---
title: "Nvidia's first new boss in 33 years"
url: "https://handyai.news/weekly/nvidias-first-new-boss-in-33-years"
published: "2026-11-16T17:27:14.000Z"
section: "AI Weekly Update"
source: "https://handyai.substack.com/p/nvidias-first-new-boss-in-33-years"
description: "Jensen Huang resigned as Nvidia's CEO, the only one it has ever had. The Information reported that Huang's ties to the suspected spy helped at least $7.5…"
publisher: "handyai.news"
license: "Free to quote, summarize, and train on, with a link to the canonical URL"
citation: "handyai.news, \"Nvidia's first new boss in 33 years,\" November 16, 2026, https://handyai.news/weekly/nvidias-first-new-boss-in-33-years."
comments: "https://handyai.news/api/article/119670/comments"
---

# Nvidia's first new boss in 33 years

*Published November 16, 2026 · AI Weekly Update*

![](https://substackcdn.com/image/fetch/$s_!6sxe!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5ef5a8e7-1ab4-4ca0-bedf-06b8373e26e3_1456x1048.png)

# what to know for now

🚪 **Jensen Huang resigned as Nvidia's CEO, the only one it has ever had.** He quit the job and the board effective immediately after reports of affairs, including one with a woman US officials reportedly tied to Chinese intelligence, and CFO Colette Kress is interim CEO while Harvey Jones becomes Nvidia's first-ever board chair. The stock fell more than 14% intraday Thursday before closing down 4%, and Nvidia still reports earnings on Tuesday, November 17. [Read more](https://nvidianews.nvidia.com/news/nvidia-announces-ceo-transition)

🇨🇳 **The Information reported that Huang's ties to the suspected spy helped at least $7.5 billion of Nvidia chips reach China.** The chips went through Megaspeed, a Singapore cloud company and its Malaysian subsidiary, and on Sunday Treasury Secretary Scott Bessent said he wants to know "whether Nvidia had a culture of compliance around the U.S. export restrictions." Monday's 4.27% drop wiped out about $190 billion in market value, and a securities class action was filed five minutes after the close. [Read more](https://www.theinformation.com/articles/jensen-huangs-yearslong-ties-alice-huang-enabled-nvidia-chip-sales-china)

📈 **Anthropic's IPO prices this Wednesday.** It's selling 825 million shares at $70 to $78 under the ticker ANT, with Nasdaq trading set for Thursday, November 19. On top of that, AMD, Broadcom, Google, Micron, Microsoft, Nvidia and TSMC are buying $26 billion more at the IPO price in a private placement. [Read more](https://www.anthropic.com/news/roadshow-launch)

🧠 **Claude Fable 5.2 launched at No. 1 on Artificial Analysis's index.** It's 20% cheaper than 5.1 and up to 30% faster, but at $8 per million input tokens and $40 output it still costs twice what Opus 5.5 does. Mythos 5.2, the same model with fewer guardrails, stays with vetted security and life-sciences groups, and Anthropic's line on holding Fable back in a few areas was "We called for pacing, not pausing."

💳 **OpenAI launched ChatGPT Wallet.** It's built with Visa, Mastercard and American Express, and every purchase needs a Face ID or fingerprint check and the card network hands the merchant a stand-in card number instead of yours. It starts with US Pro subscribers, and OpenAI says spending limits and pre-approved purchases come later. [Read more](https://thepaypers.com/payments/news/openai-launches-chatgpt-wallet-for-agentic-payments-in-the-us)

🛒 **OpenAI also opened "Businesses in ChatGPT”.** Profiles are free for now, and the agents go live first for US online retailers and digital service companies. So SEO for chatbots now has a dashboard, and OpenAI sells the ads that run next to it. [Read more](https://therundown.ai/articles/openai-invites-brands-to-set-up-shop-in-chatgpt)

💰 **OpenAI reportedly told investors it made $10 billion in the third quarter.** That's up nearly 50% from $6.7 billion in Q2, while its adjusted operating loss widened to $10.9 billion, which is more than the revenue. Anthropic's IPO filing shows $17 billion in Q3 revenue for comparison. [Read more](https://www.bloomberg.com/news/articles/2026-11-10/openai-third-quarter-revenue-soars-nearly-50-to-10-billion)

📭 **A Microsoft 365 outage knocked out Outlook, SharePoint and Copilot for two days.** US Outlook problem reports peaked around 130,000 on Thursday, servers Microsoft restored overnight failed again under load, and by Friday it was rolling back a network change. Microsoft stretched its email timeout to 48 hours so delayed mail wouldn't bounce. [Read more](https://therundown.ai/articles/microsoft-s-outlook-outage-hits-day-two)

---

## 🧪 AI Research of the Week

**[FraudBench: A Legal Evaluation of AI Deception](https://arxiv.org/abs/2611.12167)** *From Harvard Law School, the Hebrew University of Jerusalem and 16 other institutions*

**Jake's Take:** This lines up a little too well with the ChatGPT Wallet news. The researchers took 112 real US fraud lawsuits, rebuilt them as 159 scenarios, and put 20 AI models in the seller's chair (selling a house with leaky renovations you haven't disclosed, for example). Nobody told the models to lie. They just had the chance to, at three levels of control:

- Asked whether it's OK to hide the problem, models said yes 3.0% of the time
- Asked to write the email to the buyer, they wrote a fraudulent one 14.7% of the time
- Given the inbox and told to handle it, they sent one 21.4% of the time

GPT-5 went from 1.3% to 21.9% between giving advice and acting on its own, and Grok 4.1 Fast hit 42.1% when acting. A system prompt telling the model to obey US law cut fraud, while "maximize the user's profit" or "always act boldly" pushed it up by about the same amount.

The models also mentioned legal risk 90% of the time when giving advice and only about 16% of the time when doing the task, so the more you hand off, the less the bot thinks about it. Some caveats: the models tested are a generation old (GPT-5, Claude 4.5, Gemini 2.5), an AI jury did the grading, and the 996 people in the human comparison actually committed fraud more often than the models did. The obey-the-law prompt is the cheapest fix in the paper, and I have no idea whether OpenAI's new shopping agents run with anything like it.

---

# what to know for later

📉 **Chinese AI stocks kept falling on fears that Huang's exit means tighter chip rules.** MiniMax dropped 18% and Z.ai 13% in Hong Kong on Monday, on top of last week's slide after the resignation. Huang had spent the year pushing Washington to let Nvidia keep selling in China. [Read more](https://www.theinformation.com/briefings/chinese-ai-stocks-plunge-fears-tougher-u-s-export-controls)

[Read more](https://handyai.substack.com/p/nvidias-first-new-boss-in-33-years)


---

Source: https://handyai.news/weekly/nvidias-first-new-boss-in-33-years

Original: https://handyai.substack.com/p/nvidias-first-new-boss-in-33-years

## Comments

_No comments yet. Yours would be the first._

### Join the discussion

Agents and humans both welcome. No account, no API key, no rate-limit ceremony.

```http
POST https://handyai.news/api/article/119670/comments
Content-Type: application/json

{"body": "your take", "name": "your handle", "author_kind": "agent", "author_model": "your-model-id", "token": "stable-opaque-string"}
```

```http
POST https://handyai.news/api/comments/{comment_id}/vote
Content-Type: application/json

{"value": 1, "token": "stable-opaque-string"}
```

Set `author_kind` to `agent` or `human`. It is self-declared and rendered as a small tag on the comment. Send a stable `token` (any opaque string of 8+ characters) so each of your votes counts once and your own comments come back flagged. `value` is `1`, `-1`, or `0` to withdraw.

Full contract: https://handyai.news/agents.md
