Now that Jensen Huang is gone, I think America is about to discover something the hard way:
Nvidia didn’t just lose a CEO. It lost the one guy who knew how to run an Asian supply chain like an old-school merchant.
You can replace him with a brilliant American executive, an Indian executive, a Chinese executive — whatever.
The title is replaceable. The relationships are not.
Anyone who grew up around old-school Chinese business culture will recognize Jensen’s style immediately. You eat together. You drink together. You remember who helped you twenty years ago. When you decide someone is your guy, you keep dealing with him for decades. Contracts matter, of course, but when things go wrong at 2 a.m., what matters even more is whether the person on the other side will pick up your call.
Look at Nvidia’s supply chain. TSMC makes the chips. SK Hynix makes the HBM. Foxconn, Quanta, Wistron, Pegatron and others build the systems. Jensen didn’t suddenly discover these companies when ChatGPT arrived. He spent decades personally building relationships with the people running them.
TSMC is probably the best example. Back in 1998, when Nvidia was still tiny and short on manufacturing expertise, TSMC sent two production people over to help. The relationship became so close that Morris Chang later asked Jensen twice if he wanted to become CEO of TSMC. Jensen said no both times.
Decades later, Jensen was still flying back to Taiwan, eating with Morris Chang, Barry Lam and half of the island’s tech establishment. At one Taipei dinner, the guest list looked less like a restaurant reservation and more like the control room of the global AI supply chain: TSMC, Quanta, Pegatron, Wistron, Asus, Acer, Inventec, Foxconn.
That stuff looks almost comically old-fashioned in an industry obsessed with AI agents and trillion-dollar market caps. But then you see what happens when Nvidia actually needs something.
In 2024, Jensen wanted SK Hynix to move its HBM4 supply schedule forward by six months. Six months in semiconductors is insane. You’re talking about engineering, capacity and production plans that were set long before Jensen showed up asking for more.
SK chairman Chey Tae-won publicly told the story later. Jensen asked. Chey went back to his people and pushed.
That is not something you reproduce by handing the next CEO a supplier spreadsheet.
And Jensen also knew what it meant to owe somebody.
In Nvidia’s early days, its project with Sega went badly enough that the company was in real trouble. Jensen went to Japan, admitted the technology wasn’t going to work, and Sega still gave Nvidia $5 million that helped keep the company alive.
Thirty years later, Jensen went back to Tokyo and thanked former Sega president Shoichiro Irimajiri in person. His message was basically: without what you did for us back then, there would be no Nvidia today.
He did the same thing on the other side of the table.
Serve Robotics CEO Ali Kashani has talked about December 2023, when his company was so close to dying that he had already discussed bankruptcy. An investor pulled out right before Christmas. Kashani called Jensen. Nvidia helped carry the financing, other partners joined, and a few months later Serve was ringing the Nasdaq bell.
That’s the Jensen playbook. Eat with people. Remember favors. Return favors. Call directly when there’s trouble. Keep relationships alive for twenty or thirty years.
Silicon Valley has a fancy word for this: “network.”
My family would probably just call it doing business,and now Jensen is gone.
Nvidia can transfer his office, his org chart, his contracts, his procurement team and every supplier database to the next CEO.
What it cannot transfer is the invisible ledger sitting inside the heads of dozens of people across Taiwan, Korea and Japan:
He helped me once. I helped him once.
We’ve known each other for 25 years.
If he calls, I’ll listen. There is no handover document for that.
This letter was sent to the Nvidia Board today which we believe outlines the minimum acceptable standards the Board owes shareholders as it manages this difficult situation. Onward. 🫡
We've spent the better part of 2026 building the tools companies need to put AI to work: Stack, Token Spend Management, https://t.co/x0WFBLoteu, and Glass.
Today we raised $1.85B at $60B valuation. There are now 85,000 businesses on Ramp saving a combined $1B+ a month.
We still have so much to build.
After we have determined all the proofs, discovered all the drugs, found all the new materials, we will have to face the reality that our deepest problems are political, philosophical and psychological.
"People aren't ready, this is gonna change everything"
You know what people weren't ready for? Fucking mammoths. Tigers eating them. Monkeys ripping their arms off. The black plague. The ice age. Smallpox. Vikings pulling up on the beach. Drinking water that gives you diarrhea until you die. Getting a scratch and losing your whole leg. The Mongols. The Spanish Inquisition.
One volcano can send us back to the dark ages and these safety guys are gonna be shocked. Oh the robots didn't kill us? It was a volcano? Imagine their disappointment as the hot ash rains down on them.
A solar flare is gonna explode your phone in your pocket your legs are gonna be covered in battery acid and your bank account will stop working. You're afraid of a computer program doing math?
"Things are moving too fast" homie you live on a rock that periodically deletes everything. You are one the most luckiest entities in a 100 light year radius living during a golden era. Cheer up.
My ancestors crossed an ocean in a wooden box because they ran out of fucking potatoes.
"The social disruption will be enormous!"
Nigga World War 2? World War 2? Hop in the people carrier we're landing on the beach.
Are these people out of their fucking minds? Do they not know what we've survived already?
A couple decades of air conditioning and people suddenly think oh this is the natural condition of mankind. No. We used to have eight kids because seven didn't make it. You're gonna be fine. Man the fuck up.
Accelerate.
we built muse from scratch, but it is definitely heavily inspired as a product by openclaw. After I used openclaw in january I bought hundreds of mac minis for the MSL team and lots of us fell in love with using openclaw (and other personal agents). @steipete is a genius and his harness was pioneering from the jump. I think a lot of people were inspired by it. our goal with muse was to build something like openclaw that we could make safe and secure and easy to use and scale to billions of people.
something most folks who use muse won't ever know or appreciate was the degree to which this was @natfriedman's baby.
it has been one of the great honors of my career watching nat cook. he assembled and ran the team, and so many of the little details that people love came from his beautiful mind.
nat is one of the great artists of our time with products as his medium. more people should experience his brilliance, and with muse, they can ❤️
If tech was a pro sports league, it is one with no salary cap. The bet to assemble a galaxy of AI founder talent to compete is paying out with a generational product
Feels like @Muse will be @Meta’s next billion+ user app if the infra team can scale compute / DCs fast enough, and the business teams can find clever ways to cover the cost of giving away so much for free. The UX, VM + fast computer-use and connectors are truly great.
Muse is out today.
An agent like this is a different kind of system to design. It reads your mail, spends your money, runs your code. Getting that right took real care, and our technical post walks through exactly how. We designed security in.
2/ also for the first time ever, through this partnership, we are providing agentic payment protection – if Muse makes a mistake, a refund guarantee will cover it. More details here: https://t.co/QRcZHG9RaQ
Every year I get a few inbound saying the Operator vision would work now.
What people forget is now anyone could build my old version with an API call and it was really hard to do from scratch then
Instinct and this new generation of companies will crack this with very modern expressions and capabilities
(I still have the domain!)
Over 10 years ago, I led an investment in Operator - to make businesses textable.
While it didn’t quite hit the timing right, so much of this feels like it is happening right now, with AI agents on the other side.
Excited for the new wave!
https://t.co/UdyNndAcbb
Today I'm excited to share an update on @Neko's US expansion.
This launch is probably the most consequential milestone in our company's mission to make preventive healthcare accessible at population scale.
Modern medicine is extraordinary at fixing you once something has gone wrong. It is almost entirely uninterested in looking before that happens.
We've spent the past 8 years building an alternative and we are finally ready to bring it to the US for the first time.
Neko NYC opens on 24 September at 300 Lafayette Street in SoHo. A Neko Health Scan is $499.
Americans are already spending thousands of dollars to access yearly labs, on-demand scans, or concierge medicine. At Neko, there are no third-party imaging centres, no send-out labs, nobody in the middle taking a cut. We built a completely new healthcare experience ourselves to make it both better and cheaper for you.
If you're not familiar with our concept, in one hour you get a full skin check, body composition, EKG, 4-point blood pressure analysis, macro and micro vascular assessment, and comprehensive labs with results before you leave. Then a clinician sits down with you and takes you through every finding, in person, before you leave the building.
For about 1.2% of scans we have a direct life-saving intervention.
There are already 25,000 New Yorkers in line for a scan. That could mean saving 300 lives in NYC alone.
Miami, SF and Washington DC are already under construction. Waitlists for those cities open today (linked in the comments). 🇺🇸
I grew up on food stamps in rural Kentucky. I have founded a company that makes clean, safe nuclear energy. I have created hundreds of jobs and will create thousands more. My family and I have bled sweat and cried to build this.
The idea that you can just steal this from me is awful and evil. You are not the good guys
btw anthropic's internal document on this literally said "we don't want it to be known that we are working on this.”
it was called project panama.
here's exactly what happened:
1: anthropic concluded that books were the cheapest way to build a world-class model because they gave claude curated facts, structured arguments, compelling stories, and writing “an editor would approve of.”
2: once anthropic decided it needed books at enormous scale, its first solution was piracy.
it downloaded 7m+ books from online libraries including libgen. the judge later wrote that although anthropic had legal ways to buy them, it chose piracy to avoid what dario amodei called the “legal/practice/business slog.”
3: that piracy created a massive legal risk.
so in february 2024, anthropic hired tom turvey, the former head of partnerships for google books, to find a legally safer way of obtaining “all the books in the world.”
4: turvey first contacted major publishers about licensing their catalogs.
those attempts didn’t produce agreements, so anthropic chose a route that required no publisher permission: buying millions of physical books through distributors and used-book retailers.
5: within about a year, anthropic spent tens of millions acquiring and scanning millions of books, including many rare and 1/1 titles. one vendor proposal targeted 500,000 to 2 million books in six months.
6: to scan that many books within months, the vendors physically dismantled them.
a hydraulic cutter removed each spine. the pages were trimmed to size, fed as loose sheets through high-speed industrial scanners, and converted into searchable PDFs. the paper remains were then sent for recycling.
7: these PDFs were fed into claude as training data.
the complete collection became a private, searchable anthropic library that the company planned to “store forever.” the scans aren’t available to the public and were never open-sourced.