这些婚外情无疑显示了黄仁勋判断力的严重缺失，我认为这在多个层面上也反映了其不受制约的傲慢。这样的选择不仅背叛了他相伴40年的妻子，也给他的公司带来了风险。与任何英伟达合作伙伴发生关系已经够糟糕了，但他选择的对象处于地缘政治的火药桶中。即使爱丽丝·黄（Alice Huang）的公司没有被怀疑帮助中国企业规避芯片限制（而且他对此毫不知情），它仍然通过向中国公司租赁强大的AI基础设施来利用法律漏洞。
民主党人——他们将在明年1月新一届国会开幕时接管关键委员会——也会把他召到国会山作证吗？目前尚无人知晓，但黄仁勋曾炫耀他与特朗普总统的密切关系。还记得9月份在“All-In峰会”上时，特朗普给他打电话的场景吗？此外，英伟达还向特朗普的宴会厅项目捐赠了2500万美元，这是一个在美国选民中极不受欢迎的项目。几位民主党议员已经誓言要调查背后的资金来源。
换句话说，他是特朗普身边的一位亿万富翁。直到周四为止，他还是AI和数据中心领域的代言人（这也是选民眼中一个极不受欢迎的政治话题），并且他公开承认与一名涉嫌中国间谍的人有染。尽管他已不再是CEO，但这似乎对民主党人来说是一个过于诱人的目标，无法忽视。
黄仁勋一直是科技行业和政府政策制定者中开放权重AI模型最强大、最响亮的倡导者。英伟达不仅提供自己的开放权重模型，还提供大量用于智能体、AI安全等方面的开源技术。他甚至同意以129亿美元的价格收购托管这些模型的最大网站Hugging Face。（该交易预计将在明年上半年完成，需等待监管批准。）
开放权重模型为企业提供了市场选择并降低了成本。如果没有黄仁勋，下一个最大的支持者可能是微软的萨提亚·纳德拉（Satya Nadella），但他并没有站出来组织整个行业发声。
黄仁勋在他债券承销商（贷款方）和债券买家几乎对与AI建设相关的债务感到厌倦之际，摧毁了自己的声誉。据《福布斯》报道，在今年发行的约800亿美元数据中心债券中，近一半被评为高风险的“垃圾债”。
毫不气馁，黄仁勋一直在想出继续为繁荣融资并保持公司收入增长的方法。显然，他授权了令人难以置信数量的所谓循环交易，即英伟达购买公司股份以资助其扩张，以便它们购买芯片。据一些报道，截至7月，公司已承诺此类交易金额达7500亿美元，彭博社报道。8月，英伟达变得更加大胆，同意为俄亥俄州一家OpenAI数据中心的1050亿美元融资提供担保，CNBC报道。
黄仁勋还说服了一整套新的 financiers 出资。8月，他宣布拥有雄厚资金的私募股权公司Apollo、BlackRock、Blackstone、Brookfield、Goldman Sachs和KKK已承诺投入高达5000亿美元用于数据中心建设。他向他们保证，数据中心的英伟达GPU可以作为抵押品，因为英伟达担保这些芯片将保值。
黄仁勋描绘了一幅英伟达作为“可投资基础设施”的愿景，具有像飞机一样的长期转售周期。然后，他像往常一样穿梭于财经媒体之间推销他的愿景。然而，现在华尔街担心英伟达的股票会崩盘，并拖累他们的投资组合，甚至可能波及经济。
英伟达失去了其顶级销售员，AI行业也失去了其顶级吹鼓手。即使黄仁勋 somehow 重新进入公众视野再次为AI代言，每一次采访和会议都将围绕他的性生活、选择和国家安全展开——而不是他对行业的愿景。
我们将不得不等待看英伟达如何应对黄仁勋强加给它的这场丑闻，但现在该公司终于进入了后“创始人模式”时代，并在寻找新任CEO，恐怕它已没有足够的信誉像黄仁勋那样推动行业前行。
查看原文
Nvidia CEO Jensen Huang’s confession that he had an affair (and reports of another one with a direct report) that led the board to oust him on Thursday isn’t just your run-of-the-mill salacious billionaire CEO sex scandal.
In this case, one of his lovers is suspected by the U.S. government of being a spy for China. She’s also an executive of a Singapore-based neocloud that is suspected of smuggling Nvidia chips to China.
These affairs show undeniably poor judgment by Huang and, I’d argue, unchecked hubris on many levels. Such choices are not only a betrayal of his wife of 40 years, they’re also a liability for his company. Sleeping with any Nvidia partner is bad enough, but he also chose one in a tinderbox of a geopolitical situation. Even if Alice Huang’s company wasn’t suspected of helping Chinese firms evade chip restrictions (and he was somehow unaware of this), it was still exploiting a legal loophole by renting powerful AI infrastructure to Chinese companies.
It’s those potential national consequences that may land him in a hellscape of ongoing public humiliation. Republican committee chairs are already summoning him to testify in closed-door sessions about the security ramifications.
Will Democrats, who are set to take over key committees when the new Congress convenes in January, also haul him in to testify on the Hill? No one knows yet, but Huang has flaunted his close relationship with President Trump. Remember when Trump called him while he was onstage at the All-In Summit in September? Nvidia also donated $25 million to Trump’s ballroom, a deeply unpopular project with American voters. Several Democratic lawmakers have already vowed to investigate the money behind it.
In other words, he is a billionaire close to Trump. He was — until Thursday — the face of AI and data centers (also a deeply unpopular political topic with voters), and he has publicly admitted to having an affair with an alleged Chinese spy. Even though he’s no longer CEO, that seems too juicy a target for Democrats to ignore.
Beyond whatever may be coming, Huang’s influence has been lost right when he was using it to push the tech industry in a number of directions. These areas have now lost their biggest champion.
Open-weight AI models as competitors to closed-source labs
Huang has been the biggest, loudest advocate for open-weight AI models in the tech industry and among government policymakers. Nvidia not only offers its own open-weight models, it also offers a lot of other open source tech for agents, AI security, and more. It has even agreed to acquire the biggest site that hosts them, Hugging Face, for $12.9 billion. (That deal is supposed to close in the first half of next year, pending regulatory approval.)
In July, when Anthropic and OpenAI were reportedly urging the Trump administration to ban Chinese open-weight models, Huang organized a letter of support for open-weight models generally (if not Chinese models specifically), signed by hundreds of tech companies.
Open-weight models give enterprises choice in the market and reduce costs. Without Huang, the next biggest advocate for them is perhaps Microsoft’s Satya Nadella, but he’s not out there organizing the industry to speak up.
Pushing the AI industry to clean up its security
Both OpenAI and Anthropic revealed incident after incident this year as their AI agents hacked into others’ networks or performed other misdeeds. Huang loudly proclaimed the AI industry still didn’t need regulation and would deal with safety itself.
As the industry mulled over what they should do (auditors? a FINRA-like board?), Huang rallied the tech industry again. Nvidia got over 100 companies to sign on to his Open Agent Safety Platform, a largely open source effort to solve the epidemic of misbehaving AI agents.
Naturally, this effort hasn’t instantly solved the problem, and now the top spokesperson for self-regulation via technology solutions is out of his position of influence.
Arguing AI is a tool, not a life form
Huang has also been among the loudest voices insisting that AI is an engineered system that creates products, and so can be controlled like any other engineered system. This viewpoint counters the AI doomers and even the big AI labs themselves, which sometimes position their models as on the verge of consciousness (Anthropic) or as alien minds (OpenAI).
While Huang deals with the fallout of his actions, the loudest voice of this counter-narrative is, for the foreseeable future, silenced.
A new Nvidia-powered AI PC industry
Huang also just convinced a long list of PC makers to buy into his long-time dream of providing CPUs (the main processor chips) for them. In this age, when most Americans have not experienced AI beyond a chatbot, he orchestrated a new line of high-end AI PCs geared for running agents locally.
These new PCs are due out this holiday season, and they may sell well for their intended targets: creators, developers, and others already savvy with agents. But now that Huang has become a meme for cheating on his wife with an alleged spy, he’s perhaps not the best choice to help market these devices as a cool alternative to the Mac mini.
Financial engineering to fund data centers
Huang demolished his reputation right as bond underwriters (lenders) and bond buyers are pretty much over AI build-out-related debt. Of roughly $80 billion in data center bonds issued this year, nearly half were rated high-risk “junk,” Forbes reports.
Undeterred, Huang has been coming up with ways to continue financing the boom and keep his company’s revenues growing. Certainly, he’s authorized a mind-boggling number of so-called circular deals, where Nvidia buys stakes in companies to fund their expansion so they can buy chips. By some accounts, the company had committed to $750 billion in such deals as of July, Bloomberg reported. In August, Nvidia got even bolder, agreeing to backstop $105 billion in financing for an OpenAI data center in Ohio, CNBC reported.
Huang also convinced a whole new set of financiers to pony up. In August, he announced that deep-pocketed PE firms Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR had committed up to $500 billion for data center build outs. He promised them that the data centers’ Nvidia GPUs could be used as collateral because Nvidia was guaranteeing the chips would retain value.
Huang sold a vision of Nvidia as “investable infrastructure” with a long resale cycle, like airplanes. Then he made the rounds with the financial press, as he often did, to pitch his vision. Now, however, Wall Street is concerned that Nvidia’s stock will crash and take their portfolios, and possibly the economy, with it.
Nvidia has lost its top salesperson and the AI industry has lost its top hype man. Even if Huang somehow comes back into the public eye to champion AI again, every interview and meeting will be about his sex life, choices, and national security — not his vision for the industry.
We’ll have to wait and see how Nvidia navigates the scandal Huang heaped onto it, but now that the company is finally in its post “founder mode” era, and hunting for a new CEO, it’s doubtful it has the cache to push the industry the way Huang was.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
Studio