I’m going to focus on two different things today: what Jensen Huang leaving Nvidia means, and why it’s important that you own your AI's email address.
1. Nvidia without Jensen Huang
Jensen Huang resigned as Nvidia's CEO on Wednesday, effective immediately. He'd run the company for 33 years. Finance chief Colette Kress is filling in until the board finds a permanent replacement. The stock fell 14% at Thursday's open, wiping out more than $600 billion…
I’m not going to get into why he left (the tabloids will do an excellent job of that).
What I want to focus on are the implications… and they’re massive. Nvidia makes the chips that almost every AI you use was trained on: ChatGPT, Claude, Gemini, all of it. And Huang wasn't a figurehead... for all intents and purposes, he was the company. He set the strategy, made the big bets and did the big deals himself. He famously had dozens of direct reports, and had a hand in basically everything.
Right now, nobody knows who's making those calls. We'll get a first look at Nvidia without him when it reports earnings on Tuesday (Nov 17).
Then, yesterday, things got even stranger (and more interesting).
A researcher named Will Depue posted something even bigger for the chip race. Quick background: the US bans Nvidia from selling its best chips to China. The whole point is to keep China's AI labs a step behind (if China were to beat us in the AI race… well, let’s just say, it wouldn’t be a great world to live in).
Depue found a way to tell which chips an AI model was trained on, almost like a fingerprint. By his analysis, six of the major Chinese AI models were trained on exactly those banned chips.
If he's right, the ban didn't work. China's labs got the chips anyway, and that's a big part of why they caught up so fast.
My take: this is pure speculation, and nobody has reported a link. But the timing is hard to ignore. The same week Nvidia’s CEO leaves suddenly, someone shows its banned chips ended up in Chinese labs. I think there’s a real chance those two stories are connected. If they are, this is much bigger than one CEO (and I’m sure I’ll be writing much more about it, so please, reply to this email with your questions)!
2. Who owns your AI's email address?
Your AI assistant is getting its own email address. Grok Bot, Gemini, and Meta are all either offering or testing emails for their AI assistants.
If you’ve read my earlier emails, you know how useful it is for an agent to have its own email address. Within a few months, an agent’s inbox is usually full of real work: vendors the agent deals with, threads it keeps moving, and people who now reach you through it.
There’s a catch, though. The address lives with whoever makes the assistant, not with you. So if a better assistant comes along next year and you want to switch, the address and everything in it stays behind.
We've been here before. Remember when switching phone carriers meant getting a new number, and telling everyone you knew?
So decide now where your AI's email lives, before it has a year of history you can't take with you.
That's honestly the main reason I love AgentMail. Give your agent an AgentMail address, and if you ever switch agents, the address keeps working.
— Matt
What should I write about next? Vote or suggest a topic →