·2 min read

(Bloomberg) -- Vietnam will impose anti-dumping tariffs on steel from China, following South Korea and other nations in fighting back against surging supplies from the world’s biggest producer.

The Southeast Asian nation will impose temporary tariffs on some hot-rolled coil starting from early March, according to a statement from the Ministry of Industry and Trade on Friday. Outside China itself, Vietnam is the biggest single buyer of Chinese steel, and hot-rolled coil is a major export product.

China’s sent the most steel overseas in nine years in 2024, as its producers turned to global markets to offset a deep construction slowdown at home. That set the stage for President Donald Trump to propose a blanket 25% tariff on all US imports, and has prompted nations from South Korea to Brazil and India to consider levies.

The flurry of protectionism will pile pressure on Beijing to rein in its billion-ton steel industry after several years of slowing domestic demand. Steel futures in China fell as much as 1.8%, while steelmakers in Vietnam advanced.

Recent tariff decisions “should incentivize the Chinese government to launch another round of supply reform” to boost supply discipline and improve industry profitability, analysts including Jack Shang from Citigroup Inc. wrote in a note.

Vietnam’s temporary tariffs of between 19.38% and 27.83% will come into force on March 7 and last for 120 days. China exported about 8 millions tons of HRC to Vietnam last year, and the tariffs would likely cover about 50% of that volume, Citigroup said, citing discussions with industry players.

The anti-dumping probe was triggered by Hoa Phat Group and Formosa Ha Tinh Steel Corp., two major Vietnamese steelmakers who requested an investigation of imports from India and China last year. The government won’t go ahead with duties on India at this moment, it said.

HRC futures on China’s Shanghai Futures Exchange were 1.3% lower by 3:34 p.m. local time, while iron ore in Singapore was little changed at $107.50 a ton.

Translation

越南向中國徵收鋼鐵關稅抑制其激增出口

(彭博社)－繼韓國和其他國家針對全球最大鋼鐵生產國中國鋼鐵供應激增進行反傾銷之後，越南將對來自中國的鋼鐵徵收反傾銷關稅。

越南工業和貿易部週五發表的聲明稱，這個東南亞國家將從 3 月初開始對部分熱軋捲板(HRC)徵收臨時關稅。中國對外，越南是中國鋼鐵最大的單一買家，熱軋捲板是中國鋼鐵的主要出口產品。

到2024年的九年內，中國的大部分鋼鐵產品是出口到海外，因為中國鋼鐵生產商將目光轉向全球市場，以抵消國內建築業大幅放緩的影響。這為特朗普總統提議對所有美國進口產品徵收 25% 的關稅提供了基礎，並引使韓國、巴西和印度等國也考慮徵收關稅。

在國內需求連續幾年放緩之後，一系列保護主義將給北京帶來壓力，迫使其遏制數十億噸的鋼鐵業。中國鋼鐵期貨下跌1.8%，而越南鋼鐵上漲。

花旗集團的分析師 Jack Shang 在一份報告中寫道，最近的關稅決定 “應該會激勵中國政府啟動新一輪供應改革” ，以加強供應紀律並提高行業盈利能力。

越南19.38%至27.83%的臨時關稅將於3月7日生效，為期120天。花旗集團引述與業內人士的討論結果稱，去年中國向越南出口了約 800 萬噸熱軋卷，加徵關稅的範圍可能涵蓋其中的 50% 左右。

這項反傾銷調查由越南兩大鋼鐵製造商 Hoa Phat Group 及 Formosa Ha Tinh Steel Corp發起，這兩家公司去年申請對來自印度和中國的進口產品進行調查。政府表示，目前不會對印度徵收關稅。

截至下午 3:34，中國上海期貨交易所熱軋捲板期貨下跌 1.3%, 而在新加坡鐵礦石價格基本持平，每噸 107.50 美元。

       So, Vietnam will impose anti-dumping tariffs on steel from China, following South Korea and other nations in fighting back against surging supplies from the world’s biggest producer. Apparently, China is getting more pressure from abroad and a tariff war is also taking place among Asian countries.

誰是人工智能初創DeepSeek的創辦人？ （2/2）

 Recently the New York Times reported the following:

Who Is the Founder of the A.I. Start-Up DeepSeek? (2/2)

The Chinese company DeepSeek seemed to have come out of nowhere this week when it upturned markets. Here’s what to know about Liang Wenfeng, the engineer who started it.

Jan. 29, 2025

(continue)

His 2010 thesis at Zhejiang University tackled what would soon become one of the hottest topics in Chinese A.I.: improving intelligent tracking algorithms for surveillance cameras.

Later, the hedge fund he co-founded was buffeted by regulatory pressures, eventually forcing the closure of one of its main investment products, according to Peter Alexander, managing director of Z-Ben Advisors, a market consultancy, who researches Chinese hedge funds.

“Between 2019 and 2023, they wanted to have this side project so that their Ph.D.s felt like they had something to do and so DeepSeek originated from that,” Mr. Alexander said.

“But it really went into overdrive when their primary investment product had to shut down in February of 2024,” he said.

In a way, it was China’s crackdown on the private sector that nudged DeepSeek toward long-term A.I. research.

He’s willing to try things other entrepreneurs will not. He even hired lit majors.

If you polled China’s A.I. experts on who would deliver the country’s first major generative A.I. breakthrough, few would have picked Mr. Liang. That included the Chinese government.

DeepSeek was a private company with no apparent state backing, no big-name alliances and none of the institutional heft of players like Baidu, the search giant. In a system that favored insiders, Mr. Liang was not one.

Yet, there’s precedent for that. Some of China’s most disruptive tech companies — Huawei, Alibaba, ByteDance — started outside the spotlight, only to redefine their industries.

Mr. Liang’s approach has been as unconventional as his company’s rise. He has emphasized intellectual exploration over sheer grind. His hiring philosophy is equally unorthodox — DeepSeek’s engineering teams have been joined by literature buffs to help refine the company’s A.I. models.

“Everyone has their own unique journey and brings their own ideas with them, so there’s no need to push them,” he said in the 36Kr interview.

In a tech culture defined by grueling hours and hierarchy, that outlook borders on Bohemian. Yet, Mr. Liang insists change is necessary if China wants to lead in frontier A.I. innovation.

“When ChatGPT came out, the entire industry in China lacked the confidence to pursue frontier innovation,” he said.

“Innovation starts with confidence — and we often see that more from young people.”

Translation

誰是人工智能初創DeepSeek的創辦人？ （2/2）

本週，中國公司DeepSeek 似乎突然出現，扭轉了市場局勢。以下是關於這家公司的創始人工程師梁文峰的一些資訊。

（繼續）

他在2010年浙江大學的論文研究, 成為後來中國人工智慧最熱門的議題之一：改進監視攝影機的智能追蹤演算法。

研究中國對沖基金的市場顧問公司 Z-Ben Advisors 董事總經理 Peter Alexander 表示，後來，由他聯合創建的對沖基金受到監管壓力的連續衝擊，最終導致其主要投資產品之一關閉。

Alexander說: 「2019 年至 2023 年期間，他們希望開展這個副項目，以便他們的博士生們覺得有事可做，DeepSeek 由此誕生」。

他說：“但當他們的主要投資產品不得不在 2024 年 2 月關閉時，這項目才真正進入超速運作, 加倍努力的狀態。”

從某種程度上來說，正是中國對私人企業的受打擊而推動了 DeepSeek 走向長期人工智能研究。

他願意嘗試其他企業家不願意做的事情。他甚至聘請了主修文學的大學生。

如果你對中國的人工智能專家進行調查, 去看他們討論誰將會推出全國第一個重要的生成人工智能, 很少人會選擇梁先生。這包括中國政府。

DeepSeek 是一家私人公司，沒有明顯的政府支持，沒有大牌聯盟，也不具備搜尋巨頭百度等公司的機構影響力。在一個有利於內部人員的體制中，梁先生不是其中之一員。

但這樣的事是有先例的。華為、阿里巴巴、字節跳動等中國最具顛覆性的科技公司起步於受社會廣泛關注之外，但卻能重新定義了各自的產業。

梁先生的做事方法和他的公司崛起一樣是非傳統的。他強調智力探索而非單純的刻苦學習。他的招募理念同樣非傳統同 - DeepSeek 的工程團隊聘請了文學愛好者來幫助完善公司的人工智能模型。

他在接受36Kr採訪時表示: 「每個人都有自己獨特的人生旅程，帶着自己的想法，所以沒必要強迫他們」。

在以艱苦的工作時間和階級制度為特徵的科技文化中，這種觀點近乎波希米亞主義者 (Bohemian)。然而，梁先生堅稱，如果中國想在人工智能新境界創新方面佔據領先地位，就必須做出改變。

他說： “當ChatGPT問世時，中國整個行業都缺乏追求新境界創新的信心。”

“創新始於信心 - 我們在年輕人身上經常看到它。”

              So, I am impressed by Liang Wenfeng in that he points out why the China’s biggest investors and tech firms have failed to anticipate the rise of generative A.I.- because China’s companies are obsessed with quick returns in a fiercely competitive market. I am also impressed by the fact that DeepSeek’s engineering teams have literature buffs helping to refine the company’s A.I. models based on the belief that “Everyone has their own unique journey and brings their own ideas with them”.

Note:

1. Bohemianism (波希米亞主義) is a social and cultural movement that has, at its core, a way of life away from society's conventional norms and expectations. The term originates from the French bohème and spread to the English-speaking world. It was used to describe mid-19th-century non-traditional lifestyles, especially of artists, writers, journalists, musicians, and actors in major European cities. (Wikipedia)

誰是人工智能初創DeepSeek的創辦人？ （1/2）

Recently the New York Times reported the following: 

 Who Is the Founder of the A.I. Start-Up DeepSeek? (1/2)

(Source: NYT)

The New York Times

The Chinese company DeepSeek seemed to have come out of nowhere this week when it upturned markets. Here’s what to know about Liang Wenfeng, the engineer who started it.

Jan. 29, 2025

In technology, many entrepreneurs get only one defining act. Liang Wenfeng, the founder of DeepSeek, is already on his second.

The engineer, described by colleagues as introspective, first made his mark in China’s investment world in the late 2010s, cofounding a hedge fund that used artificial-intelligence models to deliver strong returns and attracted billions of dollars in capital.

Buoyed by profits and wary of Beijing’s tightening grip on speculative trading, Mr. Liang pivoted in 2023. He poured money into artificial intelligence, betting on A.I. chips and assembling a team to build China’s answer to the Silicon Valley front-runner OpenAI.

Now, just two years later, DeepSeek has upended the global tech landscape. Here is what you need to know about Liang Wenfeng.

He’s a deeply technical engineer. That puts him in a line of other successful Chinese tech executives.

When Chinese technologists debated why the country’s biggest investors and tech firms failed to anticipate the rise of generative A.I., many pointed to a single culprit: China’s companies were obsessed with quick returns in a fiercely competitive market.

Armed with those lessons — and backed by his own trading windfall — Mr. Liang has made it clear his ambitions lie far beyond commercial applications.

His focus, he has said, is on what he sees as China’s only real chance to catch up with the United States. That means taking bold, idealistic swings at fundamental A.I. challenges. His primary ambition is to create artificial general intelligence or A.G.I. — the elusive goal of building machines that can think and learn like humans.

When DeepSeek undercut its Chinese competition last year by offering its model at bargain prices, forcing larger rivals into their own price cuts, Mr. Liang dismissed the significance.

“To be honest, we didn’t really care — it was just something we did along the way,” he said in a widely shared interview with 36Kr, a Chinese tech outlet. “Providing cloud services is not our main goal. Our aim is still to achieve A.G.I.” (DeepSeek has remained mostly silent this week and has not responded to requests for comment.)

In his convictions that superhuman artificial intelligence is just around the corner, Mr. Liang sounds a lot like OpenAI’s chief executive, Sam Altman. But the similarities end there. Mr. Liang, a low-profile executive with a deep technical background in A.I. engineering, more fits the mold of Pony Ma, a co-founder of China’s Tencent, than Silicon Valley’s charismatic visionaries.

He started as a hedge fund trader. And then he turned to pure A.I. research.

In many ways, Mr. Liang’s career traces the major shifts in China’s technology landscape.

(to be continued)

Translation

誰是人工智能初創DeepSeek的創辦人？ （1/2）

本週，中國公司DeepSeek 似乎突然出現，扭轉了市場局勢。以下是關於這家公司的創始人工程師梁文峰的一些資訊。

在科技領域，許多創業家只經歷一決定性的舉動。 DeepSeek 創辦人梁文鋒是在經歷第二次決定性的舉動。

同事們都說這位工程師是很自省的，2010 年代末，他首次在中國投資界嶄露頭角，與他人共同創立了一隻對沖基金，利用人工智慧模型實現豐厚回報，吸引了數十億美元的資本。

受利潤的鼓舞，加上擔心北京方面加強對投機交易的控制，梁先生在 2023 年做出了轉變。他向人工智能投入了大量資金，押注人工智能晶片, 並組成團隊打造中國對矽谷領跑者OpenAI的回應。

如今，僅僅兩年後，DeepSeek 已經顛覆了全球科技格局。以下內容是關於你需要了解梁文峰的。

他是一位技術精湛的工程師。這使他躋身其他成功的中國科技高階主管之列。

當中國技術專家爭論為何該國最大的投資者和科技公司未能預見到生成人工智慧的興起時，許多人指出了一個罪魁禍首：在競爭激烈的市場中，中國公司沉迷於快速回報。

憑藉這些經驗教訓，並藉助自身交易所得的意外之財，梁先生已明確表示，他的抱負遠不止於商業應用。

他說，他的焦點是他認為這是中國趕上美國的唯一真正機會。這意味著對基礎人工智能挑戰進行大膽、帶理想主義的出擊。他的主要目標是創造通用人工智能（AGI）。 — 建構能夠像人類一樣思考和學習的機器這一難以實現的目標。

去年，DeepSeek 透過低價推出其模型來削弱其中國競爭對手，從而迫使規模更大的競爭對手也降價，但梁先生否認了此舉的重要性。

他在接受中國科技媒體36Kr採訪時表示: 「說實話，我們不太在意的 - 這只是我們在這個過程中所做的事情」, 這採訪受廣泛分享。 「提供雲端服務不是我們的主要目標。我們的目標仍然是實現 AGI。 （DeepSeek 本週基本上保持沉默，沒有回應置評請求。）

梁先生堅信超人類的人工智能即將到來，他的觀點與 OpenAI 的執行長 Sam Altman 很相似。但相似之處僅止於此。梁先生是一位低調的主管，但在人工智慧領域擁有深厚的技術背景, 更符合中國騰訊聯合創辦人馬化騰的形象, 多於矽谷那些魅力非凡的夢想家。

他最初是一名對沖基金交易員。然後他轉向純人工智能研究。

從許多方面來看，梁先生的職業生涯都見證了中國科技格局的重大轉變。

（待續）