I have been writing this blog long enough to know most weeks in AI are noise. This one was not. The most valuable chip company on the planet lost its founder, the biggest IPO ever happened, a pile of nonprofits asked a judge to freeze frontier AI, and Congress dumped the receipts on who got Nvidia chips in China. Let us go.
01 of 04
Jensen Huang Is Out at Nvidia, and Congress Wants a Word
On November 8, Jensen Huang sent a letter to Nvidia employees admitting he had been unfaithful to his wife. The board opened an independent investigation that same day. By November 12, after 33 years running the company he co-founded in 1993, he was out as CEO.
The affair alone would be tabloid stuff. What makes it a national security story is who it was with. According to reporting compiled by TechCrunch, U.S. officials believe the woman, Alice Huang (no relation), an executive at the Singapore AI cloud company Megaspeed, is a spy for China's Ministry of State Security. That is an allegation, not a finding. Megaspeed reportedly bought $4.8 billion in Nvidia chips over two years and has been under U.S. and Singapore investigation since October 2025 over possible diversion to China. Nvidia's board said early on it had no reason to believe national security or company secrets were compromised.
CFO Colette Kress is now interim CEO while keeping the CFO job, and longtime director Harvey Jones became Nvidia's first-ever board chair, per Fortune. On November 17, Jones said the board is not considering bringing Huang back, even after Huang posted on X that he "would be honored" to return and Elon Musk, Palmer Luckey, and Garry Tan publicly cheered for it.
Meanwhile, three House Republican chairmen (John Moolenaar of the Select Committee on China, Brian Mast of Foreign Affairs, and Rick Crawford of Intelligence) asked Huang for a closed-door briefing, and Huang has agreed to testify before Thanksgiving. Nvidia also reported $115.8 billion in quarterly revenue, up 103 percent, and Kress said the company will stop selling H200 chips in China.
Here is where I land on it. Nvidia the business is fine. Business as usual, Kress said, and the numbers back her up. The part I care about is that a single person's private life could be a back door into the most important supply chain in tech. If the allegations hold up, that is a scandal about vetting and export enforcement, not about one man's marriage. I want to hear what he tells Congress behind those closed doors, and I want a transcript.
02 of 04
Anthropic's Record IPO and the Circular Money Question
Anthropic priced its IPO at $83 a share and raised more than $80 billion in new capital, the largest offering on record. When it hit Nasdaq on November 19, the stock opened around $111, about 33 percent above the IPO price, per TechCrunch. The seven co-founders cashed out roughly $280 million each, close to $2 billion combined.
At the opening bell, CEO Dario Amodei went out of his way to say the company would not bend to business or political pressure and that "we will pace the training and deployment of AI systems." Not the usual ring-the-bell victory lap.
The more interesting read is TechCrunch's look at the financing. Amazon put in $18 billion and now holds a 14.7 percent stake worth upwards of $159 billion at the IPO price. Google has put in $12.3 billion since 2023, with another $30 billion promised on milestones, for a 13.7 percent stake worth nearly $149 billion. Alongside the IPO, a $26 billion private placement came from Nvidia, AMD, Microsoft, Broadcom, Micron, TSMC, and Google. Most of those same names sell Anthropic its compute. The company's SEC filing shows about $900 billion in committed spending, mostly compute and infrastructure. Third-quarter revenue was $17 billion, and the company is still losing tens of billions a quarter.
What I want to know is who actually carries the risk now. The suppliers invest, the money comes right back as chip and cloud bills, and the suppliers win either way. Public shareholders are the new folks at the table, and they are the ones holding the bag if revenue does not catch up to a $900 billion bill. I will give Anthropic credit for saying the quiet safety part out loud on the biggest day of its life. But a pacing pledge and a quarterly earnings call are going to pull in opposite directions, and Wall Street has a long memory for missed numbers.
03 of 04
Four Nonprofits Ask a Judge to Freeze Frontier AI
On November 12, four nonprofits filed suit in San Francisco Superior Court: the Irreplaceable Education Fund, Public Citizen, Heat Initiative, and the Alliance for Secure AI. The defendants are Anthropic, Google, Meta, OpenAI, and SpaceX's AI unit. Former New Jersey Attorney General Matthew Platkin filed the 63-page complaint, per Decrypt.
The legal theories are California's Unfair Competition Law and public nuisance. The ask is big. They want an injunction barring the companies from developing or deploying any AI system more capable than their best one as of the filing date, plus a ban on using AI to run its own research into better models. ChatGPT, Claude, and Gemini would keep running. Their successors would not. Platkin's line: "It is patently illegal for companies to develop and release products they know are unsafe in any respect."
Phil Aroneanu, who runs the Irreplaceable Education Fund, told Al Jazeera that a real freeze would need government inspections and the authority to halt prohibited work. Meta declined to comment, and the others had not responded to Reuters as of the filing coverage.
I do not think a state court judge is going to freeze five of the largest companies in America. That is not how this ends. But the clever bit is that the complaint leans on the labs' own public warnings about how dangerous this stuff could be. You cannot spend three years telling Congress your product might be an existential risk and then act shocked when somebody calls it a nuisance in court. Even if the case loses, discovery alone could get ugly for somebody.
04 of 04
Congress Publishes Who Got Nvidia Chips in China
On November 19, House Foreign Affairs Chairman Brian Mast released Commerce Department export license records for Nvidia H200 and AMD MI325X chips headed to Chinese buyers, using a provision in the Export Control Reform Act that lets the committee publish license data when secrecy is against the national interest. The table from the Bureau of Industry and Security shows status as of May 11, 2026, as broken down by TECHi.
Exporters asked for 3,444,160 chips across 54 applications. Commerce approved 879,868 of them. Here is who got the green light: ByteDance and Tencent at 150,000 each, Didi at 99,780, Alibaba, Kuaishou, VNET, and Linkwise Digital at 75,000 each, miHoYo at 57,810, Baidu at 43,000, JD.com at 41,300, Li Auto at 8,260, Great Wall Motor at 4,800, and BYD and Tesla's Shanghai unit at 4,000 each. Four of those buyers, Alibaba, Tencent, Baidu, and BYD, are on the Pentagon's list of Chinese military companies. The only flat rejection was AMD's request to sell 75,000 chips to UCloud.
Then there is the other pile. Nvidia filed five applications in December 2025 for 1,118,500 chips going to ByteDance, TikTok, and Alibaba Cloud sites in Malaysia, Indonesia, and Thailand. Every one of them was returned without action. Mast also said he believes two of the installers named on Nvidia's applications are fronts, which is his claim, not a finding.
I am glad somebody finally showed the math. For two years we have argued about chip export policy in the abstract, and it turns out the government was quietly approving hundreds of thousands of H200s to companies the Pentagon itself flags. Put that next to the Huang story and you see why this week feels different. The question is no longer whether chips leak to China. It is how many, through whom, and who signed off.