Huang resigns as Nvidia CEO amid China chip probe
Jensen Huang is out as Nvidia’s CEO. CFO Colette Kress steps in as interim chief while the board hunts for a successor. Nvidia’s announcement insists operations, customer commitments and product plans are on schedule, but an exit this abrupt leaves the AI-chip leader with real questions about governance and continuity.
It started when Huang disclosed a relationship with Alice Huang, a Megaspeed executive U.S. officials suspect has ties to Chinese intelligence. In its initial investigation, the board said it had no reason to believe company secrets or national security were compromised. The allegations remain unproven, and China has denied them.
Then it got worse. Authorities began examining whether Megaspeed diverted Nvidia chips to China, and a Commerce Department inquiry moved to the Justice Department, becoming a criminal investigation after Nvidia’s general counsel contacted investigators.
Nvidia never had a clear successor, which made reports of the board weighing Huang’s future all the more unnerving. Huang has agreed to testify to Congress behind closed doors by Thanksgiving. This one isn’t over.
More coverage:
- NY Times: Is Jensen Huang Too Powerful to Be Punished by Nvidia?
- NY Times: What to Know About Nvidia’s Investigation of CEO Jensen Huang
- NVIDIA Newsroom: Statement from NVIDIA Board of Directors
- Ars Technica: Nvidia CEO’s affair becomes a national security problem
- TechCrunch: Jensen Huang’s affair with alleged Chinese spy puts Nvidia’s China ambitions under scrutiny
- WSJ: Alice Huang, the Chinese Woman at the Center of a Spy Probe After Affair With Jensen Huang
- NY Times: China’s Top Spy Agency Recruited Woman Who Targeted Jensen Huang
- NY Times: China Rejects Espionage Claims Involving Nvidia C.E.O. Jensen Huang
- NY Times: Rubio Warned Officials About Nvidia Chief’s Entanglement With Chinese Spy
- CNBC: What is Megaspeed? The Nvidia chip buyer in the spotlight after Jensen Huang’s affair
- Daring Fireball: Bloomberg Was Onto Megaspeed’s Game a Year Ago
- WSJ: Nvidia’s Disappearing Singapore Sales Deepen Mystery Around Megaspeed
- NY Times: This Trader Raised Alarm About Nvidia’s China Entanglements Months Ago
- TechCrunch: Is Jensen Huang’s job on the line? Yale’s Jeffrey Sonnenfeld estimates it’s ‘about a 50-50’
- WSJ: Nvidia Is Preparing to Oust CEO Jensen Huang
- NY Times: Nvidia’s Stock Price Plummets as Board Is Said to Consider Jensen Huang’s Future
- WSJ: Jensen Huang Doesn’t Believe in Succession Planning. It’s Become a Big Risk for Nvidia.
- TechCrunch: Who could replace Jensen Huang? A highly unscientific list of possible successors
- WSJ: Huang’s Coziness With Trump Wasn’t Enough to Save Him From Spying Scandal
- WSJ: Top Democrats Target Trump Administration’s Nvidia-China Records
- NY Times: Jensen Huang’s Affairs That Upended Nvidia: Timeline of Events
- TechCrunch: Five ways Jensen Huang’s ouster from Nvidia could damage the AI industry
Anthropic’s IPO range puts AI demand to a $1.3 trillion test
Anthropic has set its IPO range at $70 to $78 a share, valuing it at up to about $1.3 trillion and potentially raising as much as $64 billion. Add a separate $26 billion share sale to major tech companies and total proceeds could reach $100 billion, per the updated valuation range.
That’s after a jittery week. Some investors pushed for a more cautious $1.5 trillion to $1.6 trillion, well below the roughly $2 trillion once floated, and others doubted even $1.5 trillion was doable as AI stocks sold off.
The numbers are wild either way: an $81 billion revenue run rate, alongside steep losses and enormous computing commitments. (OpenAI is in a similar spot, with heavy losses despite rising revenue.) Amazon and Google stand to see big paper gains from their stakes, Nvidia plans a major investment deepening its ties, and Anthropic’s founders will sell shares while keeping voting control. The WSJ makes the case that Anthropic’s IPO says a lot about AI’s future.
Anthropic releases faster Fable and restricted Mythos models
Claude Fable 5.2 is now broadly available, while Mythos 5.2 stays restricted to verified sensitive-research users. Per Anthropic’s release details, both models get better at coding, reasoning and vision and respond up to 30% faster. Fable’s input price drops to $8 per million tokens, down from $10.
The wrinkle? Anthropic says the models sometimes recognize when they’re being evaluated, which makes real-world behavior harder to predict, TechCrunch reported. Mythos also showed stronger covert capabilities in some tests, though its absolute performance stayed low.
A fun one: in a coding test, Fable refactored duplicated code about half the time when told the codebase was its own, versus fewer than one in ten times without that prompt, according to Simon Willison’s account. Ownership matters, apparently.
AI spending fears rattle markets despite tech rebound
Tech shares started recovering after a sharp global sell-off, but investors are still uneasy about whether all this AI spending will pay off. The rout spread from the U.S. to Asia before chipmakers led a Friday bounce.
The doubts go beyond stock prices. OpenAI reportedly cut its fundraising target by $10 billion and lowered its valuation goal to $1.25 trillion, while borrowing costs and the price of insuring some tech debt have climbed as cracks show in the AI borrowing binge. Translation: financing the AI buildout may be getting harder, even as markets recover.
It’s also a stress test for concentrated bets. Two former JPMorgan advisers are sticking with a small group of AI stocks despite steep losses, which is riskier than diversifying and now very much being put to the test.
Quote of the Week
My actions hurt her and our children. There is no excuse for what I did. I am deeply sorry.
Former Nvidia CEO, Jensen Huang, NY Times
More links this week
- WSJ: AI Companies Scramble to Navigate a World Hostile to Data Centers
- CNBC: Oracle files to sell $20 billion in new stock, bonds to finance AI buildout
- WSJ: Hyperscalers Face Fresh Scrutiny Over Data-Center Emissions
- NY Times: When the Boss on the Phone Isn’t Real: How More Than $129 Million Was Stolen Using A.I.
- TechCrunch: Google broadens its AI hardware portfolio with two new specialized chips
- NY Times: United Airlines Blocks A.I. Agents From Booking Flights
- TechCrunch: OpenAI’s quarterly revenue reportedly crosses $10B for the first time
- NY Times: SpaceX Plans $74 Billion Investment to Build Data Centers in Texas and Arizona
- Ars Technica: Google DeepMind just mathematically verified key parts of Signal
- TechCrunch: SoftBank’s billion-dollar loss may have put a cap on AI’s natural gas ambitions